Google announced that starting October 18, 2011 a new privacy protection process for organic search analytics. Organic search queries executed by those consumers logged into Google products will no longer be available to web analytics systems (including Google Analytics, Adobe SiteCatalyst, WebTrends, Coremetrics, etc.) or other Google products (Keyword Tool, Trends, etc.). Note: this does NOT impact paid search analytics, even if a searcher is logged in to a Google property. Google’s view is that advertisers pay for this information in order to optimize their programs. Google will continue to provide total search volume on the platform, and it will continue to serve query level data for those organic search queries executed on the system that are from consumers not-logged into its web products. The purpose of this note is to discuss why they are doing this, and what search marketers should expect and do.
Why is Google implementing this? The answer is simple. Privacy. Searchers logged into Google through any of its products (Gmail, Google+, YouTube) receive search results that are customized to their search history—they are personalized. Even though the data that is passed on to web analytics products is, in most cases, impossible to reconcile to an individual searcher – this is a proactive move by Google to reduce privacy concerns (both in the US as well as the EU) by saying “no search queries that have been personalized will be available to any third party.”
What is the impact of this on search marketers? Google reported that less than 10% of search queries that are executed are done by consumers logged into Google products. We have no reason to dispute this statistic, so we recommend that advertisers assume that 5 – 10% of search queries are from personalized results. So any keyword level organic analysis reporting that advertisers use from their web analytics systems will start to show 5 – 10% lower referral volumes starting immediately. We do believe that this percentage will continue to increase over time, so this statistic will require on-going adjustments as additional benchmark statistics on personalization use come available. On a technical level, the process by which Google reports referential data is still unclear. For example, in Organic Search Insight™, we get data that is unique to the data, search engine, entry URL and search query. Google is saying they will continue to provide all of this data except the search query. We are in the process of testing how this is formatted through analysis of weblogs for our own site and will report on results when they come available.
It is our understanding that this encryption process is being rolled out globally. To understand the impact, we recommend that our customers do a benchmark for their top 25 organic search queries by analyzing daily variations over a 2 – 3 week period within each geography, as personalization percentages will be different by country. The purpose should be to see if there is a substantial fall off in daily keyword level referential volume—on Google only—for a particular geography. If the results for the post effective data (October 18, 2011) query volume are lower, this should be determined using 25 days prior and post the change date and this can be used as the best proxy for how to adjust query level volumes upward.
This will impact the ability of systems to look at paid and organic synergy as these types of analytics require query level granularity to see how individual searches interact between the paid and organic listing. Search advertisers may get more traffic from organic search than is reported. They may erroneously conclude that they need to purchase additional PPC volume for particular keywords when in fact their shortfalls on organic are the results of this encryption process. No changes should be made to process until the benchmarking process above is completed – again, by geography.
by Jeff MacGurn
