Cookie Deletion Rates and the Impact on Unique Visitor Counts

In a nutshell, what we found is that about 3 in every 10 Internet users delete their cookies in a month, with an average deletion frequency of about 4 times per month. First and third party cookies showed approximately the same deletion rates. Because server logs count unique cookies, this means that, unless some adjustment is made to the server data, the deletion of cookies and the subsequent insertion of new ones will result in site servers producing inflated counts of UVs while ad servers will overstate reach and understate frequency. The over-statement in UVs and Reach is by an average factor of 2.5, which is to say by about 150%.

While there are some direct benefits to Comscore coming out of this study (it does, after all, serve to reinforce the need for panel-based measurement data such as provided by Comscore), our motivation in conducting the analysis was an effort to clarify an issue that has been the source of real confusion in the industry, and one which – to some extent – has probably hindered the growth in online advertising. Here’s why I say that. R/F metrics are such a critical part of virtually all media plans that if the actual delivered R/F is off by 150% from what was planned, it’s reasonable to believe that such a campaign won’t work as planned. Improving the accuracy of the metrics used will be to the ultimate benefit of the Internet advertising industry and we hope our study provides precisely that benefit.

Advertisers and their agencies need to feel confident about the metrics upon which they base their advertising spending decisions. By demonstrating the unreliability of cookie-based measurements alone we hope we have helped establish the important role that third party panels play in providing the accurate R/F metrics needed to help maximize the role of the Internet in multi-media advertising plans.