Universal Tag is about reducing redundancy.

The redundancy that we see typically comes in one of two forms:

1. Redundancy associated with using multiple solutions
2. Redundancy associated with data already available on the page

In this post, we’re going to discuss how Universal Tag can in fact reduce redundancy.

Redundancy Associated with Using Multiple Solutions

The trend that we’re clearly seeing today is that businesses are using more digital marketing tools than ever, which is resulting in more tags on web pages. For example, the home page of staples.com includes 7 different vendor tags. Many of the tags require their own set of variables to be declared, which results in redundant data. On this page alone, we see two different web analytics solutions: Omniture and Coremetrics.

Lets look at another example. This web site is using two different analytics solutions: HBX & SiteCatalyst. Both tags are collecting the same set of data, although the syntax may be different. If you were to look at the page weight associated with the HBX and SiteCatalyst tags on this page, you would end up with about 4 KB of added page weight per page. This is extra code on every single page.

For this same implementation, a Universal Tag page code will look something like this – which amounts to about 150 characters – a saving of about 4,000 bytes per page. To put this into perspective, for each 1 million page views per month, the bandwidth savings would be 3.7 GB/month, which is quite substantial.

<script src=”utag.loader.js”></script>
<script>
var utag_data={
page:”business”,
section:”business”,
segment:”business”
};
</script>

Compared to traditional analytics tagging, the Universal Tag offers several advantages:

• Simplified tagging
• Smaller page weight
• More flexible deployment

Redundancy Associated with Data Already Available on the Page

Often times, the data being passed to analytics solutions is already available on the page. Think about it. On a product page, you want to be able to capture the product name and attributes such as size, color, ratings, number of comments made, sku number, category, brand and other attributes. To demonstrate this, consider this example shown below of a product page.

The page itself includes just about all the information needed, including the product name, category, brand, rating, number of customer reviews, as well as the user status (“signed in” or “not signed in”). This is all information that’s useful for web analytics implementations. This information can be typically found in a number of ways, including:

• Parts of the page URL
• Query parameters within the page
• Meta tags on the page
• Microformats
• Other in-page elements such as h1 or other such tags
• Cookie values

The idea here is nothing new. See this great blog post by Dennis Mortensen from Yahoo! Web Analytics about the use of microformats for web analytics. The Tealium Universal tag fully supports the use of microformats, as well as other page elements mentioned above for data collection.

This approach creates a smaller footprint on the page, as well as lowers the deployment cycle. However, the approach also has some disadvantages. First, not everyone uses these elements consistently. For example, microformats are fairly new and there are a small number of sites using microformats today. Second, the approach directly ties web analytics deployment to the web design process which has its disadvantages. For example, what happens if you’re relying on tag IDs for web analytics and your design team redesigns the site? The dependence could break the web analytics deployment and is an extra checkpoint for your design team to consider.

In the end, you should pick the method that best works for you. However, it is important to note that whatever solution you end up selecting can fit your specific workflow.

Universal Tag is about preparing for the unknown.

Let’s face it. Implementations change because requirements change. Traditional web analytics deployments require you to know your requirements well in advance before you start tagging. Your reporting requirements (how you want to see the data) will dictate how you tag your analytics solution.

Want to see the impact of site search on cart activity? You’ll have to tag for it.
Want to see the number of times people view a product before placing it in cart? You’ll have to tag for it.
Want to see the effect of white paper downloads on lead conversions? You’ll have to tag for it.

This is costly. And what makes the process more cumbersome is the fact that requirements change. The change can be part of the evolutionary process associated with web analytics, or simply because of unexpected consequences of your implementation. In this post, we’re going to discuss what happened to a major consumer products company and how the Universal Tag saved them from re-tagging.

In this specific scenario, the company just launched a virtual world promoting many of their different brands. The virtual world is built completely in Flash and in order to track the effectiveness of the on-site promotional offers, the company decided to track the promotions as on-site campaign impressions. In other words, as a new offer (internal ad) appears on the screen, an impression tracking is sent to the web analytics tool.

This measurement framework allows the company to measure the click-through rate of the internal offers (since impressions are being tracked). However, it also results in an unexpected side effect.

Assuming that upon each offer, an impression request is sent to the analytics tool, what happens if the visitor leaves his/her computer while keeping the browser open (say heads out to lunch)?

This implementation will result in many extra unwanted calls being sent to the vendor, which results in both artificially high number of server calls (cost to the customer), as well as engagement metrics in the form of time spent on site.

With a traditional web analytics deployment, the client will have to go back to the web development team and to add a logic within the content management system that caps the number of ad impressions being sent. However, with a Universal Tag deployment, fixing this is simply a matter of adding a new plug-in to the Universal Tag library. The plug-in automatically cuts off requests after a defined number of impressions. The web development team or the agency does not have to change a thing and web analytics practitioners can be in total control of how the data is sent to the vendor.

Web analytics implementations are not always easy. Often times, as you’re getting into advanced implementations, there’s a chance that you’ll see unexpected behaviors or side effects from the implementation. With traditional deployments, these require a re-tagging exercise. The Universal Tag on the other hand lets you deploy once and fine tune without having to re-touch your page tags every time, hence preparing for the unknown.

Universal Tag is about having a better web analytics process.

Web analytics is an iterative process. A typical web analytics cycle is shown here. First, users deploy their web analytics tool. From there, they analyze the data, and make changes to their sites based on findings. The cycle then repeats itself. However, in some cases, the findings may require users to look at the data in a different angle. Often times, the new angle will require a change in web analytics implementation, which means re-tagging the site.

To demonstrate this, we’re going to discuss an analysis that we recently did for a technology company. This client sells expensive enterprise software and uses a large number of white papers in order to educate its user base. As part of the analysis, the client wanted to know if white papers have a positive impact on site conversions, which is lead generation. The client’s tool of choice is Google Analytics.

To do the analysis, we used the “Visits with Conversion” segment and looked at the downloaded files for the segment. This will show us which files were downloaded during the same session where the lead was captured. The results were initially shocking. For this particular segment, we saw about 30% less white paper downloads than an average session. Are we to believe that converting visitors are less interested in white papers than non-converting ones? This meant that we needed additional information.

The next hypothesis was that visitors download the papers, read them and then come back to the web site and submit their information. In order to prove this new hypothesis, we had to make an implementation change since Google Analytics does not provide this level of cross-session analysis without customization.

The solution was to use a visitor-scope custom variable to capture the downloaded document and look at the “download” custom variable report for the “converting” visitors.

With default web analytics deployments, this requires editing the tagging within the download pages, which is a laborious process that will involve the web development team. However, through the Universal Tag, this process be can implemented without a single page tag change.

Following this change, the discovery proved our hypothesis. In fact, we learned that it takes an average of 2 days between a white paper download and a lead registration. This exercise clearly showed the dangers of relying only on session-level data when dealing with complex sales.

Universal Tag made this discovery possible without re-tagging. Because organizations can fine-tune their implementation without costly re-tagging exercises, they can learn faster and therefore get more value from their web analytics investment than those using standard tags.

There’s more and more buzz around Universal Tagging every day and as a result, we’re getting the same questions from more people than ever: What is the Universal Tag?

In this multi-part series, we’re going to share our thoughts as to what the Universal Tag is and what it’s not.

In this post, we’re going to cover what we see as the first misconception about the Universal Tag.

Misconception: Universal Tag should be used by those that are in the process of switching vendors.

Fact: Based on two years of experience, we can categorically say that this is not the primary benefit that Universal Tag provides to clients.

Universal Tag is about fixing your current implementation.

That’s right. Out of almost 50 deployments of Universal Tag, only two clients have signed up in order to switch vendors. The large majority has no plan to switch vendors. Yet they recognize that their implementation can be vastly improved. Universal Tag provides them a platform to do just that.

The reason is because the Universal Tag provides a simplified platform for tagging compared to traditional vendor tags. It also changes the best practice implementation considerably. Today’s web analytics tools require you to think well in advance about the different types of reports that you want to get from the solution. Only after you have a good understanding of what your reporting needs are can you start the tagging process.

The Universal Tag framework changes this by letting you send generic data and map it to vendor-specific syntax at any time. This changes best practice implementations in that it lets you just send the data. The rest can be handled through the Universal Tag.

Here is a real-life example that helps demonstrate the point.

Example: Product Syntax

Consider a scenario of a product page. Within the page, you’d like to capture several components, including the product name, product size, color, and number of ratings received. As far as reporting is concerned, you’d like to get reports on top products viewed, top colors and sizes viewed, as well as average ratings of products (a numerical report) and a histogram or bar chart report of reviews (how many views for products with rating 1, rating 2, …).

Sounds simple, huh? Lets look at how you would go about implementing this with the two most popular tools on the market: Google Analytics and SiteCatalyst. The examples that we’ll use will be for a cotton shirt, color: white, size: large and a rating of 4.5.

First Google Analytics. We’re going to use custom variables to capture product name, size and color. The challenge here is that not only your developers have to know about the specific syntax, but they should also be aware of the fact that you can have visitor, visit or pageview-based custom variables. Now there’s no such thing as a numerical custom variable in Google Analytics, so you’ll have to use the event tracking feature in order to get your numerical ratings report. The implementation syntax will look something like this

pageTracker._setCustomVar(1,"product view","cotton shirt",3);
pageTracker._setCustomVar(2,"color","white",3);
pageTracker._setCustomVar(3,"size","large",3);
pageTracker._setCustomVar(4,"rating","4.5",3);
pageTrack._trackPageview();
...
pageTracker._trackEvent("product view","cotton shirt","rating","4.5");

Let’s try the same thing with SiteCatalyst. We’re going to assume that we’ll use prop1 and eVar1 for size, prop2 and eVar2 for color, prop3, and eVar3 for rating and event1 as the numerical event used to measure the average rating. The implementation syntax will look something like this:

s.events="event1,prodView";
s.products=";cotton shirt;;;event1=4.5;evar1=large|evar2=white|evar3=4.5";
s.prop1="large";
s.prop2="white";
s.prop3="4.5";

Again, you’re requiring your development team to know what different props, eVars and events are as well as the exact syntax which should be used (for example using lower case evar for merchandising).

Now let’s look at what this same implementation will look like with the Universal Tag. Here’s an example syntax:

yourdata.product="cotton shirt",
yourdata.size="large",
yourdata.color="white",
yourdata.rating="4.5",
yourdata.page_type="product view",

Now what if you wanted to deploy both SiteCatalyst & Google Analytics? No changes. The implementation will be the exact same.

This simplified implementation has several benefits. The one that’s clearly being addressed in this post is that it simplifies implementations and vastly reduces the deployment cycle. Your development team no longer has to master the analytics tool being used and can concentrate on sending the data through the simplified tag. Your business team or analytics department can then translate this data into vendor-specific syntax.

In future posts, we will share some of the other benefits that we’re seeing with the Universal Tag. Stay tuned.

Yeah baby! The Universal Tag wasn’t enough so we’re going galactic.

The Galactic Tag includes a number of exciting features, which include:

• Support for portable zip codes: the Galactic Tag includes a precise geo-location functionality, including the support for the new portable zip code system recently announced by the USPS
• ePaper tracking: the Galactic Tag is the only tag in the market that supports tracking for ePaper. In order to start tracking your printed papers, the only thing that you’ll have to do is make sure your paper has the letter “e” in it.
• Cross-Platform Video tracking: the Galactic Tag also supports the new cross-platform video platform, including videos printed on paper. This functionality  lets customers track the effectiveness of their printed commercials using the same deployment.

“The Galactic Tag represents our relentless pursuit to redefine the measurement across the marketing media,” said Mike Anderson, Tealium co-founder.  “With this release we’re redefining the market – literally.”

The Galactic Tag is available starting April 1. For more information, you can contact Tealium.

First, I need to congratulate the folks from OMS for putting together a great show. There was a record attendance of over 800 professionals covering all areas of online marketing, along with a great lineup of presenters.

During the panel discussions, one of the questions asked was how should businesses deal with multi-touch attributions.

Here’s a sample scenario to help explain the pain point involved:

A visitor is interested in running shoes and conducts a Google search for the term “running shoes”. The visitor is presented a number of search ads from competing vendors such as Nike, Adidas, and others, and decides to check out Nike and Adidas sites. The visitor gets intrigued by the Nike ID line of products and decides to conduct some further research, and even registers for the Nike newsletter. While doing research on third-party sites, the visitor sees a banner ad for the Nike ID site and clicks the banner. Finally a day later the visitor gets an attractive email offer from Nike and ends up buying the shoes.

In this scenario, the visitor has been exposed to three separate campaigns. The “running shoes” search campaign generated the awareness. The banner campaign possibly helped increase awareness and instill further trust in the product and finally, the newsletter sealed the deal. By default, web analytics providers give credit to the last campaign touched by the user. In our example, the newsletter campaign will get the credit, whereas if it wasn’t for the search campaign, the visitor would not have even been aware of the Nike ID line. In fact, two variables that make multi-touch attribution a real challenge are:

• Number of simultaneous campaigns. If you’re a company running large numbers of campaigns in parallel, you should account for multi-touch attribution
• Complex or expensive product: the more complex the product, the longer the consideration and therefore the more likely you are to have multiple touch points.

So how does one tackle this challenge? First, for the large companies running many campaigns, there are a number of commercial solutions such as ClearSaleing that help solve this challenge (and a lot more). But what about smaller companies with small budgets using free solutions such as Google Analytics or Yahoo! Web Analytics?

First, we recommend that you investigate if you even have a multi-touch attribution problem. How? Let’s take another look at our example scenario. Two metrics within your analytics solution can give insight into this. They include time to purchase and number of visits prior to purchase (or conversion).

For example, if you use Google Analytics and have e-commerce tracking, you can use the “Visits to Purchase” report to see how many times do visitors come to your site prior to purchasing. If you are a lead generation type web site and have your conversions set up as goals, you can use the default “Visits with Conversions” segment and look at the loyalty report for the segment. In both cases, if most of your conversions come from first-time visitors, then multi-touch attribution is not going to be a problem for you and the rest won’t apply.

However, if you happen to see a big difference between converting visitors and others, then you can build a quick attribution report by following these steps:

• Create a Javascript that captures your marketing campaign parameters into a persistent cookie
• The JavaScript should also be configured to append campaign values together, as visitors go from one campaign to the next
• Push the cookie value into a custom variable  – such as a visitor-centric custom variable in Google Analytics, a session-based custom field in Yahoo! Web Analytics or an eVar in Sitecatalyst.

You now have a simple yet powerful solution for seeing which campaigns your visitors are responding to, but also in what order.

Happy analyzing.

Some of the new functionality include:

• Improved multi-vendor support: the new version provides a superior method for complex implementations with multiple vendors. For example, non-technical users can map page tag values differently into various web analytics solutions, while also mapping them to their PPC bid management tool.
• Attribution management: designed specifically for clients using multiple affiliates, version 2 of Tealium Universal Tag has the ability to conditionally send data only to the winning affiliate(s).
• Multi-currency support: the new version of Universal Tag supports transactions in multiple currencies for digital marketing vendors that do not provide such support by conducting on-the-fly conversions to the supported currency.
• Universal data capture: this feature allows non-technical users to automatically capture data elements from the page and map them to their web analytics and digital marketing solutions. Examples of such data elements include microformats, meta tags, in-page style elements, query parameters, cookie values, etc.

We’ll be publishing a number of case studies on Universal Tag deployments soon. In the meantime, to see Universal Tag in action, please contact us.

First, lets start by defining home pages and their goals. The home page is typically the main gateway page for your site. It’s the first impression that your visitors will have of your site. Its role is to showcase your offerings, your value proposition and provide quick access to the most popular or important sections of your site. For this reason, web analytics should help you answer some of the following questions:

• How effective is the home page at directing visitors to product pages?
• Which part of the home page is the most effective?
• Is the home page effective at enticing visitors to learn more?

Based on these, below are some popular web analytics reports for home page analysis along with the explanation:

• Bounce Rate
• Micro Step Conversion Rate
• Conversion Rate
• Acquisition Sources
• Home Page Real Estate

Bounce Rate

The bounce rate is defined as the number of bounces (single page visits) divided by entries. It shows you what percentage of the traffic landing on the page bounces and does not view any other page on the site. It is a reflection of the home page’s ability to retain visitors. Clearly the goal is to make changes to the home page and lower the bounce rate. It’s probably one of the best reports to look for when analyzing home pages. This report is widely available in most web analytics tools such as Google Analytics, Yahoo! Web Analytics and Unica NetInsight.

Micro Step Conversion Rate

Although the ultimate goal of your site is to drive conversions, we recommend micro step conversions as a better way to assess home pages. The goal of your home page is to drive people to your product description pages. It’s at that level that you do the selling. For this reason, when assessing the success of your home page, it should be around its ability to get visitors to those ensuing pages. You can get this in a number of way. Inside tools such as Yahoo! Web Analytics and SiteCatalyst, you can tag your product description pages as events and look at the success of your home page around this event. In Google Analytics, you can create a goal for your product pages, as long as the pages have a consistent nomenclature. If not, you can create an advanced segment for your product pages and look at the home page traffic for the segment. Such metrics can pretty easily be created inside Unica NetInsight and Webtrends.

Conversion Rate

Yes, this should not be your primary report for home page analysis, but you can still use this report as a tie-breaker. For example, if two versions of home have similar bounce and micro step conversion rates, then you can use the overall conversion rate to see if one version does in fact do a better job. Unfortunately, we often see that many people use conversion rate as the primary report for assessing home page effectiveness.

Acquisition Sources

Want to lower your bounce rate? One place to start is by looking at the acquisition sources. You can start with the sources of traffic to your home page and look at their respective bounce rates. Start with referring sources with high bounce rates. Often, you’ll find a messaging gap between the referring sites and your home page. The referring site may be saying something while your home page could be promoting something else. While you cannot optimize your home page for all referring sites, you can start with those with high traffic and high bounce rates and provide messaging on your home page that helps retain this incoming traffic. You’ll typically find that a handful of sites may account for a high percentage of your bouncing traffic.

Home Page Real Estate

To understand the real estate effectiveness, you’ll have to look at the click activity on the page. Rather than looking at all page links, we recommend classifying the link into sections or categories (such as as header, footer, navigation, left box, right box, etc.), and analyzing the activity by such sections. This is different than the default site overlay that you typically get from web analytics tools and requires some additional configuration to get proper reporting. For example, if you’re using Google Analytics we recommend using Event Tracking to track the activity on various sections and links within sections. You can then see how effectively each section and each link gets visitors to product pages and to final conversion.

You can also investigate some of the in-page analytics tools such as CrazyEgg and ClickTale, which do a more thorough job of providing such reports than web analytics tools.

Of course, depending on your business, your reporting needs may vary, but we believe this list should provide a good starting page for optimizing one of your most important pages.

Page views

This is probably the best out-of-the-box metric for measuring the success of a content. The more the number of page views, the more popular the content. However, relying on this number alone has two potential shortcomings. First, it fails to differentiate between segment traffics. For example, a loyal visitor is more valuable to a content site than someone who visited the site for the first time and will likely never come back. Also, page views alone fail to report the level of engagement on the page. For example, visitors could be clicking an article and spending only a few seconds on it. The quality of traffic should therefore be accounted for.

Time spent on page

This metric clearly adds a new dimension around engagement. The more time visitors spend on the content page the more engaged they are. However, you cannot rely on this metric alone. One key reason is the fact that this metric is not always available to all visitors. For example, if the content page was the only or the last page viewed during the session, then this metric is simply not calculated within popular web analytics solutions (we’ll discuss this in a separate post).

Another shortcoming of this metrics is that like page views, it fails to segment the reports by the quality of visitor (first-time vs. loyal).

And finally, the Time Spent metric alone does not take into account the popularity of the content. For example, an article could be very engaging but only be viewed by a handful of people.

Visitor Loyalty

Web analytics solutions often provide this in context of the overall site traffic and you may have to do some tweaks to your reports to get this, but it’s important to note what percentage of your content is consumed by first-time visitors and what percentage by loyal visitors – visitors that come back to the site. The reason this is important is because in the long run, you may want to create a loyal following and create content that’s tailored to them.

Bounce Rate

This is also one of the most popular metrics within analytics solutions, but media sites should be careful not to over-analyze their bounce rates. As an example, consider a media site with an RSS feed. Through the RSS feed, visitors can see the headlines of new content using their favorite RSS aggregator. If an article looks appealing, they click the link, enter the site, read the content and then leave. That’s a bouncing visit but still a highly qualified traffic, because the visitor has subscribed to the RSS. The visitor loyalty metric indirectly takes care of this shortcoming.

Content Engagement Score

In this post, we’d like to introduce you to a content scoring KPI that we’ve used to help some of our media clients put a monetary value next to their content.

The formula is as follows:

Engagement Score = (Page Views × Avg. Time Spent  × Avg. Loyalty)

Where:

• “Page Views” is the number of times the page was viewed during the reporting time period.
• “Avg. Time Spent” in the average number of seconds or minutes spent on the page by visitors.
• “Avg. Loyalty” is the average number of visits to the site by your visitors (1 for first time visitors, 2 for those who’ve been to the site twice, and so on).

Of the three metrics needed to create this KPI, “Avg. Loyalty” is the most difficult to get, but this can be obtained done using estimates in popular tools. For example, with Google Analytics, you can use the %New Visits metric to estimate the average loyalty. You can use the following formula for this purpose:

Avg. Loyalty = (%New Visits) + 2 * (1 - %New Visits)

What this formula does is that it assigns a score of 1 for each new visitor and a score of 2 for all others, providing a reasonable approximation. You can create a similar model with Yahoo! Web Analytics – see below figure for an example of such report in Google Analytics.

Using this model, pages with the highest traffic, time spent and the most loyal visitors will get the highest scores, which is the desired outcome. You can of course use any analysis tool to create your score. One popular tool is Microsoft Excel, where the score can easily be created and analyzed. See figure below for an Excel example. It shows that our posting for tracking internal campaigns is the most engaging even though it’s an old blog post.

Overall, this model provides a simple KPI for measuring site content, while taking into account the popularity, engagement and the quality of the visitor. It does however have its shortcomings. The primary shortcoming is that it is dependent on cookies. For loyalty to be counted, visitors have to accept cookies. Furthermore as visitors delete cookies, it will impact this KPI. However, it’s fair to assume that visitor cookie deletion is not dependent on their content preference, so you should expect the same rate of deletion across the board.

The metric also depends on time spent reporting, which is not available to all visitors. Having said that, it’s also fair to assume that the time spent by those who view a certain content as their last page should be inline with those who view the content in the middle of the session. After all, the purpose of this model is to provide an approximate score for content engagement and popularity.

You may also be in a mode where loyal visitors are no more valuable than first-time visitors. For example, newer web sites fall into this category. In that case, you can simply omit the “Avg. Loyalty” metric from the formula (or replace it with the value 1).

So there you have it. We welcome your feedback on the model and hope you find it of use.

Happy Analyzing!

It’s been a super busy fall with client projects, especially with Universal Tag implementations and as a result, this blog has suffered from lack of attention. We promise that we’ll be putting more effort in the comings weeks.

Happy holidays from everyone at Tealium. We wish you a great holiday season and happy new year.