Google: 'Not Provided' Keyword 40% Of Referring Traffic, Marketers May Rejigger Metrics

• Comment (9)
• Recommend (12)
• Tweet
• 

The term "not provided" found in marketing analysis reports now accounts for nearly 40% of referring traffic data from organic search, up 171% since Google introduced the policy one year ago, according to a study released Tuesday.

Some believe it will spur the use of ad retargeting and social media.

Some 64% of companies analyzed in the study see between 30% and 50% of their traffic from Google as "(not provided)," and 81% see more than 30%, according to Optify's report, Google (Not Provided) On the Rise.

The report details the impact of Google's SSL enhancements on SEO referral data collected by publishers. The study suggests that recognized referring keywords from organic search declined by 49%.

Last year, Google gave signed-in search engine users an option to encrypt their search queries using Secure Socket Layer. The process encrypts Google search queries, meaning that data about visits from organic search queries no longer provides information, such as referring keywords, on each individual query. Rather than the exact word, Google began passing publishers the term "not provided" as the referring keyword for visits from search engine queries and clicks.

Google's intent was to provide security and privacy, but it has had an alternative outcome. Doug Wheeler, CMO at Optify, said "our No.1 keyword term is 'not provided.'"

The landing site still receives information that a site visitor came from Google, but the encrypted search terms are excluded from the referrer headers as part of the request sent to the result site visited. Wheeler believes the percentage of "not provided" data will grow — and it will force brands to do more retargeting.

Marketers may need to develop other metrics to confirm performance. In the long term, it means marketers will not have the option of measuring performance of SEO efforts by connecting a search term with Web site metrics like traffic, conversion rate, leads, engagement, and revenue.

The Optify report also suggests marketers will no longer have the ability to use referrer data to customize and personalize the experience for site visitors. For example, don't bother using referring keywords in lead nurturing rules, or score visitors and leads based on their referring keyword.

Optify analyzed the organic traffic to 424 business-to-business (B2B) Web sites between Nov. 1, 2011 and Oct. 1, 2012, tracking 17,143,603 visits from organic search -- an average of 1,428,634 visits per month. During that time, the company captured 7,241,093 referring keywords.

• Comment (9)
• Recommend (12)
• Tweet
• 

9 comments on "Google: 'Not Provided' Keyword 40% Of Referring Traffic, Marketers May Rejigger Metrics".

1. Michael Hubbard from Media Two Interactive
   

   commented on: November 13, 2012 at 9:32 a.m.
   Laurie - it's an interesting element Google has added to those of us in love with search... It's essentially said "quit focusing your efforts on organic results". By doing this, it protects Google and their vaunted algorithm, but it also steers people to paid search (AdWords) where the keyword data remains visible. Smart business move on their part - but very frustrating for people like us that are trying to focus our content on what relevant users are coming to our site for. The argument could be made that we're supposed to know - but the reality is, everyone searches in a different way, and our sites have been built around individual trends. That all being said - you mentioned "Last year, Google gave signed-in search engine users an option to encrypt their search queries using Secure Socket Layer." I consider myself pretty savvy in the space, and I don't recall that memo coming to me (but I'm sure it's buried in my ever changing privacy policies I sign off on). Just funny to see :). Thanks for the article.
2. Troy Johnson from AALBC.com, LLC
   

   commented on: November 13, 2012 at 10:39 a.m.
   Thanks for the insight. (not provided) is typically the keyword for 50% of the top 10 landing pages. Google's own "Landing Pages with Keywords report" rough 40% of the visits and 50% of the pages visits. You y'all provide information in a much more consumable fashion about Google's services than Google themselves. I guess I could have learned what this was all about by trolling through Google's discussion boards, but your emails seem to be right on time. Thanks.
3. Erik Ford from BoostCTR
   

   commented on: November 13, 2012 at 2:30 p.m.
   There's a further discussion/follow up on how this issue is addressed by professionals. Great SEM campaign builders can pass through keywords in URLs manually using the utm tag or even in the page's code tied to a passing in the URL as well. This dynamically optimizes landing pages for stronger QS provides greater granularity for segmentation in analytics.
4. Fred Jorgensen from Crosby Marketing Communications
   

   commented on: November 13, 2012 at 2:37 p.m.
   Not sure it's that big of a deal. Because the SSL issue might be considered a "relative bias", keywords 2-11 should still generally represent what's hidden by "(not provided)".
5. Michael Hubbard from Media Two Interactive
   

   commented on: November 13, 2012 at 2:40 p.m.
   Erik - not sure I'm following, and just want to clarify. When you say "SEM Campaign Builders", are you referencing PPC/AdWords - or are you referencing building pages for organic search? If it's organic, please keep the conversation going - I'd love to hear your thoughts. If it's paid search, then this particular Not Provided issue isn't relevant, as Google still relays the data when it passes through the paid side of things.
6. Michael Hubbard from Media Two Interactive
   

   commented on: November 13, 2012 at 2:49 p.m.
   Fred - I disagree and actually think it's a bigger issue for marketers. Although you're correct that you could make some relative assumptions to deduce what's (not provided), but it's not exact data and therefore you just don't know for sure - which is the entire point of Analytics. But to give you an example where we see it as a problem... We have a client who is seeing an overall 10% organic search lift for their site YOY - which is great for them. We've been also running brand awareness campaigns, yet YOY data is showing that branded organic searches are off by close to 50%. So we have no idea if the brand awareness for the client is generally way down or if it's because all of those people are now searching securely. Again, we can make some assumptions, but the frustration comes when the last couple years you've been benchmarking things, and now your data is no longer relevant because outside variables have changed. Not the end of the world - just frustrating is all.
7. David Skinner from [x+1]
   

   commented on: November 13, 2012 at 3:48 p.m.
   This is a problem for Landing Page Optimization and real-time targeting. I can't believe though that its sole cause is 40% of people having 'opted out' of this for their google searches. It sounds way too high to me unless it was a default setting. I wonder if there is another driver for it.
8. Michael Hubbard from Media Two Interactive
   

   commented on: November 13, 2012 at 4 p.m.
   You're correct David - Google actually changed the settings last year so that it's everyone's Default setting. Laurie pointed out the below link to me where it shows this: http://support.google.com/websearch/bin/answer.py?hl=en&answer=173733
9. Erik Ford from BoostCTR
   

   commented on: November 13, 2012 at 5:48 p.m.
   @ Michael- Yes to PPC. You're right in terms of passing through keywords in paid advertising, though the coordination with keyword sets and dynamically optimizing keywords on page would require a pass through in the URL. On the organic side, I've seen two routes: cookie extraction or creating two versions of a page, where on the organic initiatives, you set your URL writes to always append the Google ?utm= tracking to pass through keywords manually into Google Analytics. Neither are easy feats, but if you have a decent dev team it shouldn't be too hard to execute on.

Leave a Comment