Google: A Searcher Sees A Dozen Or More Google Tests Per Day

Dec 5, 2012 • 8:49 am | comments (9) by twitter Google+ | Filed Under Google Search Engine Optimization

Google Labs LogoIn a Google Tech Talk on October 22, 2012, Google's Chief Economist Hal Varian said that Google runs a lot of experiments. So many that a typical Google searcher/user will see a dozen or more experiments when they access Google.

Greg Linden caught this remark and posted it on his Google+ page. He said around 26 minutes into the video Hal said:

Any time you access Google, you probably are in a dozen or more experiments.

Here is the video:

The talk was about how to handle all that data and with dozens of experiments seen and acted on by each Google user several times per day - that is a lot of data.

For a webmaster and SEO - it can be a challenge managing all those variables as well.

Forum discussion at Google+.

Previous story: Google: We Are Revamping Webmaster Tools Notifications



12/05/2012 02:30 pm

As a user I'd like to know which experiments are those.


12/05/2012 03:15 pm

Some are obvious (anyone using a Youtube playlist has surely noticed the periodic change in location/look of the playlist/page or when they were testing the top-down nav in web search) but I would venture that the vast majority are algo tweaks they are testing out and you would probably never even notice them. Oh, and you will not get to know which experiments are being run, except the obvious ones, because they are not required to tell you, it would be a really bad business decision to share that information, and there are so damn many of them.


12/05/2012 03:18 pm

Almost all tests are to increase Adword revenue. Google is going for short term revenue maximation and forgetting everything else. Even tools like Danny Sullivan and Barry Schwartz are a lot less enthusiastic about Google, their patron saint. The cat is out of the bag, webmasters talk openly about results being manipulated and soon everyone will know it.


12/05/2012 03:31 pm

Oh, this guy must have some contacts! Please sir, tell us what the tests that don't fall into the 'almost all' bucket are doing? You have already explained that 'almost all' tests are accomplishing Adword revenue, so detailing what the 'nearly none' for shouldn't be difficult.


12/05/2012 03:39 pm

Yeah I have contacts Mr Google Employee: "Google paid clicks increase 28% year over year during Q3" "Google Inc.’s paid clicks during the second quarter ended June 30 increased 42% compared with the same period last year" "Google's 2Q earnings rise as clicks on ads soar" "Google Reports: Revenues Up 24% Clicks Up 39%"


12/05/2012 05:01 pm

I remember an article from Wired (maybe in 2010 ?) where the same statement came out about experiments on searches. Now, from my understanding, it's not like these tests come out of the blue. There is a careful process to pull out changes, but they also want as many concepts out there as possible. There is a huge gap between scientific experiments and industrial deployment. Most of the time, it's all about the sanity check, in order to be safe from things spinning out of control.


12/05/2012 05:53 pm

@gfyd And? How's Bing working out fer ya?

Steven Lockey

12/06/2012 04:21 pm

You should actually try reading what Danny and Barry write for a change. They critise and support Google when they deserve it in a unbiased manner. You should try been unbiased some time. You might even like it.

Colin Guidi

12/10/2012 12:13 am

Sure. Let's just completely ignore the fact that desktop search, mobile search, and tablet search, have ALL vastly grown. Ergo, AdWords usage will undoubtedly increase as well as the revenue it brings Google. I agree that there are a lot of tests which likely play favor towards AdWords usage, however Google also places a large importance upon its organic landscape, which it also constantly tests. You simply cannot make the biased statement above, that all tests are meant to improve AdWords usage/revenue.

blog comments powered by Disqus