<?xml version="1.0" encoding="UTF-8" standalone="no"?><rss xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:slash="http://purl.org/rss/1.0/modules/slash/" xmlns:sy="http://purl.org/rss/1.0/modules/syndication/" xmlns:wfw="http://wellformedweb.org/CommentAPI/" version="2.0">

<channel>
	<title>Mobiletech</title>
	<atom:link href="https://www.mobiletech.mobi/feed/" rel="self" type="application/rss+xml"/>
	<link>https://www.mobiletech.mobi</link>
	<description></description>
	<lastBuildDate>Tue, 17 Nov 2026 11:15:37 +0000</lastBuildDate>
	<language>en-US</language>
	<sy:updatePeriod>
	hourly	</sy:updatePeriod>
	<sy:updateFrequency>
	1	</sy:updateFrequency>
	<generator>https://wordpress.org/?v=7.1</generator>
	<xhtml:meta content="noindex" name="robots" xmlns:xhtml="http://www.w3.org/1999/xhtml"/><item>
		<title>Google Search Console API Adds Lens Reporting</title>
		<link>https://www.mobiletech.mobi/search-console-multimodal-api/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Tue, 17 Nov 2026 11:15:37 +0000</pubDate>
				<category><![CDATA[Mobile SEO]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=597</guid>

					<description><![CDATA[Pull Lens and Circle to Search data by page, country and device. Compare the new API results with your existing Search Console reports.]]></description>
										<content:encoded><![CDATA[<p>On November 16, 2026, Google added <a href="https://developers.google.com/search/blog/2026/11/search-analytics-api-update">Lens reporting, brand classification and multi-value filters to the Search Analytics API</a>. A retailer can now pull the product pages people find through image-based searches on phones in several countries, in one request. If your dashboard already pulls Search Console data, add a page-level multimodal series. Use the API for recurring reports; for a one-off check, select the multimodal search type in Search Console and export it.</p>
<table>
<caption>Pick the request for your report</caption>
<thead>
<tr>
<th scope="col">Goal</th>
<th scope="col">Use</th>
</tr>
</thead>
<tbody>
<tr>
<td>Lens and Circle to Search pages</td>
<td>Query <code>webMultimodal</code> by page</td>
</tr>
<tr>
<td>Google Images tab</td>
<td>Keep <code>type: image</code></td>
</tr>
<tr>
<td>Brand share in web search</td>
<td>Group by <code>brandedClassification</code></td>
</tr>
<tr>
<td>Phones and tablets across markets</td>
<td>Add country and device lists</td>
</tr>
</tbody>
</table>
<h2>Which searches count toward the new report?</h2>
<p><code>webMultimodal</code> covers searches made with inputs such as a camera image or a screenshot. The <a href="https://developers.google.com/search/blog/2026/09/web-multimodal-in-sc">September 24 Search Console rollout</a> began globally and listed Google Lens, Circle to Search on Android, image uploads in Google Search and Chrome’s Search this image command. Someone can type a shoe name and select the Images tab. That activity belongs to the <a href="https://developers.google.com/webmaster-tools/v1/searchanalytics/query">existing <code>image</code> search type</a>. Someone who uses Lens to identify a shoe from a picture is the kind of search the new report captures.</p>
<p>The reporting rollout includes the US, the UK and Europe; country filters select the audience you measure. Start by grouping results by page. You can see which destinations turn up often and which get clicks. The report combines several ways of searching by image, and <a href="https://support.google.com/webmasters/answer/17011259?hl=en-NI">specific query text is unavailable</a>. A desktop result tells you which device class was recorded. It doesn&#8217;t identify the original picture or the tool the person used.</p>
<h2>Pull one region and two device types</h2>
<p>Send a POST to the <a href="https://developers.google.com/webmaster-tools/v1/searchanalytics/query">Search Analytics query endpoint</a> using your existing OAuth connection. For a reporting job, use the <a href="https://developers.google.com/webmaster-tools/v1/searchanalytics/query"><code>webmasters.readonly</code> OAuth scope</a>. The endpoint begins <code>https://www.googleapis.com</code>. In the path <code>/webmasters/v3/sites/</code> followed by <code>{siteUrl}</code> and <code>/searchAnalytics/query</code>, encode the full property identifier as one segment. The Domain property <code>sc-domain:example.com</code> becomes <code>sc-domain%3Aexample.com</code>. A URL-prefix property uses its full URL exactly as listed in Search Console.</p>
<p>This body asks for pages found in Germany, Austria or Switzerland on a phone or tablet. The start and end are inclusive and use Pacific Time, even when the report covers European traffic. The country list accepts ISO three-letter codes. Use a device list for <code>MOBILE</code> and <code>TABLET</code>, or use a single device value if your report is phone-only.</p>
<pre><code>{
 "startDate": "2026-10-17",
 "endDate": "2026-11-13",
 "type": "webMultimodal",
 "dimensions": ["page"],
 "dataState": "final",
 "dimensionFilterGroups": [{
  "groupType": "and",
  "filterLists": [
   {
    "dimension": "country",
    "expressions": [
     "deu", "aut", "che"
    ]
   },
   {
    "dimension": "device",
    "expressions": [
     "MOBILE", "TABLET"
    ]
   }
  ]
 }]
}</code></pre>
<p>A row needs a match in each list. Germany on a tablet qualifies. Germany on desktop and the UK on mobile fall outside this request. To report UK and Irish traffic instead, use <code>gbr</code> and <code>irl</code>; for the US, use <code>usa</code>. The country is where the search was made, so changing the list changes the audience in the report. If a job already has a country rule in <code>filters</code>, replace it when you add the country list. The API rejects filtering the same dimension in both fields.</p>
<p>Each row’s <code>keys</code> array follows the requested dimension order. Here it contains the canonical page URL. Keep clicks and impressions beside CTR and average position. API <code>ctr</code> is a fraction from 0 to 1; multiply by 100 to display a percentage. For a single regional total, omit <code>dimensions</code> and leave the filter lists in place. Add <code>country</code> or <code>device</code> to <code>dimensions</code> when you need separate rows. Page and property aggregation can count impressions differently, so compare totals using the same <code>responseAggregationType</code>. Leave <code>aggregationType</code> at <code>auto</code> when grouping or filtering by page; <code>byProperty</code> is invalid in that case. Responses are sorted by clicks, highest first; date-grouped responses run from oldest to newest. The default <code>rowLimit</code> is 1,000; set it to 25,000 and advance <code>startRow</code> for larger reports. Pagination still cannot guarantee every detail row, so obtain a site total with an aggregate request rather than summing page rows.</p>
<h2>What a real regional total looked like</h2>
<p>In a <a href="https://podrez.pl/en/search-console-api-branded-lens/">November 16 test on 22 sites</a>, analyst Ewelina Podrez-Siama reported that a regional call for madeira-planer.de returned 65 clicks, 1,824 impressions and an average position of 16.07 for Germany, Austria and Switzerland from October 17 to November 13. Three individual country calls produced the same rollup when position was weighted by impressions. This was a check of the country filter. The phone-and-tablet Lens request above is a different slice.</p>
<figure>
<svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 350 300" width="100%" role="img" id="gsc-region-match" aria-labelledby="gsc-region-title gsc-region-desc">
<title id="gsc-region-title">When to replace three country requests with one</title>
<desc id="gsc-region-desc">For October 17 to November 13, 2026, both methods on Madeira Planer returned 65 clicks, 1,824 impressions and position 16.07. Compare the same scope; switch only when the totals match.</desc>
<rect x="1" y="1" width="348" height="298" rx="12" fill="#edf5ff" stroke="#829cb6"></rect>
<text x="16" y="30" font-size="18" font-family="sans-serif" font-weight="bold" fill="#152f49">Check before replacing the loop</text>
<text x="16" y="58" font-size="16" font-family="sans-serif" fill="#152f49">Same property, dates, type and filters</text>
<line x1="16" x2="334" y1="76" y2="76" stroke="#829cb6"></line>
<text x="16" y="104" font-size="16" font-family="sans-serif" fill="#152f49">DEU + AUT + CHE: three calls</text>
<text x="16" y="132" font-size="16" font-family="sans-serif" fill="#152f49">One list: [deu, aut, che]</text>
<text x="16" y="166" font-size="18" font-family="sans-serif" font-weight="bold" fill="#152f49">Both: 65 clicks · 1,824 impressions</text>
<text x="16" y="194" font-size="16" font-family="sans-serif" fill="#152f49">Weighted position: 16.07</text>
<line x1="16" x2="334" y1="212" y2="212" stroke="#829cb6"></line>
<text x="16" y="240" font-size="17" font-family="sans-serif" font-weight="bold" fill="#152f49">Match → use one regional call</text>
<text x="16" y="270" font-size="16" font-family="sans-serif" fill="#152f49">Mismatch → keep loop; check scope</text>
</svg>
<figcaption>Measured country-filter example, October 17–November 13, 2026. The migration rule compares like-for-like totals. Source: <a href="https://podrez.pl/en/search-console-api-branded-lens/">Podrez-Siama’s API test</a>.</figcaption>
</figure>
<p>The CTR for that slice is 65 divided by 1,824, or about 3.56%. Averaging the three country CTRs would give a small market the same influence as a large one. When the region request matches clicks, impressions and weighted position in your own parallel run, switch the recurring regional total to one call. Keep country rows if an editor or merchant acts on each market separately.</p>
<h2>Give brand reporting its own request</h2>
<p>For your web report, set <code>type</code> to <code>web</code> and replace the page dimension with <code>brandedClassification</code>. Keep the country and device lists if you need the same audience. The row keys distinguish <code>BRANDED</code> and <code>NON_BRANDED</code>. To request just the second group, use a brand filter with <code>equals</code> and <code>NON_BRANDED</code>. A <a href="https://developers.google.com/webmaster-tools/v1/searchanalytics/query">query-text filter is incompatible with brand grouping or filtering</a>, and Podrez-Siama’s live request combining brand classification with <code>webMultimodal</code> was rejected. Keep brand and multimodal reporting in separate request bodies. The same live test rejected branding with <code>discover</code>, but accepted it with <code>image</code>.</p>
<p>For the API, brand history starts February 21, 2026. Use a top-level property covering the site: path-only and subdomain properties are ineligible, as are sites without sufficient impressions and query volume. The UI filter was announced November 20, 2025, and reached all eligible sites March 11, 2026; those rollout dates are separate from the API history cutoff. </p>
<p>The test covered eight of the analyst’s sites and 14 client or friend properties. Twelve returned the split; visible-query coverage varied from 8% to 57% of clicks by site. Their two classes matched total clicks, while visible queries covered 45% of clicks across the sample. A regex over query rows therefore measures a different population. <a href="https://developers.google.com/search/blog/2025/11/search-console-branded-filter">Google’s classification</a> also handles misspellings and names of products associated with the site. Prefer the new split for eligible web reports after a label audit, because it can include clicks whose query text is hidden. Keep historical regex results labeled as query-text estimates.</p>
<p>For that audit, request the two dimensions together. Grouping by <code>query</code> is allowed here; filtering on query text is what conflicts with branding. This body returns each visible query and its classification in that order:</p>
<pre><code>{
 "startDate": "2026-10-17",
 "endDate": "2026-11-13",
 "type": "web",
 "dimensions": [
  "query",
  "brandedClassification"
 ],
 "dataState": "final",
 "rowLimit": 25000
}</code></pre>
<p>To reproduce a regional brand-and-device report, instead use <code>brandedClassification</code> and <code>device</code> as the dimensions and the country list from the first body. For a brand-share total, group only by classification and divide branded clicks by the sum of both classes. Audit the labels before adopting the metric, then check them monthly. The classifier handles multiple languages, but a generic phrase can still receive a branded label. These labels help reporting; they do not change rankings.</p>
<p>The label audit matters even for an exact name. On podrez.pl, the query <code>ewelina podrez-siama</code> received 195 impressions: 50 branded and 145 non-branded, with the label changing by day. On ms-fox.pl, 180 of 186 impressions for the same query were branded. Conversely, the classifier found 231 branded clicks on ms-fox.pl where a regex for the blog name found 25. Include exact names, typos, associated products and generic phrases in the label audit.</p>
<p>Across the 12 classified sites, branded CTR ranged from 3.6% to 75%, versus 0.6% to 3.3% for non-branded traffic. Nine sites had branded click shares below 5%; the other three ranged from 18% to 30%. Those are different businesses, so use your own matched-period comparison to judge change.</p>
<h2>When a request comes back empty</h2>
<p>Run a broad request for the same property and finalized dates before investigating individual pages. First remove the country and device filters. Then remove grouping and compare the aggregate with the matching Search Console report. A row with impressions and zero clicks is a useful signal. An absent row calls for a scope check. For the branded report, check the API history cutoff and property eligibility first. There is no published numeric eligibility threshold; test the intended property rather than adopting another site’s traffic level as a rule. </p>
<p>For the Lens report, check whether the <a href="https://developers.google.com/search/blog/2026/09/web-multimodal-in-sc">Web multimodal view</a> shows data in that period. Keep <code>dataState: final</code> for reconciliation. Use <code>all</code> only when you want recent, potentially changing data. For multimodal detail, group by page, country, device, date or search appearance; adding <code>query</code> produces an error. A <code>startRow</code> beyond the available rows also produces an empty successful response.</p>
<p>If your client library rejects the new field before sending it, try the authenticated REST request. When the API itself rejects it, inspect the returned field error and compare the body against the current reference. Save the property, search type, date range and filters alongside the results. Store separate statuses for an empty response, a rejected request and a row with zero clicks.</p>
<h2>Choose a page to inspect</h2>
<p>Concentration is more useful than the sample average. On podrez.pl, multimodal impressions reached 5% of web impressions and 91% of that activity went to one AI Mode guide; a quarter of the guide’s multimodal impressions were desktop. After screenshots were added on October 27, the guide rose from at most 12 impressions a day to 138–231 a day from November 1. That sequence identifies a page to investigate, but does not establish that the screenshots caused the rise. The sample’s earliest multimodal data was September 10, before the UI announcement; check your property’s available dates rather than assuming the release day is its first data day.</p>
<p>In the <a href="https://podrez.pl/en/search-console-api-branded-lens/">22-site sample</a>, 15,806 multimodal impressions produced 105 clicks, a rounded 0.7% CTR; impressions were equivalent to 0.4% of web impressions. Among sites with at least 40 multimodal impressions, average position ranged from 1.5 to 2.3 and phones accounted for 74–96% of impressions. For your own store or publication, look for a page with repeated impressions over finalized dates, then check that its main image, title and destination make sense together on a phone. If clicks look steady but sessions fall, <a href="https://www.mobiletech.mobi/google-search-referrer/">trace the landing and referrer</a> before changing the photo or the report. </p>
<p>For an existing warehouse pipeline, the <a href="https://developers.google.com/search/blog/2026/11/search-analytics-api-update">update also adds multimodal data to active BigQuery bulk exports</a>. Keep the API for a small dashboard; use the <a href="https://support.google.com/webmasters/answer/12918484?hl=en-EN">daily bulk export</a> when you need detail beyond API top-row limits.</p>
]]></content:encoded>
					
		
		
			</item>
		<item>
		<title>WooCommerce 11.3: Test the New Mobile Gallery</title>
		<link>https://www.mobiletech.mobi/woocommerce-mobile-gallery/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Mon, 16 Nov 2026 11:05:06 +0000</pubDate>
				<category><![CDATA[Mobile UX]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=594</guid>

					<description><![CDATA[See which WooCommerce gallery gets native swipes, then check thumbnails, variation photos and touch behavior before updating your store.]]></description>
										<content:encoded><![CDATA[<p><a href="https://developer.woocommerce.com/2026/11/11/woocommerce-11-3-0-release-notes/">WooCommerce 11.3.0 shipped on November 11, 2026</a>, with a different feel for mobile product photos: the Product Gallery block now uses native scrolling and scroll snap. Images follow a shopper&#8217;s finger and quick flicks carry momentum. If your store uses this block, test the update on staging now, particularly if shoppers have struggled to move between photos; ship it after the gallery and purchase flow pass.</p>
<table>
<caption>Which gallery needs which action?</caption>
<thead>
<tr>
<th scope="col">Your product page</th>
<th scope="col">Action</th>
</tr>
</thead>
<tbody>
<tr>
<td>Product Gallery block</td>
<td>Stage 11.3; test swipes and selection</td>
</tr>
<tr>
<td>Product Image Gallery + Add to Cart + Options</td>
<td>Plan and test block migration separately</td>
</tr>
<tr>
<td>Classic or third-party gallery</td>
<td>Keep configuration; test its own slider</td>
</tr>
<tr>
<td>Only one product image</td>
<td>Choose a multi-image test product</td>
</tr>
</tbody>
</table>
<h2>Find the gallery your shoppers actually use</h2>
<p>In Appearance &gt; Editor, open the product&#8217;s assigned template and inspect List View. <a href="https://woocommerce.com/document/product-gallery-block/">Product Gallery contains Thumbnails and Viewer</a>; it can also sit inside a Single Product block. Check the templates used by both simple and variable products before choosing your test pages.</p>
<p><strong>Product Image Gallery</strong> is the older block and does not support variation-image updates from Add to Cart + Options. For that pairing, migrate on staging if you need variation galleries. Since 11.1, its compatibility warning offers <a href="https://woocommerce.com/document/product-gallery-block/"><strong>Use the Product Gallery block</strong></a>. Replacement uses default settings, so review layout and image cropping before saving. Treat this as a separate change from the version test: first establish how the existing template behaves, then test the replacement.</p>
<p>On a classic template, <a href="https://woocommerce.com/document/adding-product-images-and-galleries/">slider, zoom and lightbox behavior depend on theme support</a>. A builder or extension may replace the gallery. Keep a working replacement&#8217;s configuration during the upgrade and test its own navigation; this block change is not a reason to rebuild that slider.</p>
<h2>What actually changes when you swipe?</h2>
<p>The <a href="https://github.com/woocommerce/woocommerce/pull/69066">native-scrolling fix merged on October 1</a>. Previously, custom touch handlers decided when a drag should advance a photo. Now the browser moves the track and snaps it to a slide. The <a href="https://developer.woocommerce.com/2026/11/11/woocommerce-11-3-0-release-notes/">11.3 release demonstration</a> shows the resulting finger-following movement. This is shipped behavior in Product Gallery, rather than an experimental option to enable.</p>
<p>An <code>IntersectionObserver</code> connects the visible slide to thumbnail and arrow state. There is one deliberate exception during motion: after a thumbnail tap, the destination stays selected while intermediate photos pass. An interrupting swipe returns selection to the visible image. The <a href="https://github.com/woocommerce/woocommerce/pull/69066">September 24 before-and-after recordings and test plan</a> provide a documented example and the sequence below.</p>
<h2>Run one repeatable before-and-after test</h2>
<p>Use staging with the same products, templates and settings before and after updating. Record the initial gallery order and selected image so a reset has a clear reference. Use an iPhone with Safari and an Android phone with Chrome as the default test pair; prioritize additional devices that matter in your store&#8217;s traffic. The upstream author reports Android verification, which is not a promise about every phone.</p>
<ol>
<li><strong>Prepare a simple product with three distinct photos.</strong> For example, use a boot&#8217;s front view, sole and heel, in that order. Set the front view as Product image and add the other two through Products &gt; All Products &gt; Product gallery &gt; Add product gallery images. The <a href="https://woocommerce.com/document/adding-product-images-and-galleries/">main Product Image is already the first gallery item</a>; do not add it twice.</li>
<li><strong>Drag slowly, then flick.</strong> Watch for movement under the finger, a clean resting position and the correct selected thumbnail. Also scroll the page vertically with your finger starting on the gallery: shoppers must be able to reach the size and purchase controls below it.</li>
<li><strong>Tap from the first thumbnail to the last.</strong> Let the jump finish, then repeat and interrupt it with a swipe. Judge the selection once it settles, using the outcomes in the figure. Use next/previous controls to check the return journey and their endpoint states.</li>
<li><strong>Switch a variable product and reset it.</strong> Use a product with distinct orange and black image sets. Choose orange, browse its photos, then choose black and browse again. Clear the options. Expect the parent product&#8217;s gallery to return, with the selected thumbnail and arrows consistent with the visible photo. Compare with your baseline and check the assigned variation images if the set is wrong.</li>
</ol>
<figure>
<svg xmlns="http://www.w3.org/2000/svg" id="woo-swipe-flow" width="100%" viewbox="0 0 370 270" role="img" aria-labelledby="woo-title woo-desc">
<title id="woo-title">Pass or hold: an interrupted jump in a three-photo gallery</title>
<desc id="woo-desc">Start on photo one and tap thumbnail three. Without interruption the third image and thumbnail should match. If an interrupting swipe settles on photo two, thumbnail two is a pass and thumbnail three means hold the update.</desc>
<rect x="1" y="1" width="368" height="268" rx="14" fill="#edf8f3" stroke="#477265"></rect>
<text x="17" y="35" font-size="21" font-family="sans-serif" font-weight="bold" fill="#17362d">Start on photo 1 of 3</text>
<text x="17" y="66" font-size="18" font-family="sans-serif" fill="#17362d">Tap thumbnail 3.</text>
<line x1="17" y1="86" x2="352" y2="86" stroke="#477265"></line>
<text x="17" y="119" font-size="19" font-family="sans-serif" font-weight="bold" fill="#17362d">PASS: no interruption</text>
<text x="17" y="149" font-size="17" font-family="sans-serif" fill="#17362d">Expect photo 3 and thumbnail 3.</text>
<line x1="17" y1="170" x2="352" y2="170" stroke="#477265"></line>
<text x="17" y="205" font-size="19" font-family="sans-serif" font-weight="bold" fill="#17362d">After interruption: photo 2</text>
<text x="17" y="234" font-size="17" font-family="sans-serif" fill="#17362d">Thumbnail 2: pass. Thumbnail 3: hold.</text>
</svg>
<figcaption>Illustrative three-photo verdicts. Source: the <a href="https://github.com/woocommerce/woocommerce/pull/69066">upstream test plan</a>. The second path assumes the swipe settles on photo 2; another final photo is valid if its thumbnail agrees.</figcaption>
</figure>
<p>Suppose the interrupted jump in this three-photo fixture stops on the sole, photo 2. If thumbnail 2 is selected, that check passes; the flick did not have to finish at the heel. If thumbnail 3 stays selected, hold the update for this template and investigate the mismatch. A passing simple product still needs the variable-product check: <a href="https://developer.woocommerce.com/2026/09/03/wc-11-1-release-notes/">variation galleries became a standard feature in 11.1</a>, and the new motion must work with their changing image sets.</p>
<h2>Check resizing and a route without swiping</h2>
<p>Rotate the phone after selecting a later image. The <a href="https://github.com/woocommerce/woocommerce/pull/69066/files">implementation realigns the selection when viewer width changes</a>. It uses horizontal overflow and mandatory snap points, with a 0.51 visibility threshold to avoid selecting two half-visible slides. If you maintain custom gallery CSS, compare overrides affecting overflow, widths and snap behavior before changing JavaScript. For a right-to-left storefront, include thumbnail jumps in that layout; the scrolling code accounts for direction.</p>
<p>Keep visible thumbnails or next/previous buttons as the default. Check that each can be tapped without hitting its neighbor, and that the selected state is clear. A keyboard test alone does not establish a usable tap alternative: <a href="https://www.w3.org/WAI/WCAG22/Understanding/dragging-movements.html">WCAG treats keyboard access and single-pointer alternatives separately</a>. Native browser scrolling is exempt from the dragging requirement, but custom gesture handling can change that distinction.</p>
<p>Keep your existing zoom and pop-up settings unless the test exposes a problem. In the pop-up, test navigation, the close button, Escape and focus returning to the opener. On a keyboard, check that focus is visible and every photo is reachable. Repeat thumbnail navigation with reduced motion enabled: <a href="https://github.com/woocommerce/woocommerce/pull/69066/files">programmatic navigation uses instant movement for that preference</a>. Do not require a visible smooth animation as the pass condition in that run.</p>
<p>If the store already uses video, add one check: select the clip and then swipe away. <a href="https://woocommerce.com/document/product-gallery-block/">Only the selected gallery video should play</a>. Video remains experimental, off by default and limited to uploaded media; leave it off during this upgrade if you do not use it. Enabling another feature would make a new failure harder to isolate.</p>
<h2>If the gallery fails, isolate the first mismatch</h2>
<p>Record the staging URL, product, selected variation, phone, browser version and exact gesture. Make a short recording showing both the large image and its highlighted thumbnail. “The heel stays selected after the sole stops moving” gives the template owner something reproducible; “the swipe feels wrong” does not.</p>
<p>On staging, compare the same product in an uncustomized Product Gallery with its real template. If only the real template fails, change one gallery-related theme override or extension at a time to locate the conflict. If both fail, report the minimal case to WooCommerce. Keep the failing configuration available for diagnosis and defer the affected rollout. Desktop emulation helps locate a layout problem; return to the phone to verify the gesture fix.</p>
<h2>Finish the update before reopening checkout</h2>
<p>WooCommerce 11.3 requires a database update. Back up the database and <code>wp-content</code>, test the intended software versions on staging, and deploy that same update set. After updating the plugin, use <strong>Update WooCommerce Database</strong> when prompted. <a href="https://woocommerce.com/document/how-to-update-woocommerce/">View progress opens Scheduled Actions</a>; wait for the database-update-complete notice before treating the upgrade as finished.</p>
<p>During the live maintenance window, prevent checkout while files and database updates run. Before reopening, check the product gallery, cart and a test purchase, including the shipping, tax, payment and email behavior the store relies on. If a failure blocks buying, keep checkout closed and follow the store&#8217;s recovery procedure.</p>
]]></content:encoded>
					
		
		
			</item>
		<item>
		<title>Yahoo Scout and the ILC: A Publisher’s Guide</title>
		<link>https://www.mobiletech.mobi/yahoo-scout-ilc-publishers/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Sun, 15 Nov 2026 11:12:01 +0000</pubDate>
				<category><![CDATA[Mobile Publishing]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=591</guid>

					<description><![CDATA[See what Yahoo Scout’s first ILC license means for publishers, how to scope an RSL declaration, and what to check before opting in.]]></description>
										<content:encoded><![CDATA[<p>On November 12, 2026, Yahoo Scout became the <a href="https://internetcollective.org/press/ilc-launch-2026">first commercial licensee of the Internet Licensing Collective</a>. For an independent publisher, there&#8217;s now a route to license selected work for use in an AI answer product without negotiating directly with Yahoo. Start with a section you have the rights to license. Treat it as a limited licensing pilot, with no forecast income until you have payment terms you can use.</p>
<table>
<caption>What should a publisher do first?</caption>
<thead>
<tr>
<th scope="col">Catalog</th>
<th scope="col">First move</th>
</tr>
</thead>
<tbody>
<tr>
<td>Original reporting in one section</td>
<td>Check terms for a narrow pilot</td>
</tr>
<tr>
<td>Original and syndicated work mixed</td>
<td>Separate eligible URLs first</td>
</tr>
<tr>
<td>Subscriber guide or private archive</td>
<td>Keep the paywall; review delivery</td>
</tr>
<tr>
<td>Existing direct AI license</td>
<td>Compare coverage before opting in</td>
</tr>
<tr>
<td>No clear rights to reuse</td>
<td>Leave those works outside</td>
</tr>
</tbody>
</table>
<h2>What Yahoo is signing up for</h2>
<p>The collective, formerly the RSL Collective, launched under its new name with more than 3,000 media brands, content owners and platforms, a launch figure reported by the collective. That count describes the launch group, not a count of websites already earning Scout royalties. An AI company takes one agreement for participating work. For publishers, <a href="https://internetcollective.org/publishers">membership is free and nonexclusive</a>. Publishers choose which content and uses to include and opt in to specific licenses. A direct deal stays in place; if its scope overlaps with a proposed license, compare the terms before you enroll the same catalog.</p>
<p>The agreement arrived two days after <a href="https://www.yahooinc.com/press/after-delivering-more-than-3-billion-answers-yahoo-scout-exits-beta-to-underpin-all-of-yahoo-search">Yahoo moved Scout out of beta</a> and introduced Pivot Search, which lets readers switch between traditional results and Scout answers. That experience is available on desktop and mobile web in the US, UK, Europe and Switzerland; the Scout app is available in the US on iOS and Android. Those are distribution details for Yahoo Search. The commercial territory and covered surfaces of the Scout license remain undisclosed. For a publisher with UK and US editions, ask for the covered markets before combining the two editions&#8217; results in a revenue report.</p>
<p>Seeing a page in a crawler log tells you it was fetched. Permission and reported use are recorded elsewhere. An RSL declaration expresses terms. The portal or a participating platform handles enrollment and license choices. A paywall or access rule still controls delivery. The <a href="https://internetcollective.org/platform">collective does not host publisher content</a>; an authorized system gets it from the publisher or through a supplied access URL.</p>
<h2>How to license one section without offering the whole site</h2>
<p>Use a section declaration when the rights follow a URL prefix. The <a href="https://internetcollective.org/docs/publishers/onboard">official section example</a> associates <code>/articles/</code> with the ILC payment standard:</p>
<pre><code>&lt;rsl xmlns="https://rslstandard.org/rsl"&gt;
  &lt;content url="/articles/"&gt;
    &lt;license&gt;
      &lt;permits type="usage"&gt;ai-all&lt;/permits&gt;
      &lt;payment&gt;
        &lt;standard&gt;https://internetcollective.org/license&lt;/standard&gt;
      &lt;/payment&gt;
    &lt;/license&gt;
  &lt;/content&gt;
&lt;/rsl&gt;</code></pre>
<p><a href="https://rslstandard.org/rsl">RSL 1.0 defines <code>ai-all</code></a> to include training, inference, retrieval, indexing, generation and evaluation. Do not deploy that permission unless it matches your intended license. Confirm a narrower configuration with the collective if you want answer generation without training.
</p>
<p>Serve the XML at <code>https://example.com/rsl.xml</code> for this example. Add its discovery line to that host’s <code>robots.txt</code>, preferably before any <code>User-agent</code> group so it is global. RSL also supports group-specific declarations; a matching group&#8217;s own <code>License:</code> entries take precedence over global ones:</p>
<pre><code>License: https://example.com/rsl.xml</code></pre>
<p>Enroll <code>https://example.com/</code> through the publisher portal or a platform partner, confirming licensing authority and any excluded AI systems. Enrollment identifies the host; the declaration identifies the content. For content on <code>news.example.com</code>, give that host its own discovery and enrollment.</p>
<figure>
<svg xmlns="http://www.w3.org/2000/svg" id="ilc-scope" viewbox="0 0 380 274" width="100%" role="img" aria-labelledby="ilc-title ilc-desc">
<title id="ilc-title">Which resource does the example declaration cover?</title>
<desc id="ilc-desc">The publisher enrolls example.com. Its robots file points to rsl.xml, which specifies the articles section. An article in that section matches; a page in the wire archive does not.</desc>
<rect x="1" y="1" width="378" height="272" rx="12" fill="#fcf6ef" stroke="#80666b"></rect>
<text x="17" y="39" font-size="21" font-family="sans-serif" font-weight="bold" fill="#362235">Enroll example.com</text>
<text x="17" y="70" font-size="17" font-family="sans-serif" fill="#362235">robots.txt points to rsl.xml</text>
<text x="17" y="104" font-size="17" font-family="sans-serif" fill="#362235">Declaration scope: /articles/</text>
<line x1="17" y1="122" x2="363" y2="122" stroke="#80666b"></line>
<text x="17" y="154" font-size="18" font-family="monospace" fill="#362235">/articles/flood-map</text>
<text x="17" y="181" font-size="17" font-family="sans-serif" fill="#362235">Matches the section</text>
<text x="17" y="220" font-size="18" font-family="monospace" fill="#362235">/wire/flood-update</text>
<text x="17" y="247" font-size="17" font-family="sans-serif" fill="#362235">Outside this declaration</text>
</svg>
<figcaption>One enrolled host can have a smaller licensed section. The two article paths are illustrative. Source: <a href="https://internetcollective.org/docs/publishers/onboard">the section-scoping example</a>.</figcaption>
</figure>
<p>Suppose a local newsroom owns its reporting on flood maps, and its syndicated updates live under <code>/wire/</code>. The section example covers <code>/articles/flood-map</code> while leaving <code>/wire/flood-update</code> outside this declaration. The newsroom can keep that split if it holds the required rights for <em>everything</em> it offers under <code>/articles/</code>. If a vendor&#8217;s map or photograph on that page is outside its license, hold the page back until the relevant assets are cleared or can be excluded reliably.</p>
<p>For mixed rights within a section, prefer a feed of cleared works until you can verify the finer scope. Private Access URLs can deliver cleared feeds or archives, with a different URL per authorized system. They control delivery; check that the declaration and agreement also express the scope you intend. Keep subscriber-facing access controls in place.</p>
<h3>Check the declaration before enrolling</h3>
<p>Use this preflight on your own host. It checks what you published; enrollment and reported use still need their own records.</p>
<ol>
<li>Fetch the public <code>robots.txt</code> and follow its applicable <code>License:</code> URL. Confirm it returns the intended XML, not a login page.</li>
<li>Parse the XML. Check the RSL namespace, content scope, permissions and payment URL against the accepted terms.</li>
<li>Check both example paths: <code>/articles/flood-map</code> must fall within the intended scope, while <code>/wire/flood-update</code> must remain outside it. Inspect any page, feed or header declarations too; a single file does not reveal every association.</li>
<li>Check access separately. A <code>License:</code> line <a href="https://rslstandard.org/rsl">does not change <code>Allow</code> or <code>Disallow</code></a>. Verify the intended system can retrieve the cleared material through its agreed route.</li>
</ol>
<p>If a check fails, pause enrollment and correct the declaration or delivery route. For a live mismatch, contact the collective about the affected works and reporting period before changing permissions. A homepage logo, a crawler visit or a public licensing pointer alone cannot confirm an active, paying license. Buying a placement on a participating site does not buy an AI citation; the licensing model pays for contributions the AI product actually uses.</p>
<h2>What counts toward payment?</h2>
<p>The <a href="https://internetcollective.org/license">collective&#8217;s published model</a> ties the license fee to the AI product and divides eligible revenue according to how the work contributes, how often it is used, and the value of the product and market. Its own example asks why home insurance premiums are rising. The illustrated answer combines severe-weather losses and rebuilding costs with reinsurance costs, then names the fictional Coastal Ledger in an explanation of insurers leaving high-risk regions. Primary reporting supplies the substance; supporting material adds context; background knowledge can count without a citation. Naming a publication adds value in the model. This is an illustration of allocation, not a captured Scout answer or a payout calculation.</p>
<p>Payments are allocated <a href="https://internetcollective.org/license">within the relevant product and market</a>. Yahoo usage does not entitle a publisher to revenue from an unrelated AI product that never uses its work. When assessing a report, ask which product and market its pool represents.</p>
<p>The collective <a href="https://internetcollective.org/press/ilc-launch-2026">projects its operating costs</a> at or below 10% of license revenue. At 10%, $100 in gross licensing revenue would leave $90 for the pool before it is allocated among publishers. That leaves each publisher’s share to be calculated under the allocation rules. Yahoo’s fee, usage weightings and minimum payout remain undisclosed. Leave expected income blank until the contract or statements support a figure. </p>
<p>The separately published <a href="https://www.mathereconomics.com/price-of-intelligence/">Mather Economics estimate of 15%–35% of AI-company revenue</a> models a potential market, rather than the Yahoo contract. The study does not establish Scout’s actual rate or your share.</p>
<p>Licensees report use and pay the collective, which allocates royalties. The <a href="https://internetcollective.org/faq">publisher dashboard shows reported usage by website</a>, earnings and payment history. Your site analytics answer a different question: how many readers arrived, where they landed and what they did. A citation can appear without a click, and background use can happen without a citation. Keep referral visits beside the licensing statement rather than dividing one by the other to guess a per-click rate.</p>
<h2>Start with a catalog you can check</h2>
<p>Pick a section small enough to reconcile with a statement. Before opting in, ask for a sample report, the payout schedule and any minimum, the covered products and markets, and the procedure for removing a work or a licensee. Read the agreement for how a change applies to previously acquired copies and outstanding reporting. Account choices can be updated. Removing a discovery line is not an adequate withdrawal plan: get written confirmation of the effective date and treatment of earlier copies. If those answers matter to your catalog and aren&#8217;t available, keep it out for now.</p>
<p>Save the accepted scope and terms, the deployed declaration, and a note naming the hosts and sections included. When the first statement arrives, check that it identifies the correct website and period and that excluded material still stays outside the declared scope. Expand the pilot only when you can explain what you offered, what the report measures and where the money came from.</p>
]]></content:encoded>
					
		
		
			</item>
		<item>
		<title>Silktide V4 vs V5: How to Read Your November Score</title>
		<link>https://www.mobiletech.mobi/silktide-v5-accessibility/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Sat, 14 Nov 2026 11:08:46 +0000</pubDate>
				<category><![CDATA[Mobile UX]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=588</guid>

					<description><![CDATA[Compare Silktide V4 and V5 correctly, trace changes to the November crawl, and verify the mobile accessibility issues worth fixing.]]></description>
										<content:encoded><![CDATA[<p>On November 13, 2026, the <a href="https://silktide.com/blog/silktide-accessibility-index-november-2026-update/">Silktide Accessibility Index announced a new testing engine and a second score</a>. Its November edition shows V4, the historical measure, alongside V5, which starts a new reporting series. Both still run from 0 to 100, but they assign different importance to the same findings. Start with the score label: two different November numbers can describe the same scan. If V4 itself moved since October, check which pages the scanner could reach and what it found on them.</p>
<table>
<caption>Which comparison answers your question?</caption>
<thead>
<tr>
<th scope="col">Signal</th>
<th scope="col">Check</th>
<th scope="col">Action</th>
</tr>
</thead>
<tbody>
<tr>
<td>November V4 differs from V5</td>
<td>Scoring and PDF scope</td>
<td>Keep separate reporting series</td>
</tr>
<tr>
<td>V4 moved since October</td>
<td>Coverage and changed checks</td>
<td>Reproduce the affected finding</td>
</tr>
<tr>
<td>V5 alone is missing</td>
<td>V5 recording gap</td>
<td>Leave blank; check next test</td>
</tr>
<tr>
<td>Neither score appears</td>
<td>Site access and crawl logs</td>
<td>Check refusals and challenges</td>
</tr>
<tr>
<td>A phone task breaks</td>
<td>The failing control</td>
<td>Fix and retest the task</td>
</tr>
</tbody>
</table>
<h2>One November scan, two ways to count</h2>
<p>Keep your V4 time series for reports that go back before November. Start a separate V5 series with the November value. For a comparison with a peer, use the same edition, category and score version. The <a href="https://index.silktide.com/">Index</a> runs monthly and covers nearly 15,000 sites, but the rank is relative to the sites in the chosen category. Your rank can move while your own score barely changes.</p>
<p>Both versions apply automated WCAG 2.2 checks to the same November scan. The <a href="https://docs.silktide.com/en-GB/silktide-index/scores">public scoring rules</a> give V4 four weighted categories: A (45), AA (40), AAA (15) and mobile (10). Those are relative weights, not four percentages of the final score. V4 scores each level on its own checks, so its AA result can exceed its A result. V5&#8217;s level results are cumulative: AA includes A, and AAA includes both. A V5 AA score therefore cannot exceed its A score.</p>
<p><a href="https://docs.silktide.com/en-GB/silktide-index/scores">V5 starts at 100 and deducts points</a> for each check according to the severity of the check result and the check&#8217;s maximum impact. A check is counted once in the overall score even if it belongs to several levels. The largest-impact checks retain their previous impact; the other impacts draw on priority and how confident the test is that it found a real issue. Repeated failures on a shared component can affect a check more than a single instance on one page.</p>
<p>The calculation is <code>deduction = impact × (100 − check score) / 100</code>. The official example uses a check with a maximum impact of 10 and a result of 70 out of 100. The shortfall is 30%, so V5 takes off 3 points. If that&#8217;s the only deduction, the total is 97. Put that figure in the November V5 column. Comparing it with an older V4 result would mix a new scoring rule with whatever happened to the site between scans.</p>
<p>There are differences in coverage too. Public V4 includes PDFs, while public V5 scores web pages only: its newer PDF checks do not run in the Index. If you maintain forms or guides as PDFs, keep a separate document queue even when V5 improves. The public scan sees the site as a first-time visitor and ignores exceptions your team approved in its own account. A private account can also scan more pages and run additional checks. Label exports accordingly.</p>
<h3>What if V5 is missing?</h3>
<p>Leave the V5 value blank. When V4 is present but V5 is absent, the V5 result was not recorded for that test; the site has no V5 rank that month and should reappear after its next test. Check that next result before treating the gap as a crawling problem.</p>
<p>V4 is still displayed first. Its <a href="https://docs.silktide.com/en-GB/silktide-index/scores">retirement is planned for 2027</a>, with advance notice and no fixed date yet. Keep both series until a firm switchover date is announced.</p>
<h2>Why V4 can move even if the site stayed put</h2>
<p>The V4 formula stayed the same, and Silktide expects most V4 results to remain close to October’s. October was the last Index edition on the previous engine. November retested every site from scratch with the engine already used for customer accounts since October 2026. It can load and test a different set of pages. Most sites are sampled on 25 pages, with an upper limit of 125, and the sample can change month to month. After a redesign, inspect the templates that changed. If you didn&#8217;t deploy anything, check coverage and the underlying finding before filing a regression.</p>
<p>Start with the report&#8217;s scan date and page count. Then check your release log for new templates, banners or widgets that ran during the scan. When you have the tested URLs, compare the common pages and look at which kinds of pages entered or left the sample. Twenty-five pages can include mostly articles one month and more product or category pages the next. If you can&#8217;t recover the original sample, use a fixed internal set for the next comparison and keep the public series labeled as a sample.</p>
<p>The <a href="https://docs.silktide.com/en-GB/silktide-index/november-2026-changes">new engine backs off after refusals</a>: it halves concurrent requests, waits and retries. It recognizes browser challenges, records pages it still can&#8217;t load as rate limited and leaves them out of the score. It stops waiting for slow decorative backgrounds and records why remaining pages cannot be tested. The older engine could end up testing an error page or fail to reach enough pages to produce a score. Silktide identifies blocked or throttled automated visits as the likeliest cause of a sharp V4 move; check that branch first when your release log is quiet. In a vendor test of 500 Index sites, 37 that previously lacked enough pages became scorable; 13 moved in the opposite direction. If both scores disappear, check CDN and firewall logs around the scan window for 429 responses or challenge pages. The scanner now comes from different servers. Confirm what was served before changing a security rule.</p>
<p>The <a href="https://docs.silktide.com/en-GB/silktide-index/november-2026-changes">corrected checks can move either score</a> even on pages that loaded successfully. Use the changed check to choose the retest, rather than opening a generic “score dropped” ticket.</p>
<table>
<caption>Which November correction matters to your site?</caption>
<thead>
<tr>
<th scope="col">Feature</th>
<th scope="col">Change</th>
<th scope="col">Action</th>
</tr>
</thead>
<tbody>
<tr>
<td>Duplicate IDs or malformed lists</td>
<td>4.1.1 exits score</td>
<td>Review markup separately</td>
</tr>
<tr>
<td><code>color-mix()</code> or wide-gamut colors</td>
<td>Correct color values</td>
<td>Recheck previously flagged text</td>
</tr>
<tr>
<td>Open shadow roots</td>
<td>Checks and links inside</td>
<td>Inspect newly tested components</td>
</tr>
<tr>
<td>Visible CSS child</td>
<td>Now tested</td>
<td>Inspect the visible child</td>
</tr>
<tr>
<td>Small, separated controls</td>
<td>Correct spacing geometry reduces false failures</td>
<td>Measure nearby targets</td>
</tr>
<tr>
<td><code>::after</code> focus ring</td>
<td>Ring detected</td>
<td>Recheck the flagged focus state</td>
</tr>
<tr>
<td>Silktide cookie banner</td>
<td>Hidden-focus false flag fixed</td>
<td>Test focus inside banner</td>
</tr>
<tr>
<td>PDF image alt text</td>
<td>Check removed</td>
<td>Keep auditing documents</td>
</tr>
</tbody>
</table>
<h2>Use a target-size finding to pick the first mobile fix</h2>
<p>The <a href="https://index.silktide.com/accessibility/website/warehouse/november-2026">November report for Warehouse</a>, a UK fashion retailer, shows what a public result gives you. The scan assessed 25 pages on November 6. Under Recommendations it lists “Ensure interactive components are far enough apart,” WCAG 2.2 AA 2.5.8, with progress of 2%. That progress value describes the tool’s check result across the tested sample, rather than the proportion of shoppers who can use the site. It identifies a class of touch controls to investigate. To act on a similar report for your own site, find a representative page and reproduce the flagged control at a narrow viewport.</p>
<p>The November engine now uses the <a href="https://www.w3.org/WAI/WCAG22/Understanding/target-size-minimum.html">24 CSS pixel circle in the WCAG spacing test</a>. For an undersized target, center the circle on its bounding box; it must avoid other targets and the circles around other undersized targets. If your isolated mobile toolbar has two 18 by 18 CSS pixel buttons with centers 20 pixels apart, their circles overlap by 4 pixels. Move their centers to 28 pixels apart and those circles have a 4-pixel gap. That clears the spacing test for this pair, assuming no other controls intersect. For frequent actions, enlarge the actual tap area to at least 24 by 24 CSS pixels by default; use spacing when the layout prevents that. The icon itself can stay small.</p>
<figure>
<svg xmlns="http://www.w3.org/2000/svg" id="silktide-target-spacing" viewbox="0 0 360 288" width="100%" role="img" aria-labelledby="spacing-title spacing-desc">
<title id="spacing-title">Target spacing for two undersized controls</title>
<desc id="spacing-desc">Two 18 by 18 CSS pixel square targets. Circles of diameter 24 overlap by four pixels at center spacing of 20, and leave a gap of four at spacing of 28.</desc>
<rect x="1" y="1" width="358" height="286" rx="14" fill="#fff8e8" stroke="#6f5862"></rect>
<text x="17" y="32" font-size="19" font-family="sans-serif" font-weight="bold" fill="#33212d">20 px apart: fix spacing</text>
<rect x="59" y="56" width="54" height="54" fill="#e5c5f1" stroke="#35213a"></rect>
<rect x="119" y="56" width="54" height="54" fill="#e5c5f1" stroke="#35213a"></rect>
<circle cx="86" cy="83" r="36" fill-opacity="0" stroke="#713232" stroke-width="2"></circle>
<circle cx="146" cy="83" r="36" fill-opacity="0" stroke="#713232" stroke-width="2"></circle>
<text x="191" y="88" font-size="17" font-family="sans-serif" fill="#33212d">4 px overlap</text>
<line x1="16" y1="132" x2="344" y2="132" stroke="#6f5862"></line>
<text x="17" y="161" font-size="19" font-family="sans-serif" font-weight="bold" fill="#33212d">28 px apart: circles clear</text>
<rect x="59" y="185" width="54" height="54" fill="#e5c5f1" stroke="#35213a"></rect>
<rect x="143" y="185" width="54" height="54" fill="#e5c5f1" stroke="#35213a"></rect>
<circle cx="86" cy="212" r="36" fill-opacity="0" stroke="#713232" stroke-width="2"></circle>
<circle cx="170" cy="212" r="36" fill-opacity="0" stroke="#713232" stroke-width="2"></circle>
<text x="219" y="218" font-size="17" font-family="sans-serif" fill="#33212d">4 px gap</text>
<text x="17" y="270" font-size="15" font-family="sans-serif" fill="#33212d">18 × 18 px targets · 24 px circles</text>
</svg>
<figcaption>Illustrative pair of targets, drawn at the same scale. Clearing this pair does not clear its other neighbors. Source: <a href="https://www.w3.org/WAI/WCAG22/Understanding/target-size-minimum.html">WCAG 2.2 target size and spacing</a>.</figcaption>
</figure>
<p>The <a href="https://www.w3.org/WAI/WCAG22/Understanding/target-size-minimum.html">other exceptions to 2.5.8</a> cover an equivalent conforming control on the same page, inline targets, unmodified browser-controlled targets, and an essential or legally required presentation. Check these before redesigning a flagged element: a link within a sentence is a different case from a toolbar button. CSS pixels are layout units, not the phone display’s physical pixels.</p>
<p>Check the live state where the control matters. A quantity button can have enough room on the product page and sit too close to Remove once the mobile cart is open. Measure the clickable element, not the icon. If you fix a shared component, test it in the page that raised the finding and in another template that uses it. For focus findings, tab through the opened dialog and back to its trigger; an <a href="https://www.mobiletech.mobi/intera11y-interactive-web-accessibility-checks/">interactive accessibility check</a> captures failures a snapshot can&#8217;t.</p>
<p>A public result of 100 means the scan found no issues it could score. Checks that require judgment don&#8217;t contribute to either number. Use <a href="https://www.w3.org/WAI/test-evaluate/">human evaluation</a> to decide whether someone can complete the journey, and keep the benchmark as a separate reporting measure. In the issue ticket, record the month and score version, the page and viewport, the check that changed, and the action a reader or shopper couldn&#8217;t finish. Assign the fix to the owner of that component and repeat the same task after the change.</p>
]]></content:encoded>
					
		
		
			</item>
		<item>
		<title>Google Search Referrers in Mobile Traffic</title>
		<link>https://www.mobiletech.mobi/google-search-referrer/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Fri, 13 Nov 2026 11:06:20 +0000</pubDate>
				<category><![CDATA[Mobile SEO]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=585</guid>

					<description><![CDATA[Trace Google Search referrals from Android, iOS and the web, and find out why Search Console clicks and analytics sessions can diverge.]]></description>
										<content:encoded><![CDATA[<p>On November 12, 2026, Google <a href="https://developers.google.com/search/updates">published a guide to recognizing Search traffic in analytics and server logs</a>. It lists what a click can send from the web, the Google app on Android and the Google app on iOS. If your mobile traffic looks down in analytics but steady in Search Console, start with the request for the page people land on. This is documentation of existing signals, so a correctly classified site needs no migration. A new reporting gap deserves a trace before it prompts changes to content or rankings strategy.</p>
<p>Look for the HTTP <code>Referer</code> header. Search uses the <a href="https://developers.google.com/search/docs/monitor-debug/identify-google-search-traffic"><code>origin</code> policy</a>. In the <a href="https://w3c.github.io/webappsec-referrer-policy/#referrer-policy-origin">specification&#8217;s example</a>, a link from <code>example.com/page.html</code> to another HTTPS site carries only <code>https://example.com/</code>. The search result URL is trimmed the same way: the host can come through, but the path and search terms don&#8217;t. Find the queries in Search Console&#8217;s Performance report instead. The request header is spelled <code>Referer</code>; the controlling response header is <code>Referrer-Policy</code>.</p>
<table>
<caption>Where to look when mobile Search numbers disagree</caption>
<thead>
<tr>
<th scope="col">Signal</th>
<th scope="col">Check</th>
<th scope="col">Action</th>
</tr>
</thead>
<tbody>
<tr>
<td>Clicks down too</td>
<td>Web report, mobile</td>
<td>Compare pages and queries</td>
</tr>
<tr>
<td>Only sessions down</td>
<td>First page request</td>
<td>Trace one click</td>
</tr>
<tr>
<td>Referrer reaches final page</td>
<td>Session report</td>
<td>Check tag and rules</td>
</tr>
<tr>
<td>First referrer blank</td>
<td>Same phone and app</td>
<td>Retest; leave source open</td>
</tr>
</tbody>
</table>
<h2>What a Google click can leave in the logs</h2>
<p><a href="https://developers.google.com/search/docs/monitor-debug/identify-google-search-traffic">Web Search referrers include</a> <code>https://www.google.com/</code> and local domains such as <code>https://www.google.co.uk/</code> and <code>https://www.google.co.jp/</code>. The Google app on iOS uses the same kind of value. A UK host in the header does not tell you the visitor&#8217;s country, and the common origin doesn&#8217;t tell you whether they came from the iOS app or a browser.</p>
<p>Android can leave a more specific clue. When the Google app opens a browser tab, the app can send the following documented value:</p>
<pre><code>android-app://com.google.android.googlequicksearchbox/https/www.google.com</code></pre>
<p>Parse the scheme, app authority and path together as an Android Google app signal. A rule that only accepts HTTPS web origins would miss it. The app can also send the ordinary Google origin, so counting app URIs alone will miss some Android Google app traffic.</p>
<figure>
<svg xmlns="http://www.w3.org/2000/svg" id="referrer-trace" viewbox="0 0 440 410" width="100%" role="img" aria-labelledby="referrer-title referrer-desc">
<title id="referrer-title">Example: a Google app referrer changes at a client-side hop</title>
<desc id="referrer-desc">Hypothetical trace. The first article request has the documented Android Google app URI. After a JavaScript redirect, the final document request has the publisher's origin. Inspect that hop, then repeat the same route and verify analytics.</desc>
<rect x="1" y="1" width="438" height="408" rx="13" fill="#eef4fa" stroke="#627e96"></rect>
<text x="20" y="32" font-size="20" font-family="sans-serif" font-weight="bold" fill="#142e43">Example trace, not measured traffic</text>
<text x="20" y="68" font-size="19" font-family="sans-serif" font-weight="bold" fill="#142e43">1. First document: /old-story</text>
<text x="20" y="95" font-size="16" font-family="monospace" fill="#142e43">Referer: android-app://</text>
<text x="20" y="118" font-size="14" font-family="monospace" fill="#142e43">com.google.android.googlequicksearchbox</text>
<text x="20" y="141" font-size="16" font-family="monospace" fill="#142e43">/https/www.google.com</text>
<line x1="20" y1="157" x2="420" y2="157" stroke="#627e96"></line>
<text x="20" y="187" font-size="19" font-family="sans-serif" font-weight="bold" fill="#142e43">2. JavaScript hop to /story</text>
<text x="20" y="214" font-size="16" font-family="monospace" fill="#142e43">Referer: https://publisher.example/</text>
<text x="20" y="240" font-size="17" font-family="sans-serif" fill="#142e43">The request arrived; its source changed.</text>
<line x1="20" y1="259" x2="420" y2="259" stroke="#627e96"></line>
<text x="20" y="290" font-size="19" font-family="sans-serif" font-weight="bold" fill="#142e43">3. Inspect this hop first</text>
<text x="20" y="319" font-size="17" font-family="sans-serif" fill="#142e43">Remove an unnecessary client-side hop.</text>
<text x="20" y="346" font-size="17" font-family="sans-serif" fill="#142e43">Repeat the click on the same phone.</text>
<text x="20" y="373" font-size="17" font-family="sans-serif" fill="#142e43">Verify the tag and session source.</text>
</svg>
<figcaption>Hypothetical request trace using the documented Android value and a possible redirect failure. Line breaks split one incoming header for readability. Source: <a href="https://developers.google.com/search/docs/monitor-debug/identify-google-search-traffic">Search referrers and redirect effects</a>.</figcaption>
</figure>
<h2>Count landing requests before counting referrals</h2>
<p>If you already use a standard analytics product and Google traffic is classified as expected, leave its channel rules alone. Build a separate count in logs only when you need to trace a missing or mislabeled visit. Use the log layer that actually saw the request. If a CDN answered from cache, an origin log alone can miss the landing entirely.</p>
<p>Filter for requests for the HTML entry pages you want to measure. A phone that lands on an article can then request images, scripts and stylesheets; those requests may carry the article&#8217;s own URL as their referrer. Counting every request swells the denominator with assets. Keep redirects and their destinations in the trace, but avoid counting both as separate landings.</p>
<p>For web referrers, compare the parsed origin to an explicit list of Google origins relevant to the sites you operate. For example, accept an exact parsed origin of <code>https://www.google.com</code> and explicitly supported local origins; reject <code>google.com.example.org</code>. A substring test for <code>google.com</code> would accept that unrelated host. Keep the list aligned with your audience&#8217;s observed Google domains, rather than assuming these three examples are exhaustive. Handle the documented Android value separately and leave blanks unclassified.</p>
<p>Keep a separate bucket for known paid traffic: a Google origin on its own isn&#8217;t evidence that the click was organic. Search Console excludes Google Ads clicks. If a redirect drops ad markers such as <code>gclid</code>, paid visits can lose their paid classification and inflate another channel.</p>
<p>Keep the landing route on HTTPS. The <a href="https://w3c.github.io/webappsec-referrer-policy/#referrer-policy-origin"><code>origin</code> policy can send the origin even on an HTTPS-to-HTTP navigation</a>, while strict variants suppress it on that downgrade. Do not weaken the site&#8217;s privacy policy to force attribution; inspect the particular redirect that changes the signal.</p>
<h2>Follow one click from the phone to the session report</h2>
<p>Open the actual result in the app or browser where the change shows up. Record the time with its time zone, the phone OS and app/browser version, and the clicked URL. Keep the complete route, including any result-link wrapper. In a network trace or the matching server/CDN records, compare the <code>Referer</code> on the first document request with the header on the final document that runs your tag. Pay particular attention to region selectors, mobile interstitials and old URLs that send people to replacements. A JavaScript or meta refresh redirect can make your own site appear as the referrer. A stricter policy on a redirect can remove it, and a redirect that discards the query string can lose paid attribution even if the Google host survives.</p>
<p>Next inspect <code>document.referrer</code> in the final page&#8217;s browser console, then check whether the analytics tag runs and which channel gets the session. The JavaScript value can be blocked even when a request header was sent; inspect both. </p>
<p>In GA4, open <a href="https://support.google.com/analytics/answer/12923437">Reports → Acquisition → Traffic acquisition</a> and choose <code>Session source / medium</code>. Look for <code>google / organic</code> and inspect referral or direct rows for the same landing page if it is missing. Those rows identify what GA4 classified, not the source of every unassigned visit. Check custom channels, referral exclusions and report filters before rewriting your log matcher. If the tag behaves differently after a consent choice, <a href="https://www.mobiletech.mobi/mobile-tracker-audit-consent-journeys/">replay each consent path in a fresh visit</a> so the previous choice doesn&#8217;t carry over.</p>
<p>If the first request already has a blank referrer, the destination site can&#8217;t recover the source from that header. <a href="https://developers.google.com/search/docs/monitor-debug/identify-google-search-traffic">Private browsing and anti-fingerprinting, strict settings or extensions, opening results in a new tab in some browsers, and a native search widget handing a link to an external browser</a> can all leave that gap. Repeat the click on the same phone in a normal browsing session. A working control helps isolate the conditions of the gap; it doesn&#8217;t identify the source of all the unlabeled visits.</p>
<h2>What if Search Console clicks are steady but sessions fall?</h2>
<p>Check the landing and measurement path before changing a page&#8217;s SEO. Start in the <a href="https://support.google.com/webmasters/answer/7576553">Search results Performance report</a> with Web search and the mobile device filter. Compare the same dates and relevant country and pages on both sides. <a href="https://developers.google.com/search/docs/monitor-debug/google-analytics-search-console">Align the reporting scope</a>: Search Console normally uses Pacific Time, while GA4 uses its property time zone. Search Console generally credits the canonical page; Analytics sees the tagged destination URL. In GA4, compare <code>google / organic</code> sessions for mobile devices and the corresponding landing pages, not all organic sources or all Google channels. Group parameter variants with their canonical article before comparing totals.</p>
<p>Search Console counts an organic click; client analytics needs the visitor to reach the page and the tracking to run. Closing a tab before the page and tag finish loading can leave a click without a measured session. A small, stable gap is normal; investigate a new or widening divergence rather than trying to make the totals identical.</p>
<p>Also check <a href="https://developers.google.com/search/docs/monitor-debug/google-analytics-search-console">non-HTML destinations and bot handling</a> before comparing site-wide totals. A PDF can receive Search clicks without running the page&#8217;s analytics tag, and raw request counts need their own bot filtering. For a typical article site, start with one affected HTML route so these differences don&#8217;t obscure the handoff.</p>
<p>Say a publisher sees 1,000 mobile Search Console clicks in each of two comparable periods, while GA4 shows Google organic sessions falling from 820 to 610. That&#8217;s 210 fewer measured sessions, about a 26% drop from the earlier session count. In this example, an Android test records the Google app URI at the first request but the publisher&#8217;s own origin at the final page after a newly added JavaScript redirect. Repair that route first by removing the unnecessary client-side hop. Keep required redirects on HTTPS and retain the query parameters your attribution setup needs. Then repeat the same result click, verify the request and JavaScript referrers, and confirm the expected session source under the same consent choice. The test shows a break you can repair; it does not tell you how many of the 210 sessions it explains.</p>
<h2>What you can report with confidence</h2>
<p>A common Google origin can cover Search, Discover or Google Images. Use the separate Discover performance report for feed traffic and the Image search type for image results; a log matcher cannot split those shared origins. <a href="https://developers.google.com/search/docs/appearance/ai-features">AI Overviews and AI Mode are included in Search Console&#8217;s Web data</a>, without a separately documented referrer label in this guide. Report Web Search as a combined total; the common origin cannot produce an AI-only count.</p>
<p>The November 12 documentation gives no minimum browser or app version and no separate US, UK or EU rollout schedule. Test the actual client and consent path serving your audience. It also supplies no correction factor for missing headers or for Android clicks that use the app URI; leave unobserved sources unknown.</p>
<p>When you hand a case to the developer who owns the landing page, give them the original entry URL, the first and final referrer values, the phone and app used, and the session channel if one was recorded. Include the redirect URL if there is one, so they can reproduce the exact route.</p>
]]></content:encoded>
					
		
		
			</item>
		<item>
		<title>CrUX Adds Soft Navigation FCP and Visit Types</title>
		<link>https://www.mobiletech.mobi/crux-soft-navigation-fcp/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Thu, 12 Nov 2026 11:10:32 +0000</pubDate>
				<category><![CDATA[Mobile Web]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=582</guid>

					<description><![CDATA[Read the new CrUX paint and navigation metrics on phones, compare them with the existing series, and check which routes need investigation.]]></description>
										<content:encoded><![CDATA[<p>The November 11, 2026 <a href="https://developer.chrome.com/docs/crux/release-notes">Chrome UX Report update</a> adds two experimental measurements that count page changes inside single-page apps. One tracks First Contentful Paint (FCP), the time until visitors first see content after a navigation. The other reports the mix of navigation types. The fields are available through the CrUX APIs and CrUX Vis. The next monthly BigQuery release is scheduled to add them there.</p>
<p>Start with the navigation mix. If most visits to a page come from links inside your app, a paint time that includes those visits answers a different question from the first load alone. A shopper who taps a product card can reuse the app they already have open. Someone arriving at the same product from a search result may need to download and run the app first.</p>
<table>
<caption>Pick the view for the question</caption>
<thead>
<tr>
<th scope="col">Question</th>
<th scope="col">First check</th>
</tr>
</thead>
<tbody>
<tr>
<td>How does a document load paint?</td>
<td>Keep the existing FCP series</td>
</tr>
<tr>
<td>How common are in-app transitions?</td>
<td>Open experimental Navigation Types</td>
</tr>
<tr>
<td>How does including in-app views change FCP?</td>
<td>Compare both FCP series</td>
</tr>
<tr>
<td>Which route needs work?</td>
<td>Query its URL, then measure the route</td>
</tr>
</tbody>
</table>
<h2>What the two fields measure</h2>
<p>A soft navigation keeps the current HTML document and updates the visible URL and page content. A tap on a product card that changes both the address and the rendered view can qualify. A menu expanding in place usually has no new URL to attribute. <a href="https://developer.chrome.com/docs/web-platform/soft-navigations">Detection is enabled by default from Chrome 151</a> and requires a user interaction, a visible URL change and a content paint. The resulting <code>soft-navigation</code> entry exposes <code>paintTime</code> and <code>presentationTime</code> for its first contentful paint. When CrUX has enough data for the destination URL, experimental results can account for the route under that address.</p>
<p>The new FCP variant combines eligible document navigations and detected soft navigations. A contentful paint is the first eligible text, image or other content that gets drawn. The product name can appear well before the product image, so use <a href="https://web.dev/articles/fcp">FCP</a> to investigate the first sign of content and LCP to investigate when the main content appears. FCP is a supporting metric. LCP, INP and CLS make up the <a href="https://web.dev/articles/vitals">Core Web Vitals</a>.</p>
<p>Navigation Types reports how the sampled views are divided. The <a href="https://developer.chrome.com/docs/crux/methodology/metrics#navigation_types">existing breakdown</a> includes regular navigations, cached documents, reloads, history returns, back/forward cache and prerenders. The experimental version adds soft navigations to its mix. On Android, a restored tab is counted as <code>reload</code>, rather than the separate <code>restore</code> category. Read each fraction against its own series: adding soft navigations changes the denominator for the other categories too. If soft navigations account for a large share, the blended number can move even when the document-load experience stays steady. Subtracting the standard p75 from the blended p75 cannot recover a soft-navigation-only paint time: each percentile describes its whole distribution.</p>
<h2>Read two reports for the same phone traffic</h2>
<p>Open <a href="https://cruxvis.withgoogle.com/">CrUX Vis</a>, enter a URL, choose the phone view and open Loading Performance. Expand First Contentful Paint and Navigation Types. Each has an Include soft navigations control. Keep the same URL, device and collection dates while you toggle the feature. Note whether the tool is showing URL or origin data. An origin aggregates eligible pages, which can hide an expensive transition in a small part of the site.</p>
<p>The tool can display FCP even when the Navigation Types graph lacks a point for the same window. On the <a href="https://cruxvis.withgoogle.com/#/?view=loadingperf&amp;url=https%3A%2F%2Fwww.youtube.com%2F&amp;identifier=url&amp;device=PHONE&amp;periodStart=0&amp;periodEnd=-1&amp;display=p75s">YouTube homepage report</a> for phones, standard FCP was 3,940 ms at the 75th percentile for October 11 through November 7. The standard navigation chart displayed no value for that same window when checked. At 3.94 seconds, that standard FCP exceeds the <a href="https://web.dev/articles/fcp">poor threshold of 3 seconds</a>; the good target is at most 1.8 seconds at p75. Investigate the document-navigation paint experience. Keep the navigation mix unknown for this window: the missing point gives no basis for choosing a route-heavy explanation or calculating an experimental improvement.</p>
<h2>Request the new fields from the API</h2>
<p>For a dashboard, send the following JSON body to the <a href="https://developer.chrome.com/docs/crux/api">CrUX API’s <code>queryRecord</code> endpoint</a> using a Google Cloud key enabled for the Chrome UX Report API. Send a POST to <code>/v1/records:queryRecord</code> on <code>chromeuxreport.googleapis.com</code>. Pass your key as the <code>key</code> query parameter and set <code>Content-Type</code> to <code>application/json</code>. Replace the sample URL with an eligible page. Start with a route that users reach both from outside the app and from within it.</p>
<pre><code>{
  "url": "https://example.com/product",
  "formFactor": "PHONE",
  "metrics": [
    "first_contentful_paint",
    "experimental_first_contentful_paint_with_softnavs",
    "navigation_types",
    "experimental_navigation_types_with_softnavs"
  ]
}</code></pre>
<p>Look under <code>record.metrics</code>. Each paint field can return a histogram and <code>percentiles.p75</code> in milliseconds. A p75 of 3,940 ms places about three quarters of the sampled observations at or below that time. Each navigation-types field returns a <code>fractions</code> object. Read the soft-navigation share in the experimental breakdown and multiply a fraction by 100 for a percentage; use it to decide how much weight in-app views carry. Preserve the field names in stored data and retain a missing metric as missing. If the URL is too sparse, replace <code>url</code> with <code>origin</code> for a separate, clearly labeled rollup.</p>
<p>Save the <code>firstDate</code> and <code>lastDate</code> from the record’s <code>collectionPeriod</code> beside each result. The window rolls over 28 days and the latest data is about two days behind. A report appearing after a deploy can still contain mostly pre-deploy visits; for an initial trend, compare like-for-like weekly points, then wait for a window wholly after the change before judging the full effect. The <a href="https://developer.chrome.com/docs/crux/history-api">History API</a> groups overlapping windows at weekly intervals. Adjacent points therefore share three weeks of traffic. Use the documented <code>queryRecord</code> fields for the current comparison and confirm each field is returned in the historical periods your reporting needs.</p>
<h2>A reporting change can move the number immediately</h2>
<p>The <a href="https://developer.chrome.com/docs/crux/release-notes">October dataset release</a> gives a real comparison covering 18,070,207 origins, down 1.7% from the preceding release. The share with good scores on all three Core Web Vitals was 56.3% with the existing calculation and 60.3% with soft navigations included. That is a 4.0 percentage point gap within the same dataset, or a 7.1% relative increase as the release reports it. Reporting it as a site improvement would compare two measurement definitions as if they were two dates.</p>
<figure><svg xmlns="http://www.w3.org/2000/svg" id="crux-oct-coverage" viewbox="0 0 400 214" width="100%" role="img" aria-labelledby="crux-oct-title crux-oct-desc"><title id="crux-oct-title">October origins with good Core Web Vitals</title><desc id="crux-oct-desc">On a zero to one hundred percent scale, 56.3 percent pass all three with the existing series and 60.3 percent with soft navigations included. The difference is 4.0 percentage points.</desc>
<rect x="0" y="0" width="400" height="214" rx="12" fill="#edf2ed"></rect>
<text x="20" y="35" font-family="sans-serif" font-size="20" font-weight="700" fill="#20372d">Good on all three Vitals</text>
<text x="20" y="69" font-family="sans-serif" font-size="17" fill="#20372d">Existing series</text>
<rect x="20" y="79" width="360" height="18" rx="4" fill="#ced9d2"></rect>
<rect x="20" y="79" width="202.68" height="18" rx="4" fill="#426557"></rect>
<text x="232" y="94" font-family="sans-serif" font-size="17" fill="#20372d">56.3%</text>
<text x="20" y="131" font-family="sans-serif" font-size="17" fill="#20372d">Soft navigations included</text>
<rect x="20" y="141" width="360" height="18" rx="4" fill="#ced9d2"></rect>
<rect x="20" y="141" width="217.08" height="18" rx="4" fill="#426557"></rect>
<text x="248" y="156" font-family="sans-serif" font-size="17" fill="#20372d">60.3%</text>
<text x="20" y="192" font-family="sans-serif" font-size="16" fill="#20372d">0%</text>
<text x="380" y="192" text-anchor="end" font-family="sans-serif" font-size="16" fill="#20372d">100%</text>
</svg><figcaption>Same October origins, two measurement definitions. Data from the <a href="https://developer.chrome.com/docs/crux/release-notes">CrUX release notes</a>.</figcaption></figure>
<p>The individual good-score shares were 68.8% versus 73.6% for LCP, 81.6% versus 85.0% for CLS, and 85.2% versus 87.0% for INP, standard versus experimental. Keep those same-month differences separate from the release’s month-over-month changes in the standard series: LCP +0.7%, CLS −0.1%, INP −0.2% and all three +0.7%, as reported in the release notes.</p>
<p>The <a href="https://web.dev/blog/crux-experimental-soft-navs">October 13 launch analysis</a> found 66–75% of origins had fewer than 5% soft navigations. More than half of views were soft on 7.0% of Android origins and 10.6% of desktop origins; among the top 1,000 sites, those shares were 35% and 42%. Its Android cohort with 75–100% soft navigations moved from 26% good LCP under the standard calculation to 87% under the experimental one. That earlier cohort comparison shows why a route-heavy app deserves attention; it is separate from November’s whole-dataset comparison.</p>
<h2>Keep the series separate while the metrics are experimental</h2>
<p>For a site that loads a document on most clicks, keep the standard series as the default report. For an app with frequent in-app views, add the experimental series alongside it and use URL-level results to choose a route to profile. The experimental LCP, INP and CLS fields arrived on October 13; November 11 adds FCP and navigation types. The new paint timing helps investigate a route that leaves visitors waiting for its first content.</p>
<p>CrUX includes eligible Chrome users on Android and desktop. Chrome on iOS and Android WebView are excluded; Android Chrome Custom Tabs can contribute. URLs must be publicly discoverable and sufficiently popular under the <a href="https://developer.chrome.com/docs/crux/methodology">collection rules</a>. A missing route therefore needs a coverage check before it becomes a performance conclusion. The same platform limits matter for US, UK and European audiences: a phone report is not a complete picture of iPhone or embedded-app traffic.</p>
<p>The stable Core Web Vitals remain unchanged during the experiment. <a href="https://web.dev/blog/crux-experimental-soft-navs">Google targets May 2027 for graduation, subject to feedback</a>, with Search Console switching after graduation. A faster experimental FCP today does not change a Search Console Core Web Vitals result: FCP is not one of those three metrics.</p>
<h2>Test the route the report points to</h2>
<p>Use the same destination in two recordings: first open its URL directly, then reach it by tapping a link from another view in the app. Start with Chrome 151 or later and the <a href="https://developer.chrome.com/docs/web-platform/soft-navigations">DevTools Performance panel’s soft-navigation support</a>. Match the phone and network conditions to the audience you are investigating. Check that the transition produces a soft-navigation entry and that its destination URL is correct before interpreting its paint timing.</p>
<p>If the transition stalls, inspect the request, script and rendering work between the tap and the first content paint. If CrUX cannot cover the route, use real-user monitoring that supports soft navigations to find how often that stall affects visitors; confirm support with the provider before relying on its route metrics. Keep the direct landing and in-app arrival separate in that diagnosis. A blended percentile tells you where to look; the two recordings reveal which experience needs the fix.</p>
]]></content:encoded>
					
		
		
			</item>
		<item>
		<title>WordPress 7.2 Beta 5 Fixes Focus and Motion</title>
		<link>https://www.mobiletech.mobi/wordpress-ipsum-focus-reduced-motion/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Wed, 11 Nov 2026 11:13:08 +0000</pubDate>
				<category><![CDATA[Mobile Publishing]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=579</guid>

					<description><![CDATA[Test visible button focus and reduced-motion page transitions on a staging copy before adopting the new WordPress theme, Ipsum.]]></description>
										<content:encoded><![CDATA[<p>WordPress <a href="https://wordpress.org/news/2026/11/wordpress-7-2-beta-5/">7.2 Beta 5 arrived on November 10, 2026</a>, with two accessibility fixes for its new default theme, Ipsum. Page crossfades now follow the visitor’s reduced-motion preference, and several button focus rings move inward. If you are evaluating Ipsum, test these changes now on a staging copy. Keep your stable WordPress release on production. Beta 5 is for testing and development. If you use another theme, test its own focus and motion behavior.</p>
<table>
<caption>Ipsum Beta 5: what your test result means</caption>
<thead>
<tr>
<th scope="col">Observed result</th>
<th scope="col">Decision</th>
</tr>
</thead>
<tbody>
<tr>
<td>Default motion fades; reduced motion does not</td>
<td>Pass the theme’s motion check</td>
</tr>
<tr>
<td>Both motion settings still fade</td>
<td>Find the remaining transition opt-in</td>
</tr>
<tr>
<td>Neither motion setting fades</td>
<td>Establish a supported-browser control first</td>
</tr>
<tr>
<td>Each tabbed button has visible focus</td>
<td>Pass for the tested controls and palette</td>
</tr>
<tr>
<td>Inset ring disappears against its background</td>
<td>Fix the focused colors before adoption</td>
</tr>
</tbody>
</table>
<h2>Why the page fade needed a theme fix</h2>
<p>In Beta 4, <a href="https://github.com/WordPress/ipsum/issues/196">the theme enabled cross-document transitions even with reduced motion requested</a>. The browser’s ability to animate the navigation did not itself make Ipsum honor that preference. Beta 5 gates the opt-in with a media query:</p>
<pre><code>@media (
    prefers-reduced-motion:
    no-preference
) {
    @view-transition {
        navigation: auto;
    }
}</code></pre>
<p>This is the <a href="https://github.com/WordPress/ipsum/pull/198">condition introduced by the motion fix</a>. The crossfade remains available when the visitor has no reduced-motion preference. When reduced motion is requested, Ipsum no longer enables it through this rule. Independent transition rules in a child theme or plugin still need their own check.</p>
<p>For an eligible <a href="https://developer.mozilla.org/en-US/docs/Web/CSS/Reference/At-rules/@view-transition">cross-document transition</a>, both documents must opt in and share the same origin, with no cross-origin redirect in between. Click a link within the page for your test. Typing a destination into the address bar or following an external booking link does not exercise the same route.</p>
<h3>Run the same navigation with both preferences</h3>
<p>Start with stock Ipsum on Beta 5 in Chrome or Edge, the browsers used in the patch’s test instructions. Use your index page and a linked post. You only need Beta 4 if you are investigating the original regression.</p>
<ol>
<li>Leave the device’s motion preference at its default. Load the index, click the post link and check that a crossfade appears.</li>
<li>Enable reduced motion in the operating system. Reload the index, then click the same post link. Expect ordinary navigation without the crossfade; the post must still open.</li>
<li>Switch to one non-default Ipsum colorway and repeat both passes. Then repeat the route on a phone using its accessibility setting and your audience’s browser.</li>
</ol>
<p>Keep the URLs and browser fixed while changing the preference. A fade in the first pass supplies the positive control: without it, two unanimated navigations cannot establish whether the preference gate worked. The table above gives the next action for each result. Judge the animation separately from how long the destination takes to load.</p>
<h2>What changed in button focus</h2>
<p>The <a href="https://github.com/WordPress/wordpress-develop/pull/15845">bundled Ipsum update</a> removes the Button block’s <code>4px</code> focus-offset overrides, including its Outline variation. They can inherit the shared <code>-4px</code> offset already used by the theme’s button element styles. Those shared styles specify a <code>1px</code> dotted outline using <code>currentColor</code>, the element’s text color. Moving that line inward changes which background sits behind it.</p>
<figure><svg xmlns="http://www.w3.org/2000/svg" viewbox="0 0 400 264" width="100%" role="img" id="ipsum-focus-offset" aria-labelledby="ipsum-offset-title ipsum-offset-desc"><title id="ipsum-offset-title">Check the inset ring and the unchanged Search control</title><desc id="ipsum-offset-desc">Offset schematic, not a screenshot or scale drawing. A positive four-pixel offset puts the ring outside the button; negative four pixels puts it inside. Check the background under that ring. Search already used negative four pixels.</desc>
<rect x="0" y="0" width="400" height="264" rx="14" fill="#eeeae0"></rect>
<text x="20" y="34" font-family="sans-serif" font-size="21" font-weight="700" fill="#212936">Check the ring’s background</text>
<text x="22" y="70" font-family="sans-serif" font-size="17" fill="#212936">Before: +4 px</text>
<rect x="24" y="90" width="150" height="60" rx="9" fill="#28394d"></rect>
<rect x="19" y="85" width="160" height="70" rx="12" fill="none" stroke="#fff8df" stroke-width="2" stroke-dasharray="4 3"></rect>
<text x="200" y="70" font-family="sans-serif" font-size="17" fill="#212936">After: −4 px</text>
<rect x="202" y="90" width="150" height="60" rx="9" fill="#28394d"></rect>
<rect x="207" y="95" width="140" height="50" rx="6" fill="none" stroke="#fff8df" stroke-width="2" stroke-dasharray="4 3"></rect>
<text x="23" y="188" font-family="sans-serif" font-size="17" fill="#212936">Ring on page color</text>
<text x="200" y="188" font-family="sans-serif" font-size="17" fill="#212936">Ring on button color</text>
<text x="23" y="232" font-family="sans-serif" font-size="17" fill="#212936">Control: Search stays at −4 px</text>
</svg><figcaption>Check the color behind the ring, not just its offset. Colors and distances are schematic. See the <a href="https://github.com/WordPress/ipsum/pull/187">Ipsum focus patch</a>.</figcaption></figure>
<p>A <a href="https://github.com/WordPress/ipsum/pull/187">November 2 contributor test</a> shows the mechanism: on <code>7.2-beta3-src</code>, an intermediate patch changed Fill, Outline and Post Comment from <code>4px</code> to <code>-4px</code>; Search stayed at <code>-4px</code>. Chromium and Firefox at 1200 × 800 showed the near-white ring more clearly against the dark focused button than against the light page. The contributor estimated roughly 6.4:1 contrast for that test.</p>
<p>The final patch also makes Outline’s focused colors match its hover colors. That earlier contrast estimate therefore cannot certify every Beta 5 button or palette. Use it as a worked example of why placement matters: for a pale ring on a pale page, bringing the ring onto a dark button can solve the visibility problem. Your decision still rests on the colors of the control you ship.</p>
<h3>Test actual controls, including Outline</h3>
<p>On a test post, add Fill and Outline Button blocks, a Search block and an open comment form. Tab through all four controls. Check the ring as focus arrives and while it stays there; <a href="https://www.w3.org/WAI/WCAG22/Understanding/focus-visible.html">visible keyboard focus must remain visible</a>. Use a real keyboard action rather than a mouse hover or tap.</p>
<p>Repeat with your site’s palette, including a button in a narrow card or callout, then check high-contrast mode. On a phone or tablet, use a connected keyboard for the keyboard-focus check. For each control, record its computed outline color and offset alongside what you can actually see. A correct negative offset cannot rescue a ring that blends into its focused background.</p>
<h2>If a check fails, isolate the override</h2>
<p>Compare the failing route with an uncustomized Ipsum copy on the same build. If only your customized site fails, inspect child-theme CSS, saved styles and plugin styling before changing the whole palette. For a persistent fade, look for another <code>@view-transition</code> rule; for a missing ring, inspect the winning focus declarations. Keep the PHP version fixed while comparing the theme.</p>
<p>Test the public-facing page separately from the signed-in editor. Start with the index and single-post templates, then cover page, archive and 404 templates where their backgrounds or controls differ. The <a href="https://make.wordpress.org/test/2026/10/20/help-test-wordpress-7-2/">Ipsum testing plan</a> also calls for style variations, patterns and desktop, tablet and mobile layouts.</p>
<p>For a reproducible stock-theme failure, include WordPress and theme versions, plugins, browser, OS and motion setting. Supply both navigation URLs, or the block style and palette, plus steps and a capture. The <a href="https://www.mobiletech.mobi/intera11y-interactive-web-accessibility-checks/">interaction-testing guide</a> explains why the sequence that produces the state matters. Use the <a href="https://wordpress.org/support/forum/alphabeta/">Alpha/Beta/RC forum</a> for help. The <a href="https://make.wordpress.org/test/2026/10/20/help-test-wordpress-7-2/">official reporting guide</a> directs reproducible core bugs to Trac and editor bugs to Gutenberg’s issue tracker. To join organized testing, use the <a href="https://make.wordpress.org/test/">test team</a> and its <a href="https://make.wordpress.org/chat/">#core-test Slack channel</a>.</p>
<h2>Choose a test build for Ipsum</h2>
<p>For a stock check, open the <a href="https://wordpress.org/news/2026/11/wordpress-7-2-beta-5/">release page’s Playground</a>. For your site, use staging. Download the <a href="https://wordpress.org/wordpress-7.2-beta5.zip">Beta 5 ZIP</a>, or install and activate WordPress Beta Tester with <strong>Bleeding edge</strong> and <strong>Beta/RC Only</strong>. To pin an existing disposable installation, run <code>wp core update --version=7.2-beta5</code>.</p>
<p>Other Beta 5 work covers application-password creation emails, admin-toolbar keyboard navigation, Visual History accessibility, Notes reaction permissions, MP3 uploads and Git-worktree automatic updates. Ipsum also gains block-style, pattern and stylesheet refinements; the <a href="https://github.com/WordPress/wordpress-develop/pull/15845">theme sync</a> lists those changes. The <a href="https://wordpress.org/news/2026/11/wordpress-7-2-beta-5/">release notes</a> link core tickets and Gutenberg changes since November 3, and report known core PHP 8.6 issues resolved ahead of its scheduled November 19 release. Verify plugin and theme PHP compatibility separately.</p>
<p>The <a href="https://make.wordpress.org/core/2026/10/06/wordpress-7-2-release-party-schedule/">planned milestones</a> are RC1 on November 17, RC2 on November 24, RC3 on December 1, a dry run on December 7 and final release on December 8, 2026. Save your failing and passing routes now. Rerun them on the release candidate and final build with your content and plugins; adopt Ipsum when those checks pass alongside your normal upgrade checks. If one fails, keep your current theme while you resolve it.</p>
]]></content:encoded>
					
		
		
			</item>
		<item>
		<title>Google Adds Things to Do to EEA Search Units</title>
		<link>https://www.mobiletech.mobi/google-eea-things-to-do-search/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Tue, 10 Nov 2026 11:09:22 +0000</pubDate>
				<category><![CDATA[Mobile SEO]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=576</guid>

					<description><![CDATA[See how Google's EEA activity results work, when tour operators need a feed, and how to test the mobile path from search to booking.]]></description>
										<content:encoded><![CDATA[<p>On November 9, 2026, Google <a href="https://developers.google.com/search/updates">confirmed support for “things to do” queries</a> in two search features for the European Economic Area: the aggregator unit and the supplier unit. For activities, tours and events, a marketplace can apply to submit listings to the aggregator unit. A museum or tour operator can qualify for the supplier unit through its own website. If you run the activity yourself, check the page people land on before you budget for a feed.</p>
<p>The location that matters is the searcher&#8217;s. The <a href="https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:European_Economic_Area_(EEA)">EEA</a> covers 30 countries: the EU member states plus Iceland, Liechtenstein and Norway. The UK and US are outside these units&#8217; documented coverage. A business based there can still serve EEA users, so checking from head office may miss the layout your customers see.</p>
<table>
<caption>Which route fits your site?</caption>
<thead>
<tr>
<th scope="col">Site</th>
<th scope="col">First action</th>
</tr>
</thead>
<tbody>
<tr>
<td>Museum’s own tickets</td>
<td>Supplier: check your ticket page</td>
</tr>
<tr>
<td>Tour marketplace</td>
<td>Aggregator: check feed eligibility</td>
</tr>
<tr>
<td>City guide</td>
<td>Web search: keep activity links</td>
</tr>
<tr>
<td>US-only operator</td>
<td>Outside EEA: keep current setup</td>
</tr>
</tbody>
</table>
<h2>What a customer sees in the two units</h2>
<p>The <a href="https://developers.google.com/static/search/docs/images/aggregator-unit.png">official aggregator example</a> uses Paris hotels. On the phone mock-up, the top-ranked provider, Example.com, opens by default. Its cards show hotels with prices and ratings, while the other provider choices are collapsed below. Pick an alternative, if there is one, and its inventory opens in the same unit. A click goes to the page supplied by that aggregator. For the hotel, that click is a referral opportunity through the marketplace. The hotel’s own landing page is a separate result.</p>
<p>The <a href="https://developers.google.com/static/search/docs/images/supplier-unit.png">supplier illustration</a> shows a separate set of cards headed “Hotel websites.” That is the part for direct providers. Think of a walking tour sold by both a marketplace and the guide who runs it. The two units offer different routes: through an intermediary or through a direct provider. These hotel illustrations show the layout rather than a live activity search.</p>
<figure><svg xmlns="http://www.w3.org/2000/svg" id="eea-activity-units" viewbox="0 0 400 344" width="100%" role="img" aria-labelledby="eea-unit-title eea-unit-desc"><title id="eea-unit-title">How the two EEA search units depend on each other</title><desc id="eea-unit-desc">When an aggregator unit appears, one provider is expanded and a direct supplier unit can appear alongside it. If supplier cards are present, inspect the direct landing page; if absent, check eligibility. When the aggregator unit is absent, no supplier unit appears.</desc>
<rect x="0" y="0" width="400" height="344" rx="16" fill="#edf1e9"></rect>
<text x="22" y="36" font-family="sans-serif" font-size="21" font-weight="700" fill="#203b37">Is the aggregator unit present?</text>
<rect x="20" y="56" width="360" height="154" rx="10" fill="#ffffff" stroke="#577870"></rect>
<text x="37" y="86" font-family="sans-serif" font-size="20" font-weight="700" fill="#203b37">Yes: check both routes</text>
<text x="37" y="114" font-family="sans-serif" font-size="17" fill="#203b37">1 aggregator unit; 1 provider opens</text>
<text x="37" y="140" font-family="sans-serif" font-size="17" fill="#203b37">Other providers may be selectable</text>
<text x="37" y="166" font-family="sans-serif" font-size="17" fill="#203b37">Supplier card? Test its destination</text>
<text x="37" y="190" font-family="sans-serif" font-size="17" fill="#203b37">No supplier card? Check eligibility</text>
<rect x="20" y="224" width="360" height="96" rx="10" fill="#ffffff" stroke="#577870"></rect>
<text x="37" y="255" font-family="sans-serif" font-size="20" font-weight="700" fill="#203b37">No: rule out this supplier unit</text>
<text x="37" y="284" font-family="sans-serif" font-size="17" fill="#203b37">0 supplier units by design</text>
<text x="37" y="306" font-family="sans-serif" font-size="16" fill="#203b37">Inspect other results separately</text>
</svg><figcaption>Read the aggregator result before diagnosing a missing supplier result. Schematic, not a screenshot. Sources: <a href="https://developers.google.com/search/docs/appearance/aggregator-unit">aggregator behavior</a> and <a href="https://developers.google.com/search/docs/appearance/supplier-unit">supplier eligibility</a>.</figcaption></figure>
<h2>If you run the activity, start with your own page</h2>
<p>A <a href="https://developers.google.com/search/docs/appearance/supplier-unit">direct provider serving EEA users</a> needs no separate application or additional data beyond what Google can crawl to be eligible for the supplier unit. Optional feeds can enrich results where supported. For a museum or tour business, the public activity page is the participation route. Optional enrichment can wait until a supported integration solves a specific need.</p>
<p>Start with an activity people can book. A guided museum visit needs a page that identifies the visit, explains where it starts and what the ticket includes, and lets customers check dates. Keep those details public even if availability comes from a booking engine. A Book button that opens a generic calendar and forgets the chosen visit wastes that referral.</p>
<p>In Search Console&#8217;s <a href="https://support.google.com/webmasters/answer/9012289">URL Inspection</a>, check Page indexing for the Google-selected canonical. Fix it if Google has selected an unintended page. Then run Test live URL and open View tested page to inspect the rendered offer. Canonical selection belongs to the indexed report; the live test checks the current page and cannot predict which canonical Google will select.</p>
<h3>A €25 museum visit: which problem should you fix?</h3>
<p>Take a hypothetical Paris museum selling its own guided visit for €25. Its public visit page is indexed under the intended canonical. A mobile search sample taken in France shows an aggregator unit but no card for the museum. The museum meets the business-role and audience requirements, but that does not guarantee a card. Buying an aggregator-feed integration is the wrong first response for this direct supplier.</p>
<p>Now follow the museum&#8217;s booking button on a phone. Suppose the visit page shows €25, but the booking engine opens a €40 private-tour product. Fix that product mapping first: the broken journey is observable and affects visitors however they arrive. If the correct €25 visit opens with the chosen language and date intact, keep monitoring search appearances rather than rebuilding a working booking path.</p>
<h2>If you run a marketplace, ask about approval before building</h2>
<p>Marketplaces, online travel agencies, comparison services, metasearch engines and directories use the aggregator route. Begin with the <a href="https://support.google.com/websearch/contact/search_aggregator_features">Google Search Aggregator Features interest form</a> and confirm eligibility before assigning a feed build. Participation requires VSS approval, relevant inventory, the required data and compliance with Search content policies. The <a href="https://developers.google.com/actions-center/verticals/local_vss/integration-steps/overview">Local integration</a> also expects an aggregator with direct contractual merchant relationships; a few affiliate links do not establish that relationship.</p>
<p><a href="https://developers.google.com/search/docs/appearance/aggregator-unit">Local businesses, things to do and lodging use direct data feeds</a>; flights and long-distance transport use real-time APIs. An activity marketplace should follow the POI route. Existing Things to Do booking inventory is a separate integration: the <a href="https://developers.google.com/actions-center">Actions Center catalog</a> lists bookings and Local &amp; Things to Do in the EEA separately.</p>
<h3>Match the place, then check the destination</h3>
<p>The <a href="https://developers.google.com/actions-center/verticals/local_vss/reference/poi-feed-doc">POI specification</a> requires <code>poi_id</code>, <code>name</code>, <code>location</code>, <code>category</code> and matching category-specific data. Its restaurant example uses <code>test_restaurant_1</code> with <code>category: LOCAL</code> and <code>local_data</code>. The top-level <code>url</code> matches the entity; destinations live under <code>local_data.brand_landing_pages</code>, with required <code>brand_id</code> and <code>default_url</code>. Optional <code>localized_landing_pages</code> pair URLs with language tags. Check those fields if the wrong language opens.</p>
<p>Validate before upload. Send one JSON file containing all entities as a daily full refresh over SFTP; gzip is recommended. Keep each entity&#8217;s translations together, with <code>default_locale</code> present in <code>localizations</code>. Do not split the archive into country files. In the Partner Portal, use Configuration &gt; Feeds for SFTP details and Ingestion &gt; History for processing status. A successful upload still needs ingestion checks.</p>
<p>Use the <a href="https://developers.google.com/actions-center/verticals/local_vss/integration-steps/overview">sandbox and production review stages</a> to verify matching and destinations before launch. Start with a representative sample, including a multilingual entity, then build the complete refresh. A feed record represents a place, and its brand landing page determines the onward route; neither is proof of exposure for a particular search.</p>
<p>Once records match, supply details that help someone choose: a precise category, descriptions, hours, relevant amenities and original photos. Include verified ratings and their counts where available. Use factual, descriptive titles in title case, avoiding all caps, excessive punctuation, emoji and promotions such as “BEST DEALS.” Keep prices and availability aligned with the landing page and remove expired offers. These are <a href="https://developers.google.com/search/docs/appearance/aggregator-unit">aggregator quality practices</a>, not a promise of placement.</p>
<h2>How to tell a search change from a booking problem</h2>
<p>Pick a few non-branded queries for the activities that matter to you and keep the list stable. Have someone in the target EEA market capture the mobile result with the country, language, device, timestamp and who or what collected it. Keep each market in a separate view before combining totals. Record whether you saw an aggregator unit, which provider was open, and whether the supplier unit appeared. If you can identify only a generic carousel, record it that way. Save branded queries separately so a shift in activity discovery does not get hidden by people searching for your name.</p>
<p>In Search Console, look at mobile clicks and impressions for the same country and landing pages. As a starting point, compare consecutive 28-day windows with the same query set, then check seasonal demand and your own site changes before attributing a movement to these units. Keep rank in the report, but <a href="https://support.google.com/webmasters/answer/7042828">average position</a> doesn&#8217;t tell you how far a customer had to scroll. Match that evidence with mobile bookings. If clicks hold up but bookings fall, replay the ticket journey before blaming the layout. If the booking happens on a separate domain or after a consent step, <a href="https://www.mobiletech.mobi/mobile-tracker-audit-consent-journeys/">test tracking across the whole journey</a> before reading the drop as lost sales.</p>
<h2>What November 9 changes, and when to wait</h2>
<p>November 9, 2026 dates the explicit confirmation of activity support in these two units. The broader EEA search redesign began earlier: the <a href="https://developers.google.com/search/blog/2024/02/search-experiences-in-eea">February 15, 2024 announcement</a> tied new search experiences to preparations for the Digital Markets Act. That post&#8217;s interest-form list now includes events, tours and activities, but the list and supported query types were updated on April 10, 2025.</p>
<p>The <a href="https://developers.google.com/search/updates">2026 documentation sequence</a> is September 8 for regional search experiences covering hotels, flights, long-distance trains or buses and products; September 18 for local businesses; and November 9 for things to do. Those dates identify documentation changes, not the first impression of every result. There is no query-by-query appearance schedule or stated mobile browser/version requirement in these guides.</p>
<p>At the next check, repeat the same queries in the same market and save the result beside today’s sample. If the units appear, follow the actual links and compare the destination with the advertised offer. If they remain absent and your page works, keep monitoring; a site rebuild has no observed search problem to solve.</p>
]]></content:encoded>
					
		
		
			</item>
		<item>
		<title>Shopify Bulk Jobs Now Need Separate Permissions</title>
		<link>https://www.mobiletech.mobi/shopify-bulk-operations-access/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Mon, 09 Nov 2026 11:13:02 +0000</pubDate>
				<category><![CDATA[Mobile Publishing]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=567</guid>

					<description><![CDATA[Check Shopify's new bulk query and mutation permissions, the 31-day rule, and the IP deadline before your next catalog job.]]></description>
										<content:encoded><![CDATA[<p>A Shopify app can export a catalog on Sunday and get an access denied error when it tries to update prices on Monday. On November 6, 2026, Shopify made <a href="https://shopify.dev/changelog/posts/partner-apps-must-enable-bulk-operations-before-use">bulk reads and bulk writes separate permissions</a> for Partner-built apps. Each can lapse on its own. If you’re counting on a scheduled update to put the right price or stock status in front of shoppers, check the job that writes those values before you trust a green export.</p>
<p>Check two independent controls: the operation type in the Partner Dashboard, and the worker’s outbound network in the Dev Dashboard. Public apps face an additional IP allowlist deadline on November 12. A successful export proves neither that bulk writes are enabled nor that every worker can pass the network check.</p>
<table>
<caption>What to check before a Shopify bulk job</caption>
<thead>
<tr>
<th scope="col">Who runs the job</th>
<th scope="col">When to check</th>
<th scope="col">Action</th>
</tr>
</thead>
<tbody>
<tr>
<td>Public Partner app</td>
<td>Before next job and November 12 cutoff</td>
<td>Check operation type and enforce allowlist</td>
</tr>
<tr>
<td>Custom Partner app</td>
<td>Before next bulk job</td>
<td>Enable the operation type it uses</td>
</tr>
<tr>
<td>Merchant-created app</td>
<td>Before troubleshooting access denied</td>
<td>Check API permissions and request path</td>
</tr>
<tr>
<td>Merchant using an outside app</td>
<td>Before a scheduled catalog change</td>
<td>Ask for proof the test completed</td>
</tr>
</tbody>
</table>
<h2>Check the operation that will actually run</h2>
<p>In the Partner Dashboard, open <strong>Apps</strong> and select the app that sends the request. Confirm its identity against the scheduled integration, then open <strong>API access requests → Run bulk operations</strong>. The extra permission controls whether Shopify accepts a bulk operation; it does not grant access to the underlying store data.</p>
<ul>
<li><strong>Allow bulk queries</strong> enables <code>bulkOperationRunQuery</code>, used for reads such as catalog exports.</li>
<li><strong>Allow bulk mutations</strong> enables <code>bulkOperationRunMutation</code>, used for batch writes such as price changes.</li>
</ul>
<p>Enable only the types the app uses. If a type already shows as enabled, leave it enabled. <a href="https://shopify.dev/changelog/posts/partner-apps-must-enable-bulk-operations-before-use">Access applies to every installed store, across all GraphQL Admin API versions</a>. No app configuration change, release or merchant reinstall is required. Changing API versions will not bypass this permission.</p>
<h3>Count starts separately for reads and writes</h3>
<p>Existing bulk users had access enabled automatically, but an unused type can lapse. The <a href="https://shopify.dev/docs/apps/build/apis/graphql-admin/bulk-operations/queries#enabling-access-to-bulk-operations">31-day window counts whether the app started that type</a>, rather than whether a job completed.</p>
<p>Take a hypothetical seasonal promotion: the app starts an export on November 8, 2026, last started a bulk price update on October 1, and plans its next update for November 9. The write side has been idle for 39 days. That recent export does not preserve mutation access. Check <strong>Allow bulk mutations</strong> before the update and re-enable it if disabled.</p>
<p>Partner-created custom apps need the switches too. Apps merchants create in Shopify admin or the Dev Dashboard are exempt at this time. An older <a href="https://shopify.dev/changelog/blog/app-security-updates-ip-allowlist-is-now-required">November 5 security announcement</a> describes future scopes, app configuration changes and a 180-day lookback. For the permission available now, use the November 6 dashboard instructions and 31-day window.</p>
<h2>Run the documented product test</h2>
<p>Use a dev store to verify bulk-query admission and result handling. With <a href="https://shopify.dev/docs/api/usage/access-scopes">product-read access such as <code>read_products</code></a>, run this <a href="https://shopify.dev/docs/apps/build/apis/graphql-admin/bulk-operations/queries#step-1-submit-a-query">official product query</a> through its backend Admin API client:</p>
<pre><code>mutation {
  bulkOperationRunQuery(
    query: """
    {
      products {
        edges {
          node { id title }
        }
      }
    }
    """
  ) {
    bulkOperation { id status }
    userErrors { field message }
  }
}</code></pre>
<p>The documented response returns operation ID <code>gid://shopify/BulkOperation/720918</code>, status <code>CREATED</code> and an empty <code>userErrors</code> list. Your ID will differ. Save it: the request has been admitted, while Shopify fetches the products asynchronously. It will later expose a download URL.</p>
<ol>
<li>Prefer the <code>bulk_operations/finish</code> webhook to repeated polling. A finish notification can mean success, failure or cancellation, so inspect its status.</li>
<li>After success, query the <code>node</code> field with the webhook’s <code>admin_graphql_api_id</code> to retrieve the result <code>url</code>. Download the JSONL and check product IDs and titles against the dev store.</li>
<li>If a webhook does not arrive, recover by polling. On API versions <code>2026-01</code> and later, use <code>bulkOperation(id:)</code>; older versions use <code>currentBulkOperation</code>. Check status and error details before downloading.</li>
</ol>
<p>For writes, use a disposable dev-store product and change a field the app actually updates through its normal worker. Inspect the final result and fetch the product again to verify the value. A query test only exercises query access; it cannot approve the mutation path.</p>
<h3>Start diagnosis at the first failed step</h3>
<ul>
<li><strong>HTTP 403:</strong> inspect the actual outbound address and allowlist first, including requests whose source Shopify cannot confirm.</li>
<li><strong>Bulk access denied:</strong> verify the app identity and matching operation switch. Record the operation name, time and redacted response.</li>
<li><strong>Data permission error:</strong> test an ordinary query for the same fields. Bulk queries require the same data permissions as ordinary queries; check scopes and the store installation.</li>
<li><strong>Operation admitted, then failed:</strong> inspect final status, errors and available output before replaying. For a write, reconcile what changed before retrying the batch.</li>
</ul>
<p>Use the GraphQL Admin API backend for these tests. <a href="https://shopify.dev/docs/apps/build/apis/graphql-admin/bulk-operations/queries#admin-ui-extensions">Admin UI extension Direct API access does not support bulk operations</a>; connect the extension to its backend. The Storefront API cannot run them either.</p>
<h2>Test the outbound network the worker uses</h2>
<p>Public apps must turn on the IP allowlist by <a href="https://shopify.dev/changelog/blog/app-security-updates-ip-allowlist-is-now-required">November 12, 2026, at 11:59 p.m. Eastern</a> to keep bulk queries and mutations running. That is November 13 at 04:59 UTC, 4:59 a.m. in the UK and 5:59 a.m. in Central Europe. Custom apps are not required to have an allowlist, though Shopify recommends one.</p>
<p>The next public-app deadline is November 20, 2026, at 11:59 p.m. Eastern: missing the allowlist means delisting, blocked new installs and a flag in merchant settings. Existing installations retain ordinary API access until April 1, 2027, with an admin warning from January 15, 2027. Bulk jobs are already subject to the earlier cutoff. Expiring offline tokens have a separate January 1, 2027, listing deadline.</p>
<figure>
<svg xmlns="http://www.w3.org/2000/svg" id="shopify-bulk-dates" viewbox="0 0 420 280" width="100%" role="img" aria-labelledby="shopify-bulk-dates-title shopify-bulk-dates-desc">
<title id="shopify-bulk-dates-title">Three Shopify access checks in November 2026</title>
<desc id="shopify-bulk-dates-desc">November 6: check operation switches for Partner apps. November 12 at 11:59 p.m. Eastern: public apps must enforce IP allowlists for bulk jobs. November 20 at 11:59 p.m. Eastern: public app listing and new installs require the allowlist.</desc>
<rect x="0" y="0" width="420" height="280" fill="#f0f7f4"></rect>
<line x1="24" y1="33" x2="24" y2="221" stroke="#214e41" stroke-width="3"></line>
<circle cx="24" cy="33" r="6" fill="#214e41"></circle><circle cx="24" cy="127" r="6" fill="#214e41"></circle><circle cx="24" cy="221" r="6" fill="#214e41"></circle>
<text x="44" y="38" fill="#142f27" font-size="20" font-weight="bold">Nov 6 · Partner apps</text>
<text x="44" y="65" fill="#142f27" font-size="18">Check query and mutation switches</text>
<text x="44" y="132" fill="#142f27" font-size="20" font-weight="bold">Nov 12 · 11:59 p.m. ET</text>
<text x="44" y="159" fill="#142f27" font-size="18">Public apps: allowlist for bulk jobs</text>
<text x="44" y="226" fill="#142f27" font-size="20" font-weight="bold">Nov 20 · 11:59 p.m. ET</text>
<text x="44" y="253" fill="#142f27" font-size="18">Public apps: listing / install cutoff</text>
</svg>
<figcaption>November 2026 action timeline, unscaled. The public-app bulk cutoff precedes the listing cutoff. Sources: <a href="https://shopify.dev/changelog/posts/partner-apps-must-enable-bulk-operations-before-use">bulk access release</a> and <a href="https://shopify.dev/changelog/blog/app-security-updates-ip-allowlist-is-now-required">security deadlines</a>.</figcaption>
</figure>
<p>Open the app in <strong>Dev Dashboard → Settings → IP allowlist → Manage</strong>. Prefer public outbound IP addresses when they are stable. For a worker behind NAT, use the gateway’s public addresses and include IPv6 if it reaches Shopify that way. Under <strong>Allow requests from</strong>, select <strong>IP addresses</strong> or <strong>Networks (ASN)</strong>.</p>
<ul>
<li>IP mode accepts up to 100 addresses or CIDR ranges. Use IPv4 <code>/16</code> or narrower and IPv6 <code>/32</code> or narrower. A range must start at its network address; for example, <code>203.0.113.0/24</code>, not <code>203.0.113.7/24</code>.</li>
<li>For dynamic egress, use up to 10 public hosting-provider ASNs and narrow them with up to 10 countries. A hosting network is shared with other customers, so static addresses are the safer default where available.</li>
</ul>
<p>ASN and country matching uses MaxMind GeoLite’s classification of the request’s source address. A missing ASN, or a missing country when you apply a country filter, causes rejection. Check the worker addresses in Dev Dashboard API logs and your provider’s network details before selecting those rules. <a href="https://shopify.dev/docs/apps/build/dev-dashboard/ip-allowlist">Allowlist rules and compatibility</a> apply to both GraphQL and REST Admin API requests.</p>
<p>Click <strong>Run test</strong> under <strong>Test against recent traffic</strong> to check the past 30 days, correct blocked legitimate addresses, then turn on <strong>Enforcement</strong> and <strong>Save</strong>. Exercise occasional workers too: historical traffic can miss a seasonal process, and recent requests can take a few minutes to appear. Confirm a normal backend token request succeeds from the allowed worker and fails with HTTP <code>403</code> from outside the list.</p>
<p>Use backend requests for that network test. Dev preview requests on dev stores, embedded-page Direct API requests and POS UI extension Direct API requests are exempt from rejection. Conversely, Shopify Flow’s <strong>Send HTTP request</strong> action and GraphQL Admin API calls from Oxygen-hosted storefronts are incompatible with enforced allowlists because Shopify cannot confirm their source IP. Identify those paths before enabling enforcement and resolve them with the integration owner.</p>
<h2>Make the next dormant job predictable</h2>
<p>Add the app ID, operation type, last accepted start, owner, worker egress and latest result evidence to the integration’s runbook. Put the dashboard preflight before an infrequent job’s next scheduled run. Prefer checking real job activity over starting synthetic bulk jobs merely to keep access alive: those starts change the inactivity window and can hide the condition you need to detect.</p>
<h2>What merchants can verify before a price change</h2>
<p>Open <a href="https://changelog.shopify.com/posts/review-app-access-and-security-information-in-shopify-admin">Settings → Apps</a> in Shopify admin to see recent activity and whether the app uses expiring tokens and enforced IP restrictions. Send the developer the exact job name and ask them to run a small batch from the worker that normally handles it. For an export, ask when the feed last refreshed. For a write, ask which product was tested and verify its new value on the store.</p>
<p>If that proof is missing on launch day, hold the affected batch. Check a representative product page on a phone and compare its displayed price or availability with the intended value in Shopify admin. Keep unrelated jobs running. Release the batch when the owner can point to a completed test and the correct value where a shopper sees it.</p>
]]></content:encoded>
					
		
		
			</item>
		<item>
		<title>Clarity Agent Tests: What a Checkout Pass Shows</title>
		<link>https://www.mobiletech.mobi/clarity-agent-checkout-tests/</link>
		
		<dc:creator><![CDATA[mobiletech_admin]]></dc:creator>
		<pubDate>Sun, 08 Nov 2026 15:08:22 +0000</pubDate>
				<category><![CDATA[Mobile UX]]></category>
		<guid isPermaLink="false">https://www.mobiletech.mobi/?p=563</guid>

					<description><![CDATA[Clarity’s published ski-jacket example flags missing cart options. Learn to read its results, choose a repair, and verify checkout yourself.]]></description>
										<content:encoded><![CDATA[<p>Microsoft Clarity&#8217;s November 4, 2026 update gives store owners a useful new question to ask: can a shopping agent finish the job on this website? Its <a href="https://clarity.microsoft.com/blog/clarity-connects-agents-LLMs-people/">simulation assessments report completion, time, interaction count, and improvement opportunities</a>, initially for commerce and travel. The interesting part is what those results can expose: a cart that holds the right product but gives the agent too little information to verify it.</p>
<p><strong>My recommendation: try an assessment now if you already use Clarity, receive agent traffic, and have access to simulations.</strong> Start with a purchase path you care about. Use the result to choose a repair, then verify that repair against the cart and order data. A team without access should keep its existing checkout tests running rather than install a shopping assistant just to chase this feature.</p>
<h2>Start with the assessment, not another storefront widget</h2>
<p>Agent Activity observes visitors; simulations attempt tasks. That distinction matters. The dashboard became <a href="https://clarity.microsoft.com/blog/agentic-analytics-what-agents-do-on-websites/">generally available on October 29 for projects receiving agent traffic</a>. The November update explains what the assessments return. Watching an agent reach checkout tells you where to investigate; a controlled task gives you a result to compare after a change.</p>
<table>
<caption>Where to begin with Clarity</caption>
<thead>
<tr>
<th scope="col">Your situation</th>
<th scope="col">Best starting point</th>
<th scope="col">Next action</th>
</tr>
</thead>
<tbody>
<tr>
<th scope="row">Agent visits, simulations available</th>
<td>One important purchase task</td>
<td>Assess and inspect the failure</td>
</tr>
<tr>
<th scope="row">Agent visits, no simulation access</th>
<td>Agent Activity</td>
<td>Investigate a recorded failed journey</td>
</tr>
<tr>
<th scope="row">No detected agent visits</th>
<td>Existing checkout regression tests</td>
<td>Check coverage before buying more tools</td>
</tr>
<tr>
<th scope="row">Need product discovery</th>
<td>Separate Brand Agents decision</td>
<td>Review catalog and data permissions</td>
</tr>
</tbody>
</table>
<p>Brand Agents is the customer-facing assistant and catalog-discovery service, <a href="https://learn.microsoft.com/en-us/clarity/brand-agents/brand-agents-overview">currently a beta for Shopify and WooCommerce websites, not native apps</a>. Eligible merchants can choose discovery across Copilot, Bing, and MSN without adding the on-site assistant; existing installations need no extra catalog upload. If checkout diagnosis is your goal, begin with analytics. Adding a conversational widget changes the experience you are trying to assess.</p>
<h2>The published example: a blue ski jacket that becomes hard to verify</h2>
<p>The <a href="https://clarity.microsoft.com/blog/wp-content/uploads/2026/11/Agentic-Analytics-dash.png">official dashboard screenshot</a> shows the last completed simulation for <code>www.alpineskihouse.com</code>, dated July 6, 2026. The visible task involves selecting a medium blue ski jacket and adding it to the cart. The exact starting prompt is not displayed. This is a published product example, not an independently measured store benchmark.</p>
<p>The assessment shows <strong>45.1% average task completion, 3.2 minutes average time to complete, and 21.0 average interactions</strong>. Its improvement opportunity is specific: after the jacket was added, the cart omitted the selected size and color. It recommends making those options visible so the agent can verify the item before continuing to checkout.</p>
<figure><svg xmlns="http://www.w3.org/2000/svg" id="clarity-jacket-assessment" viewbox="0 0 420 420" width="100%" role="img" aria-labelledby="clarity-jacket-title clarity-jacket-desc"><title id="clarity-jacket-title">Clarity's published ski-jacket assessment</title><desc id="clarity-jacket-desc">The example shows 45.1 percent average task completion, 3.2 minutes average time to complete, and 21 average interactions. The cart did not display selected size and color. The recommended action is to show both before checkout. This is a schematic of published results, not a screenshot.</desc><rect x="0" y="0" width="420" height="420" rx="12" fill="#f4f1e8"></rect><text x="24" y="42" font-family="sans-serif" font-size="22" font-weight="700" fill="#142c35">Ski-jacket assessment</text><text x="24" y="91" font-family="sans-serif" font-size="28" font-weight="700" fill="#146a68">45.1%</text><text x="155" y="91" font-family="sans-serif" font-size="18" fill="#142c35">Average completion</text><text x="24" y="139" font-family="sans-serif" font-size="28" font-weight="700" fill="#146a68">3.2 min</text><text x="155" y="139" font-family="sans-serif" font-size="18" fill="#142c35">Average time</text><text x="24" y="187" font-family="sans-serif" font-size="28" font-weight="700" fill="#146a68">21.0</text><text x="155" y="187" font-family="sans-serif" font-size="18" fill="#142c35">Average interactions</text><line x1="24" x2="396" y1="216" y2="216" stroke="#142c35" stroke-width="2"></line><text x="24" y="253" font-family="sans-serif" font-size="20" font-weight="700" fill="#142c35">Missing in cart:</text><text x="24" y="287" font-family="sans-serif" font-size="20" fill="#142c35">Selected size and color</text><text x="24" y="341" font-family="sans-serif" font-size="20" font-weight="700" fill="#142c35">Next: show both options</text><text x="24" y="377" font-family="sans-serif" font-size="18" fill="#142c35">Then repeat the same task</text></svg><figcaption>Schematic of the published assessment; repeating the task is our recommended follow-up. Source: <a href="https://clarity.microsoft.com/blog/wp-content/uploads/2026/11/Agentic-Analytics-dash.png">Clarity's dashboard example</a>.</figcaption></figure>
<p>The useful finding is the missing confirmation at the cart. A product selector can work perfectly and still leave the next step ambiguous. For the owner, the sensible repair is to show the selected variant beside the item, then check that the text agrees with the cart&#8217;s actual variant identifier. That second check matters: displaying “medium, blue” would be worse than useless if the cart stored a different SKU.</p>
<p>I would make that repair before optimizing the number of clicks. Then I would repeat the same selection task and inspect whether the agent can verify the variant and proceed. Those are recommended next steps, not a published outcome: the example gives no post-repair score.</p>
<h2>What those three numbers let you conclude</h2>
<p>A 45.1% completion rate is a reason to investigate, even if the average duration looks tolerable. It is not a diagnosis by itself. The screenshot does not reveal the number of attempts, the full task set, or how the averages were calculated. You cannot turn it into a count of failed orders or blame every incomplete task on the missing cart labels.</p>
<p>Read completion first, the observed obstacle second, and time and interactions after that. In this case, the recommendation gives the owner a small, testable change. The other metrics become useful when that same task is repeated under comparable conditions. A faster run that selects the wrong size is a regression.</p>
<p>For your own assessment, save the task, expected result, environment, and reported values together. Keep product availability, delivery destination, and approval boundaries steady when comparing runs. If any of those changes, explain it before crediting the storefront repair. An agent&#8217;s route can vary; one shorter attempt is a weak basis for declaring the problem solved.</p>
<h2>A checkout event still needs an order behind it</h2>
<p>The case also explains why Clarity&#8217;s traffic statistics need careful reading. Its <a href="https://clarity.microsoft.com/blog/agentic-analytics-what-agents-do-on-websites/">study of more than 5.2 million agent-involving sessions across 230,511 projects</a>, from August 13 to September 9, recorded an 88% increase in session volume over three weeks. Checkout appeared in 25.21% of agent sessions versus 14.65% of human sessions. The published checkout count, 306,621, implies a denominator of about 1.22 million at that percentage, rather than the headline cohort.</p>
<p>Use those figures as a reason to examine agent journeys, not as a conversion forecast. Events overlap across different sites, and an agent-classified session can include a human handoff. For a purchase task, verify the requested variant, quantity, delivery choice, total, order identifier, and payment state in your own systems. For a task ending before payment approval, judge the prepared basket and handoff instead.</p>
<p>Keep transaction checks in a controlled test environment with fulfillment disconnected. On WooPayments, <a href="https://woocommerce.com/document/woopayments/testing-and-troubleshooting/testing/">enable test mode under Payments &gt; Settings and save</a>; after a test purchase, compare <strong>WooCommerce &gt; Orders</strong> with <strong>Payments &gt; Transactions</strong>. Test mode creates real records labeled as tests. Resolve any pending payment or existing order before repeating a failed submission.</p>
<p>One route distinction matters for Shopify stores: <a href="https://help.shopify.com/en/manual/online-sales-channels/agentic-storefronts/microsoft">Copilot direct checkout can complete inside the chat</a>, without an online-store visit or client-side analytics pixels firing. It displays only to US customers. Check the channel order for that route; use the website assessment for purchases that actually reach your storefront, including the website journeys you serve to UK and European customers.</p>
<h2>Check access and coverage before relying on a pass</h2>
<p>Start in your Clarity project with Agent Activity. If simulation controls are absent, use the <a href="https://learn.microsoft.com/en-us/clarity/faq">top-right feedback control to contact support</a> about access. Confirm the available task types, simulator, run allowance, price, device coverage, and treatment of simulated traffic before scheduling routine checks. Clarity is advertised as free, but the announcement supplies no simulation-specific quota or charging schedule.</p>
<p>Supported Agentic Analytics operators are Google, OpenAI, and Anthropic. That coverage list is not a guarantee that the simulator reproduces every visiting agent. Nor does the announcement&#8217;s global availability language establish a country-by-country simulation entitlement. Use a pass as evidence about the task and environment assessed.</p>
<p>Keep the adjacent Clarity features separate in your purchasing decision. <a href="https://about.ads.microsoft.com/en/blog/post/november-2026/whats-new-in-microsoft-clarity-agent-insights-faster-answers-and-ai-discovery">Conversational Analytics</a> is the Ask Clarity experience: analysis linked to session evidence. It remains a closed retailer beta, with AI Visibility and advertising inputs planned. Those answers can help investigate a journey, but they do not supply missing transaction evidence.</p>
<p>If you also choose Brand Agents, review its <a href="https://learn.microsoft.com/en-us/clarity/brand-agents/brand-agents-overview">catalog and conversation-data terms</a>: accepting publishes a catalog copy to Microsoft Merchant Center, and customer conversations and interaction events are shared with Microsoft. For Clarity visitors in the EEA, UK, and Switzerland, <a href="https://learn.microsoft.com/en-us/clarity/faq">pass their cookie-consent choice through a supported integration</a>; missing consent can limit recordings and journey continuity. Check the consent path before interpreting a missing session.</p>
<p>Finally, verify the repaired cart on a real phone in Safari and Chrome. Check that the selected options remain readable with the keyboard open and after an error. Agent task duration cannot tell you whether a person can use that screen comfortably. Keep deterministic checkout tests for repeatable transaction checks, and use Clarity&#8217;s assessments to uncover paths and assumptions those tests missed.</p>
<h2>Turn the assessment into one repair ticket</h2>
<p>For the ski-jacket case, the ticket is straightforward: show selected size and color in the cart, confirm both match the stored variant, then repeat the selection-to-checkout task. Attach the original assessment, the changed cart state, and the new result. Accept the repair when the evidence shows the requested item survives the journey correctly; use time and interaction count to decide what to improve next.</p>
]]></content:encoded>
					
		
		
			</item>
	</channel>
</rss>