Is your on-line advertising delivering more than your site analytics is telling you?

As someone who has been analyzing on-site web data for a number of years, I have observed a phenomenon that for the lack of a better term, I call the 'fuzzy factor'. It refers to the differences observed in conversions reported by the analytics tool (in my case I have been using Google Analytics) from paid advertising and the conversions reported by the advertising platforms (Adwords, Yahoo, Bing, Facebook). Note that when I refer to the term 'conversion' in this context, I am referring specifically to a conversion that is synonymous with a 'purchase' transaction and not to any other conversions that I may be tracking such as a newsletter signup.

My process is as follows:

• When I create an ad on any third party ad delivery system (like Bing, Yahoo, Facebook) I ensure that the ads are tagged accurately. Since I use Google Analytics I encode the destination URLs using their utm_campaign, utm_medium, utm_source, utm_term and utm_content parameters for every ad that is run. I use Google's autotagging feature so that Google does the work for me for all the Adwords ads.
• I install the conversion tracking code supplied by the ad platform on the conversion page and then let the data flow.
• I then compare the transactions attributed to the various ad platforms by Analytics to the attributions made by the individual ad platforms and measure the differences.

I know there are a number of reasons for the differences, a major one being that Google Analytics uses a 'last' clicked source to attribute the source of the conversion while the ad platforms (including Adwords) begin tracking a cookie from the time the ad is first clicked and attribute a conversion to that click provided the cookie has not expired. 

So, if someone comes to a site by clicking on an Adword ad, leaves without making a purchase, returns the following day by virtue of a Google organic search on the brand name and makes a purchase, Google Analytics will attribute the conversion to Google organic search while Adwords will record a conversion that is attributed to the ad that was first clicked. For GA the score is Organic 1 CPC 0 while the score for Adwords is CPC 1 Organic 0.  

My analysis has shown a consistent discrepancy as I look at data going back about 2 years. (Caveat: Facebook only began conversion tracking on their ad platform in early Feb 2010 so my Facebook experience dates back to late Feb 2010 when most of the bugs were ironed out).

In the case of my personalized book sites, Alphakid, Printakid and Livrepersonnalise, this discrepancy is about 15%. For these sites, the advertising platforms are attributing 15% more conversions to the ads than Google Analytics is attributing to them and the overstatement percentages are consistent from ad platform to ad platform.

This is actually not surprising since not everyone who makes a purchase on the sites does so during their first visit. It may take a couple of visits with subsequent visits more likely to originate from non-paid sources.

So what does this mean for me? Here are some of the implications: 

1. Calculating advertising ROI based solely on Analytics conversions will understate the value that the advertising brings to the business.  
2. The initial ad that drives someone to a site is an important source of value and some of the value is hidden. Since not everyone's first purchase coincides with their first site visit, the purchase may never happen without that initial click on the ad. (I can also confirm this via GA's stats on days to purchase and visits to purchase)
3. A more accurate ROI calculation is needed.

For each channel/campaign I determine an Adjusted ad contribution:

Adjusted Ad Contribution ROI = (Adjusted Revenue from Campaign based on Ad Platform * Gross margin - Cost of Advertising) / Cost of Advertising.

I then re-adjust the calculation for lifetime customer value rather than one off purchases. While this adjustment cannot be precise (do you know how many purchases someone is likely to make if your site is new, can you estimate how ad costs might change over time, how your product mix may change thereby affecting average purchase value and your cost of goods) at least it will provide some insight. As a proxy you can use Google's new custom variables to set up visitor based segments for your ad channels and measure lifetime value over a two year period (the 2 years before the cookie expires).

While this might sound a bit complicated, the results are certainly worth the effort, particularly if your web business includes a significant paid ad component.

Has anyone else had similar experiences comparing the data from their ad platforms with the data from their analytics tools?

Happy analyzing!