The Impact of “Not Provided” Keyword Data on SEO

Excerpts from my latest article at Resource’s weThink blog: “Empty Analytics: How “Not Provided” Keyword Data Affects SEO.”

Search engine optimization is based on content relevance and authority—how closely does the content match the search query’s context and intent, and how many other sites across the Internet consider it to be valuable enough to link to? The growing “not provided” bucket of organic search keyword referral data in a site’s web analytics reports, therefore, can have a profound impact on our ability to effectively optimize our sites to drive organic search visits and conversions.

The “not provided” issue was originally predicted to impact less than 10% of all Google searches. Today, the impact varies between 20% and 50% of all organic search keyword data.

Among Resource’s clients, sites have lost visibility into between 21% and 39% of their organic search keywords. There’s no detectable pattern based on the site’s size, brand strength, industry or ecommerce ability on Which to predict the size of the keyword data loss. A large ecommerce site faces the same data challenge in this instance as a small consumer brand site.

The problem stems from Google’s secure search. Stating privacy concerns, Google defaults to secure search for every user logged into a Google account. That includes Gmail, Google Calendar, YouTube, and any of the other Google products. When signed into Google, the searcher’s keyword referral string is stripped from the information passed to analytics packages in the referring URL, rendering the keyword data “not provided.”

In addition, most major browsers have defaulted to Google’s secure search. Most recently the problem has been exacerbated by mobile devices using Android 4 and iOS 6, which have also defaulted to using Google’s secure search. With so many of the major software players funneling their search queries through Google’s secure search, you can see how the “not provided” issue has increased in scale from the original prediction.

What does it matter? Can’t we just ignore the “not provided” data? Read on to find out.

Read the article in full at Resource’s weThink blog »


Web PieRat logo.

Originally posted on Web PieRat.