The Clickable Search Engine Marketing Blog

One of the most important components of an effective search marketing campaign is conversion tracking. With a few simple snippets of code, you can track the networks, keywords, and ads that generate desired conversion events on your Web site, and then use that information to optimize your campaigns. This is the major advantage of online media over radio, print, outdoor and television. However, if you advertise on multiple online ad networks and use a variety of conversion tracking tools, you have likely found that they never seem to perfectly agree. There is almost always a discrepancy between different tracking sources. For advertisers who are carefully measuring performance and optimizing their campaigns, this can be quite vexing. How do you know which numbers to base your decisions on? How big of a discrepancy is normal? If there is a problem, how do you identify the source?

In this post, I will provide some techniques, strategies and rules of thumb for coping with this frustrating but unavoidable problem. You can be a fastidious, quantitative optimizer, in the face of inconsistent data — it just takes some care and patience. Many advertisers run at least two different conversion tracking systems on their sites: Web analytics, such as Google Analytics, and a publisher conversion tracking pixel from Google AdWords, Yahoo Search Marketing, or Microsoft adCenter. Some advertisers I know run still more: pixels for other ad networks like Facebook or MyAds, or a unified multi-publisher conversion tracking system like Clickable Conversion Tracking. Usually, when advertisers implement conversion tracking pixels, it's because they're measuring stats like conversion rate, and doing optimization or making other economic decisions based on that data. But, when tracking systems disagree, what do you do?  Which numbers do you use as the basis for your optimization?

If you encounter a situation in which there is a dramatic difference between two sources, there is likely something wrong. To troubleshoot the situation, I recommend these steps:

1. See if you really have a conversion tracking problem, or a click/visit counting problem. Compare the clicks/visits numbers, and see if they are as far off as the conversion numbers. If you have a click/visit counting problem, it probably means some of the traffic is getting miscategorized. The surest symptom of this is unexpected traffic showing up in places where you didn't expect it. For example, if you're finding that many of your clicks from Facebook aren't showing up as visits or conversions in Google Analytics, check and see if you've been getting larger-than-expected referral traffic from Facebook.com. (I've provided more details on this problem, known as "breakage," in a previous post.)
2. If you have a conversion counting problem, check to make sure the pixel really does appear on the conversion page. (The problem may be that the pixel is not there, or is only there for a certain sub-set of your traffic.) To do this, it's best to actually go through to the conversion page yourself, and see what shows up in the source code of that page. Perhaps you inadvertently put one of your pixels on the wrong page, or perhaps there is a cut and paste error in the pixel.
3. If everything seems like it really should be working, but isn't, another great diagnostic tool is to go through the entire conversion process, using a tracked link, and then look at the data.  In order to see your specific visit, you should try to uniquely identify yourself. In Google Analytics, you can do this by adding a parameter to your URL that you don't normally use. For example, if you don't usually use the utm_term ("keywords") or utm_content ("ad content") parameters, append an extra one of these to your URL (e.g., &utm_content=Feb23ConversionCountingTest). See if your conversion gets counted as you expect or not. Or, with a Google AdWords or Yahoo Search Marketing pixel, try your visit through a link in a paused ad group or campaign. There should be no other conversions counted besides yours.
   

Obvious and major discrepancies are easy to spot, and usually easy to fix.  A more difficult problem is the fact that conversion counting systems can disagree by small amounts. What do you do when your systems are off by 5%, or 8%?  These cases are usually not major problems with the overall setup. They are just discrepancies within the systems themselves. Some systems do a better job than others filtering out duplicate conversions caused by users reloading or navigating back to the conversion page. Some systems have better performance for users with no javascript. Some pixels are higher on the page than others, and are thus more likely to fire in the case of an incomplete page load. How do you cope with this? I have two suggestions:

1. Implement a reliable multi-source conversion tracking solution. You can't expect Yahoo and Microsoft conversion tracking to function in the same way. But you can expect Google Analytics to treat them in the same manner, provided you have your links formatted correctly. Convenient options for a multi-source conversion tracking system include Google Analytics and Clickable Conversion Tracking.
2. If you are experiencing a click or visit-counting problem, don't optimize off of a conversion rate number. Instead, use CPA (cost per action), and figure out how much you're paying per conversion. Your bottom line is that you spend X dollars, and in exchange you get Y number of conversions, which have a monetary value for you. It is crucial for you to know what your costs and revenues are, but knowing how many clicks/visits were involved in generating your costs and revenues is actually beside the point.

There is no perfect solution for Web tracking, whether you're attempting to measure clicks/visits or conversions. And because all conversion tracking systems include some error, any two systems are very likely going to disagree.  The key to managing and optimizing your online advertising in the face of this challenge is finding a system that you're confident in and sticking to it. If you don't have a primary system with numbers that are reliable, you need to troubleshoot that system and get it working right. Once you do have reliable numbers, don't get too alarmed when other sources disagree -- just make the best decisions you can off the most reliable information you have.  That's as good as it gets, and it's actually pretty good.
 
Ehren Reilly, Clickable Guru
Note: Clickable employees volunteer several hours a week to helping other search marketers succeed. "Clickable Gurus" participate in numerous online search communities to provide straightforward answers to numerous questions, and, each week, one of the gurus posts a search-marketing tip to the Clickable Blog.