October 18, 2011, Google announced it would limit logged-in users’ search query terms from being delivered to analytics reporting tools.  Historically, analytics platforms and other third party networks receive the keyword phrase used to enter a page, within the data passed along with the visitor. Since Google enabled people to log into a Google account and use encrypted search to increase their security back in May 2010, a limited number of organic referral strings have been blocked.

Google’s stated reason for expanding this to the “default” option instead of an opt-in is to further increase the privacy of logged-in users. However, paid search referral data will continue to include search phrases regardless of logged in state. Only the primary google.com searchers will be affected (when using secure https), with the plan to roll out to additional country code top level domains (such as .co.uk, or .ca) in the near future.

The Rosetta Analytics and Optimization and Search teams are preparing a more detailed analysis and will release it when sufficient data is gathered.  The following provides some initial thoughts and measurements from our team and coverage of reaction within the industry.

What are the Implications?

For dynamic web marketing campaigns
There will be a limit on the ability to optimize in real time and match content with the intent expressed in the search phrase. This will make life harder for marketers to fulfill the needs of customers online in this scenario. This data is highly useful and valuable to a marketer.

Google’s Matt Cutts has told Search Engine Land that “single digit” percentages of organic search data would be affected.  However, there is no denying that some data will be lost.  Some marketers feel this could be a slippery slope. A more “doomsday scenario” would be if Google continued to gain a greater percentage of logged-in users, and as many as 40-50% of users used encryption.

For search engine optimization
The majority of the industry concern is coming from search engine optimizers who rely heavily on keyword referral data to further optimize their campaigns.  For large organizations that enjoy huge volumes, there is likely be enough data “left over” to make very safe assumptions about the missing still-small segment.  Branded searches as a whole would probably be mirrored, and the “top 10” traffic drivers would likely have little variance.

However for smaller and more nimble SEO teams that capitalize on current trends and ever-shifting landscapes, conversion optimization could take a hit.  Extremely competitive affiliate spaces will likely lose a slight degree of confidence in sophisticated campaigns where a single keyword change can mean thousands of dollars a day, or even an hour.

For third party networks and tools
Google strikes a blow against their ability to deliver. That is, to collect data across a wide swath of the Internet and then leverage that data to understand a cookied visitor across the Internet with the intent of serving targeted advertising.  Additionally, some value will be lost within affected attribution-tracked conversions.

Adobe’s Omniture Blog has commented that their Organic Search traffic will be underreported because of this change, since SiteCatalyst only counts a “search” when there is a visible keyword.  They provide a way to measure the delta between the measured searches and the “missing ones,” as a temporary aid.  Google Analytics is handling this differently – still crediting a search, but with the keyword “(not provided)”.

Will the keyword data be available in Google Premium?
No. This change will affect all Google Products, including unpaid Google Analytics (GA) and paid GA.

Will this change affect Paid Search?
No. We’ll still be able to see paid search keywords, even if from a secure search page.

Timo Josten from Google shared,

If a user chooses to click on an ad appearing on our search results page, their browser will continue to send the relevant query over the network to enable advertisers to measure the effectiveness of their campaigns and to improve the ads and offers they present to you.

Without query information, advertisers can’t meaningfully make these decisions. We think that better ads are ultimately a better experience for users. Remember, the information passed in the referrer to the advertiser’s web server doesn’t identify you specifically.

The Impact

Initial estimates for a subset of Rosetta clients show about 1.22% of organic search traffic will be affected.

Rosetta’s Analytics team looked at a subset of their clients for October 19, 2011 (the first full day of live Encryption) to see how the change impacted the search volume that is now missing or encrypted. The percent of Organic Search traffic (422,000 visits across the sample) affected by this ranged from as low as 0.4% to as high as 3.42%.  Rosetta will continue to monitor the effect of this change.

Google is going after third party ad networks, and squashing their ability to use signed in search data as a means to target advertising. The impact on analytics and marketing is just collateral damage.

As far as privacy concerns go, why wouldn’t Google just strip off the search phrase for all natural search? Some on our analytics team tend to take a skeptical view of Google’s concerns about privacy. They still have the data, and will presumably use it for targeting their display and contextual text ads across the internet. The reality is they would like to stop others from having that data so that they have a competitive advantage to sell more advertising.

Thank you to my colleagues Ben Meck and Michael Helbling for the additional support in writing this article. I look forward to your thoughts and questions on this topic.