Here at Red McCombs Media, online advertising is central to our company. One of the things that has come up a handful of times in recent weeks is a discrepancy between the number of clicks being reported, and the number of visits being recorded by the client’s analytics package. I thought it would be helpful to explain this discrepancy publicly for the benefit of all.

Before discussing the differences it is important to understand the terms, so here are a few definitions.

DART is DoubleClick’s campaign administration tool and is one of the industry standard tools available today. Red McCombs Media uses DART for Advertisers to count the number of clicks a banner receives.

The term analytics refers to any web analytics package. Google Analytics is probably the most widely known, but the same information applies if you’re using Clicky, Omniture, WebTrends, Mint or any other.

A cookie is a small piece of text stored on a user’s computer by the browser, and may contain user preferences, referral information, or other data used by websites. Most web analytics packages, including Google Analytics, require the use of a cookie.

A publisher site is the site on which a banner or advertisement appears.

A landing page is the target webpage for a campaign.

A click is counted by DART when a person clicks on a banner or advertisement on a publisher’s site and is redirected to a landing page to find out more information about the promotion. This is in accordance with IAB click measurement guidelines. A click is not counted when a user interacts with the banner content in some way (expanding to find out more, retracting or closing an auto expandable, adjust volume levels, play a game within the banner, etc). And, a click is not counted when it is automated via a script or web crawler.  The only way a click is counted is if a person’s web browser exits the publisher’s site.

A visit is counted by analytics when a person interacts with a landing page through a web browser, and ends when the browser is closed or shut down. This term is interchangeable with the term session. A script or web crawler would not be counted as a visit, because they are not operating within a browser.

Because clicks and visits are two different metrics being measured by two different services across two different websites, it should come as no surprise that discrepancies exist between them. Regardless of the campaign it is unlikely that there will be a one-to-one correlation between clicks and visits, and in most cases the number of clicks will be greater than the number of visits.

There are a number of reasons for discrepancies between these two metrics. According to Google Analytics Help there are four explanations for a discrepancy between clicks reported by AdWords and visits recorded by analytics. Three of these four factors also apply to banner advertisements:

• A visitor may click on your ad multiple times. If a person clicks on your ad more than once during the same session, each click will be counted but Analytics recognizes each pageview as one visit. This is a common behavior for anyone that is comparison shopping. Multiple clicks equals one visit.
• A user may click on an ad, then later, during a different session, return directly to the site through a bookmark. In this case the referral information would be retained so one click would result in multiple visits. One click equals multiple visits.
• A visitor may click on an advertisement then click back, stop or close the browser before the page fully loads. In this case the click would be counted but Analytics would not record the visit. One click equals zero visits.

The issue of correlating clicks and visits becomes more complicated when searching within analytics.  It can be challenging to identify which visits are from a specific banner campaign. The easiest way to identify the campaign traffic is through referring sites. This is how most clients identify their campaign traffic. Yet, according to documentation within DART for Publishers (login required) this is an unreliable measurement. Because there are a number of ways for an ad to be implemented on a publisher’s website, analytics could report this traffic as being referred by the page where the ad is displayed or DoubleClick.com.

We at Red McCombs Media have even seen instances where the referral information is not passed at all. This can happen when a campaign banner is served by two or more ad servers, and the cookie is either corrupted or stripped. This is an artifact of the way that the banner for a campaign is distributed to publishers. In this case the click would be counted by DART and the visit would be counted by analytics, but it would not be attributed to a referring site at all.

For these reasons it is unlikely that the number of referrer visits will match the number of visits generated by a campaign, increasing the discrepancy between clicks and visits. There are, however, steps that can be taken to ensure that analytics counts as many visits as possible.

• Use Google’s URL Builder to tag the campaign. This ensures that all of the visits that are counted will be correctly segmented and labeled, making them easy to find. If you’re using something other than Google Analytics, there should be a similar means of campaign tracking available for that specific tool.
• Ensure that the same Analytics tracking code is installed on every page of the landing website. This ensures that if a visitor navigates to another page of the site it isn’t recorded as a bounce.

Note that even with these measures the number of visits will still not match the number of clicks.

Since any comparison between clicks and visits requires data from DART and analytics, how accurate is the data being reported by these services?

According to DART’s counting methodologies whitepaper (login required), their algorithm takes measures to prevent automated clicks and they are excluded from reports if detected. Over time this strategy has proven effective, very few clicks by real users are discarded, and once a user is identified as fraudulent all subsequent clicks are discarded.

Analytics packages are generally slightly less accurate when reporting the number of visits. As mentioned above, most analytics require the use of a cookie. Users have the option of blocking or deleting cookies and ad servers may also affect cookies. According to Brian Clifton, author of Advanced Web Metrics with Google Analytics and a whitepaper on the accuracy of analytics software, these factors lead to most analytics services (including Google Analytics) slightly under reporting the number of visitors. This would be in addition to the factors suggested by Google, increasing the discrepancy between clicks and visits.

Now that you understand the factors that contribute to a discrepancy, consider its magnitude. Our experience has shown that the size of the discrepancy between clicks and visits is different for each campaign. Because of this, it’s not safe to assume what the discrepancy should be before a campaign begins. Any assumption about what the discrepancy ‘should be’ is arbitrary at best.

Because of the discrepancies between clicks and visits, in our experience, a much more meaningful metric is the landing page conversion rate. This is the rate at which visitors to your landing page convert (complete a purchase, fill out a form, download a file, etc). While it is not possible to predict what the conversion rate will be before a campaign begins, once the data begins to appear there are a number of things that can be done to optimize the campaign to convert as many visitors as possible. By taking steps to optimize the conversion rate, advertisers can maximize the return on their advertising dollar. There are quite a few books available that cover the topic of landing page optimization. If you want to avoid all that reading, Red McCombs Media offers this as a service.

There are simply too many factors that contribute to discrepancies between clicks and visits for them to be considered the same, they are two different things measured by two different tools across two different domains. The supporting articles provided should show that the industry agrees. In short, this is not an actionable metric. Landing page optimization, on the other hand, is an effective means of evaluating the performance of a banner or search campaign. Do you agree or disagree with the conclusions drawn here? Please post questions or comments below.

In the coming weeks this blog will be updated with tips for optimizing the conversion rate of your campaign.