Tracking the Effectiveness of On-Site Promotions

The hard work has been done and you've got the visitor to your site, now's the time to sell to them. Google Analytics, Omniture, WebTrends, Unica and other tools can tell you a lot about how the visitor reached the site and whether they became a lead or a customer. But what about the effective are the messages displayed within your website?


Landing pages, special offers, key messages and other factors within the website can strongly influence the final outcome. A poor message, an ineffective call to action or weak content could be preventing your site from producing the best results.

There has been a lot of discussion around testing landing pages during the past few years and this technique has a lot of potential, but what do you test and when? What is the impact of adding a promotional tile to the side bar on your site or adding a slide in overlay? How useful is the headline on the persuasive copy page in increasing the visitor's awareness and desire to act?

Using Custom Variables to Track On-Site Promotions

With some adjustments to your website you can track a number of these influencing factors through the use of custom variables. All of the tools mentioned above provide this feature and using it effectively can lead to significant insights and these insights can lead into a systematic testing program to improve performance.

To illustrate how this can be done, I will outline the process tracking a promotional tile which appears across many pages within a website. This is to answer the business question; "do visitors use this tile to get to the product page and purchase or do they navigate a different way?" Our initial assumption is that the tile is driving sales of the product, but as the product can also be found by browsing the site and via the site search engine, is this tile actually boosting sales or does it need work improving?

To start the process, adding an additional tracking parameter to the link on this tile will leave a marker that we can use to track whether the visitor clicked on the tile or not. As Google and other search engines use the URLs on the site to crawl and index the content it is important that we either use the # value or apply the tracking parameter via JavaScript so that the modified URL doesn't make its way into the index of the search engines. For simplicity in the implementation I tend to use the # value. E.g. http://www.yoursite.com/product-name.html#ad=hp-offer1. As search engines don't index this part of the URL it won't impact search rankings.

When the user clicks on this link the resulting URL has the # value and this can be checked by JavaScript and a custom variable set based on the value in the URL. As this custom variable relates to tracking a sale, the variable must be set to be valid for the entire visit. In Google Analytics this means that the scope is set to the visit level (2), within Omniture this is an eVar and WebTrends set this to "most recent value" or another suitable value but not "all hits".

Next we need to define a counter that tracks the sale. This can be the revenue generated, the number of transactions or goals achieved. This combined with the number of visits to the product page provides the basics to do this analysis.

Analysing the Results

Once we have run the tile for a period of a week or so we can start to analyse the results. Ideally you want to create a table that shows the following data:

The results show that there is a difference in the number of visitors who clicked on the tile to the final sale of the product. Checking these figures for statistical significance indicates that the difference is not due to random errors but the results seem counter intuitive. Why would visitors clicking on the promotional tile have a lower conversion rate than visitors who didn't? So we succeeded in answering our original business question but in doing so we just raised another question which needs to be answered.

The answer may lie in whether the tile is triggering sales or not. For instance, it may be quite plausible that making a special offer is driving these visitors to check whether they can get a better offer elsewhere. Also the mere presence of the tile itself actually influences sales without visitors feeling the need to click on it.

As such our business question is changed to "what is the impact of displaying a promotional tile on the site to our revenue for those products?"

To answer this and similar questions we need to run controlled experiments to measure the impact on sales of not having a tile at all. I will cover this topic in a future blog post.

For now I hope that I have raised your interest in measuring the performance of your "on-site" advertising and its importance in improving the performance of your website.

For assistance with any of the issues raised in this post please contact Panalysis panalysis.com/contact_us.php

Why Not Use Google Analytics Campaign Tagging?

You may think that just adding in the campaign tracking tags to the promotional tiles will work. For instance Google Analytics makes tracking advertising campaigns easy; just add in the campaign tracking tags add the link and the campaign starts tracking right away.

These campaign tracking tags are designed for external campaigns and should only ever be used for campaigns that refer visitors to your site from another channel. It is very important that they are never used for tracking clicks on advertisements within a website e.g. a tile with a special offer on the home page. The reason for this is that setting the campaign variables for internal campaigns plays havoc with the data that is collected from the external campaigns. Internal tracking should be kept separate as it has its own methods of analysis.