Why Tracking Software Programs Yield Different Results

I’ve heard from a number of insiders at big general job boards like Monster, Careerbuilder, and HotJobs that 90 percent or more of their revenues are derived from the sale of job posting ads and resume searching. If that’s true, it is no wonder that they are reluctant to follow our lead by eliminating resume searching in order to help the candidates using our sites obtain the security and privacy they deserve. But that’s another topic for another day. Well, almost.
Unlike the big general boards, our niche board has for years generated most of its revenue from non-traditional job board products such as targeted email campaigns and targeted cell phone text messaging (SMS) campaigns. I know that we do a great job for the vast majority of our clients and occasionally fall flat on our faces. But as frustrating as it can be when we can’t drive the right traffic to our client’s web site, it is even more frustrating when we’ve driven the right traffic yet they’re not seeing it. More often than not, the problem is with the web traffic tracking software they’re using or how they’re using it.

Judah Phillips recently published an article that did an excellent job of outlining a number of causes for why a publisher like CollegeRecruiter.com may track X number of click throughs to our client’s web site yet the client tracks Y number. You’d think that if we tracked 1,243 clicks that the client would track the same or virtually the same. But if so then there would be an excellent chance that you’d be wrong.
Quite simply, “counts of the identically named metrics from ad servers don’t match the web analytics tools, which don’t match the for-pay third-party audience measurement tools, which don’t match the free audience measurement tools, which never match any of the homegrown internal measurement tools. And none of them ever match each other.” The differences are due to a variety of reasons, including:

• Different data collection methods are often used by different software programs. Banner ad serving software such as ours uses javascript page tags but web analytics software such as Google Analytics use page tags, log files, packet sniffers, or a combination of these. Oh, and each of the tools have different algorithms for processing the data so even if your software appears to use the same method as mine, they don’t.
• Some clients rely on audience measurement tools such as Alexa or Compete but these panelists must install toolbars or other software. Few workplaces allow the use of such software so the audiences are heavily skewed to home users and heavily skewed to those who are willing to have such software on their computers. Quite simply, these are not random panelists so to rely on the traffic measurements that they produce is pure folly. Perhaps a decent rule of thumb, but not accurate and definitely not scientific.
• A number of software packages rely on small files called cookies being placed on the computers of those who view and click on the ads. Unfortunately for the vendors who sell this software and the clients who buy it, cookies are not a reliable measurement tool. The new browsers routinely silently block third party cookies, meaning that if someone comes to CollegeRecruiter.com and we use a third party to serve our banner ads (we do) and that third party serves a cookie onto the computers of our visitors to help better target the ads (it does) then those cookies will often be blocked by the browsers of those users and any software trying to accurately count how many visitors we have or how many pages they’ve looked at will end up with inaccurate counts. Similarly, if we deliver a targeted email with cookie-based tracking then the vendor trying to count how many people looked at the email, clicked on it, or otherwise responded to it is in for a very difficult process. Actually, they’re in for an impossible process because the data simply isn’t there in order to accurately measure the traffic.
• Different data models are used by the software that serves ads and the software that measures traffic to web pages. Ad server software serves and counts impressions, clicks, and other such measurements. The metrics are based on how many ads have been requested from the ad server, not on which page the ad was served to. In other words, ad serving software isn’t looking at impressions the same way that web page tracking software does. The definitions are simply different.
• Some web pages are missing the code needed for the web tracking software to do its job. This isn’t really a software problem because the software is counting what it is told to count and that doesn’t include those pages, but it is another problem that occurs because ads might send people to a page that doesn’t have the web traffic software code on it or there may be one or more links on the page to other pages that don’t have the software properly added. No code means no tracking of traffic.
• More users are blocking javascript or other ad serving software. Some of these allow the ad to be served but then don’t display it on the screen. So the ad serving software counts the impression but the web tracking software doesn’t because the ad didn’t appear.
• Web traffic tracking software is notorious for not properly tracking unique visitors or page views whereas ad serving software tends to do a much better job at this. If one visitor goes to a page 10 times, web traffic tracking software will tend to count that as one views. But ad serving software is concerned with displaying 10 different ads to that person so counts that as 10 views.

Unfortunately, there is no consensus in the industry for how to measure web-related traffic. A number of industry organizations, vendor, and users recommend competing standards and many of them have great merit even though their recommendations are inconsistent. Which one is “the best” approach? Put 100 of these stakeholders in a room and you’ll get 100 opinions.
So what can we do to help our clients when they see a different number of click throughs to their web page than we report? We need to do a better job of explaining that the software they’re using may be using the same terminology as our software but they’re actually tracking two different things. Apples are great and so are oranges and they’re both fruits but they’re different. And so it is with web traffic and ad serving software.