Making Sense of the Numbers
April 1st, 2008 by Joy BrazelleOne of the biggest frustrations of running campaigns on search engines or other ad networks is the fact that the numbers just don’t add up. When logging into Google, you see the number of clicks. But, looking to your Web analytics for the number of clicks can be quite different.
Much has been written about the fact that the numbers don’t and shouldn’t match. But, what is not clear – is why they are different and how to reconcile the numbers.
The first thing that must be done on any ad campaign driving traffic to your site is to make sure the campaign is ‘trackable’ in your Web analytics. To do this, you must have a unique URL.
A few years ago it was not uncommon to make a copy of a page and name it something different (e.g. www.mysite.com/homepage2.asp). However, this quickly becomes an unmanageable option, especially if you are running multiple campaigns and need to make changes and updates to pages.
Luckily, someone figured out that you can modify a URL and make it unique simply by adding parameters. Rather than setting the destination URL to:
www.mysite.com
add parameters
www.mysite.com/?source=google&c=campaign&ag=adgroup
Your analytics package may require specific parameters in order to automatically see data in specific reports. For example, to see data in Paid Search reports in WebTrends -you must use the parameter ‘WT.srch=1’ so the URL would be:
www.mysite.com/?WT.srch=1&source=google&c=campaign&ag=adgroup.
That way you could see data populated in the pre-configured Paid Search reports, but you still can configure a Parameter Analysis for ‘source’ to compare different campaigns (e.g. Google, Yahoo, email). Other analytics packages ‘don’t care’ which parameters you use, so you can create your own naming convention for parameters.
One last note about tracking parameters and Google – You can choose to set destination URLs up at a creative level if you want to track the performance of different creatives (ads) or at a keyword level if each keyword in you want to track performance of individual keywords (or if each keyword in an ad group needs to go to a different page). Setting tracking parameters up at a keyword level is often more work (since most ad groups have more keywords than creatives) but it also can give you a lot more insight to what people are actually searching on, especially if your keywords are set to ‘broad match.’
Once campaigns are set up to be trackable in your web analytics, you can feel much more comfortable about comparing numbers. The numbers that you want to be able to reconcile are cost, clicks and conversions.
The obvious place to find the cost of the campaigns is directly in the search engine or ad network, as that will show you what you’ve been charged for the campaign. It is also worth comparing the cost in your web analytic package. It may be that you can import your campaign information into your analytics package, like ClickTracks (below). In that case it is simple to see the information.
What you are likely to see is that the number of clicks is less in your web analytics than the search engine/ad networks report. The discrepancy will likely be between 5 – 20% fewer clicks in your web analytics than reported in the search engine.
There are several causes for this. Someone clicks your ad and they don’t wait for the page to load (especially if you are using java script for data collection in your web analytics), they hit the back button, close their browser or maybe their computer crashes. The ad was clicked, but the visitor just never made it to your site.
For discrepancies higher than 20 – 25% make sure that your links are good, that you are not sending traffic to a page that does not exist on your site. If everything is set up correctly, it may be worth contacting the search engine for further investigation as it may be a sign of click fraud.
To reconcile conversions you can likely see conversions in your web analytic package and in the search engines and ad networks (if you’ve set the campaigns up to track conversions). These numbers can be slightly off due to how each system handles and credits latent conversions. If the numbers differ greatly, then it is worth further testing to see which system is set up correctly.
One final note about conversions, it is highly likely that the number of conversions will never 100% match between your web analytics reporting and your ecommerce database. The number could be over-reported in your web analytics if it is not set up to only count one conversion during a visit (so if someone reloads the page or hits the submit button several times, the order will be counted several times).
The number could be under-reported for several different reasons in your web analytics. For example if you are using java script for data collection, there are a small percentage of people who have javascript disabled on their computers. Your ecommerce database is obviously the most accurate place to look when determining sales and revenue information.
Some shopping carts, like Volusion, track sales and revenue information, as well as campaign information which is great. But for many online retailers who are using a different shopping cart reconciling conversions ends at comparing the search engine/ad network number to web analytics.
It takes practice and disciple but it is definitely worth understanding these differences and tracking them to make sure that you are spending your ad dollars wisely.
Related posts:
