Forget The Theorizing: How Does Google Instant Actually Perform?

To search marketers, new technology is like being a kid in a candy store. When Google Instant hit the SERP, there was a flurry of speculation and press. Although Google steadily rolls out minor improvements to it home page, many of the changes fly under the radar of the average consumer. Google Instant was the most dramatic consumer-facing change to the company's homepage in years.

Google Instant displays real-time search results while you type. While a jarring experience initially, you find your eyes quickly adapting to the flashing page and ultimately find what you're looking for faster.

Aside from being a fun gimmick, there is a concrete business case for Google Instant. In addition to its commitment to constantly evolving the product, Google has stated that the goal is to reduce the amount of time needed to find the results you're seeking. What's good for the user experience is good for Google -- both in market share -- and, more importantly, the faster you find what you seek, the sooner you will come back and be exposed to even more ads. Mission accomplished.



There has been a lot of chatter over whether or not these changes would affect performance. Our assumption was that the greatest impact would be on search query length, or the amount of keywords a consumer enters for a given search. The implications of a permanent change in the number of keywords used in a typical search could have a wide-ranging impact on future SEM campaigns.

As marketers we can be a skeptical bunch, and rightfully so. I's our job to pick apart new technologies and see if they help or hinder ad spend. Gut instincts told many of us that since the concept was built around delivering relevant search results faster than before, consumers would find what they were looking for in far fewer keystrokes, meaning they would be much more reliant on "head terms" of two keywords or less. Admittedly, there was a lot of skepticism around Google's true intentions; it appeared the search giant was compressing traffic into the expensive head terms, and as a result we'd see ever-increasing competition and therefore higher CPCs. Others debated this theory and hypothesized that the keyword length would increase because the technology would draw out longer searches faster than the user can type.

Without data, however, arguments don't amount to much. Armed with proven information, it was time to quell some of the speculation and state the facts. What we found was surprising: nothing changed. The keyword count post-Google Instant was virtually the same. This was an aggregated view of all campaigns across the agency. There were a few exceptions. A car manufacturer saw a 12% increase in query length, while a CPG brand saw a 7% decrease. Comparing these to similar accounts by vertical didn't yield any explanation.

With the "head term vs. tail term" argument deemed moot, we did find some data that we didn't anticipate. Overall performance improved. The metrics that most often reveal market trends -- CTR and CPC -- saw an uptick in performance. There was a 3% increase in CTR and a 6% decrease in CPC, or seven cents less on average per click. What we can take away from this is that Google Instant is, in fact, delivering more relevant ads. As a result, consumers are responding by clicking more often on sponsored search. A decrease in CPC suggests that users aren't crowding expensive real estate.

As search marketers, we should be taking a closer look at the keywords and ad groups that have fared the best post-Instant. Careful attention should be paid to the keywords that are most closely aligned with Google Suggestions. Armed with the knowledge that Google Instant has improved performance, it's critical that we leverage the gains where ever possible.


Opportunities For Search Insiders: MediaPost has a limited number of opportunities for columnists who write intelligent commentary on search trends, and/or how-to pieces on SEM and SEO. Please see our editorial guidelines here, and then send pertinent info on credentials, plus one column idea, to

3 comments about "Forget The Theorizing: How Does Google Instant Actually Perform?".
Check to receive email when comments are posted.
  1. Joe Bencharsky from iNet Entertainment, November 29, 2010 at 2:59 p.m.

    Making irrelevant or flat wrong suggestions does not speed anything. As I said in my blog post: Google Instant: Much Ado About Bupkis there is not practical use for Instant as a stand-alone function. if predictive search was based upon past search patterns or interests and relevance based upon social media interaction and searches web profile, then it might have some significant suggestions that are relevant. As it stands, just a cute feature that is designed more to distract than to assist.

  2. Kevin Lee from Didit, November 29, 2010 at 3:02 p.m.

    Of course, no one knows if Google made tweaks to it's quality score algo along with the deployment of instant. Quality score changes might manifest themselves as increases in CTR (for more relevant ads being shown) and lower CPCs (because bids need not be as high for an ad with a higher predicted CTR).
    Variance of data across advertisers makes it even more difficult for anyone other than Google to really know what is going on. I also checked with publishers who rely on SEO traffic and didnt' see a material impact. (Perhaps the comScore's and Neilsen's of the world can shed light).

  3. Nathan Safran from Conductor, November 29, 2010 at 4:02 p.m.

    At Conductor, we studied the impact of Google Instant by analyzing the distribution of search traffic for 10 major sites (2.7 million visits in total) pre and post Instant and came to similar conclusions: little changed when it came to searcher habits as a result of Instant.

    See for more information.

    -Nathan Safran
    Senior Research Analyst, Conductor

Next story loading loading..