Mymotech

Comscore recently published a press release about 1st and 3rd party cookie deletion rates that they measured in the month of December.

Here are the important details.

• Duration: December 2006
• Sample Size: 400,000
• Sample Site: Portal with content and an Ad serving 3rd party cookie
• Average number of 1st party cookies per computer measured: 2.5 (100 visitors becomes 250 visitors)

So, the gist of this press release is to say that 1st party cookies are an imperfect method for tracking unique visitors over time, and that using them will inflate visitor counts by as much as 150%.

OK, then what should a web analytics person do?

1. Don’t panic - Everybody already knows that 1st party cookies are an imperfect form of tracking unique visitors over time. Cookie deletion happens, people surf at work then at home, and a whole multitude of other scenarios that pretty much make 1st party cookies a crap shoot.
2. Look and see what metrics you are using that depend on unique visitors. Here are two samples:
• Visitor Conversion - This could be a biggie, but guess what; you are not overstating conversion you are understating it. Grab last months Visitor total and divide by 2.5 now use that number to calculate top level conversion. (Orders/Comscore Visitors) Boo Yaa right? For all you web analysts that get paid by conversion boosts, it is time to figure out where the pool goes. And practically speaking this doesn’t really change all that many things except to normalize and baseline at a different level. Obviously I am keeping it simple. This is a blog post after all.
• Campaign Response - Ok this one could tighten a little bit, especially when the sales cycle for your product is longer than a day or two. So what can you do? Start by understanding the difference between regular old visitor conversion and a campaign visitor conversion. Then get about learning how to understand your entire conversion process from offline to online and vice versa. Do some surveys to understand visitors who purchased better, and quit demanding a 4 to 1 ROI on Adwords. In actuality these numbers should drive you to your customers for answers and that is a good good thing.
3. Next start to redefine your metrics realistically. You have to fight against the current of web analytics vendor marketing speak, but you can choose and use metrics that do not leave you over-extended and washed up when Comscore points out the existing flaw in your underlying data. This is actually one of the harder steps, because it takes a special analyst not to jump on the next greatest metric band wagon, and to realize that every metric that gets used in the rest of the business world is not necessarily ready to be used in web analytics.
4. Don’t start writing the business case to move to a complete session based cookie tracking system yet. Comscore only surveyed one site and that with their panel of users who are provided spyware tools that may or may not recommend the removal of 1st party cookies. See Eric Peterson’s Blog for more on that.

So, these are interesting times. My advice: beware anyone who gets up in arms on either side of this issue about cookie deletion. Most likely they are trying to sell something. Get to know your visitors and customers. Cold stats are good for breakfast but you need more than that to sustain you.