Matt Cutts: Organic Algo Does Not Use Any Chrome Data

[bhartzer]

Last week at the Search Engine Strategies conference held in San Francisco, Matt Cutts did a keynote address. After the keynote, there was some time to personally discuss a few issues out in the hallway.

Here on WebmasterWorld there have been some discussions regarding the data that Google is collecting via their Chrome browser. There have been some rumors and even a recent presentation from a former Googler that said that clicks on links from Google Chrome were more powerful than other links--because Google could measure that those clicks. (I can't find the exact URLs of those threads right now though.)

I personally asked Matt Cutts about the use of Google Chrome data in the the Google organic algorithm.

We can now put this all to rest. Matt told me, in person, that Google's organic algorithm does not use any Google Chrome data. The same goes for the Google Toobar, as well.

[Robert Charlton]

How does this relate to personalized results and search history (assuming, ie, that you have search history enabled)? Did you touch on that at all?

I ask because in order to use Chrome, really, and to save any settings, you need to be signed into a Google account. At that point, it's likely that many people will not turn off their search histories.

[claaarky]

Does that mean that user metrics are not part of the algo, or just that they're not getting the user metrics from chrome and the toolbar?

I've seen urls get into googles index for my site that only I knew about (dynamic pages with modrewrite urls so not part of a server directory structure and I visited them once during testing). The only way google could know about those is from analytics (which they say they don't use) or via the ISP (which has been discussed here as a possibility).

If user metrics are not part of the algo (and I find that hard to believe) the only other explanation is they really have created some form of artificial intelligence (which I find even harder to believe!).

[tedster]

I wish we had that kind of answer. It may not be Chrome data or toolbar data, or even Google Analytics but it seems to me that Google accesses user engagement from somewhere. Free wi-fi? Buying click stream data from a big ISP or two? Who knows.

I think my opinion on the issue has been distorted a bit by several conferences where Matt shared the stage with Bing's Duane Forrester. Duane has given some extensive comments about they're user-data signals that sound like browser data to me (did the visitor scroll, hover on a link, bookmark, etc? During those sessions, Matt did not chime in for Google's POV.

[aristotle]

user-data signals that sound like browser data to me (did the visitor scroll, hover on a link, bookmark, etc?

It hard for me to imagine that Google wouldn't use such a gold mine of statistical information about users' interest in a web page. If Google really wants to improve the quality of their search results over the long run, then surely this is the way to go.

[scooterdude]

Because we are certain that he would know the algo inside out, and he would certainly tell his local friendly SEO society if he knew,

it being G's well known policy to keep SEO's fully appraised on how the Algo works, kool

[mrguy]

Maybe as far as Matt knows, but with all the different areas in the Plex you can bet they are doing something with the data.

[bloard]

Along these lines, one of my sites has ranked really well on Google in a very crowded space for a long time. It got there by 5 year old SEO techniques. But it has always had great user metrics, eg bounce rate, time on site etc.

There really is no basis in such a crowded space for it also to rank at the top in Bing. I kind of viewed this site as an "I got lucky google loves it out of 10 million results". But Bing continues to show it equal love over the past few years, and I'm convinced that it ranks well on Bing simply due to the traffic and metrics from users who find it through google.

Could be wrong, but Bing results match Google results far too much for pure coincidence from two totally separate algos. If I were Bing or another search engine that wanted to emulate Google, I'd rely heavily on user data.

[minnapple]

"I personally asked Matt Cutts about the use of Google Chrome data in the the Google organic algorithm."

But, does google show google analytics users that data? And how much if any of the data is used in Adwords calcuations?

[edited by: minnapple at 1:39 am (utc) on Aug 24, 2012]

[Robert Charlton]

Duane has given some extensive comments about they're user-data signals that sound like browser data to me (did the visitor scroll, hover on a link, bookmark, etc? During those sessions, Matt did not chime in for Google's POV.

I've assumed user metrics are at play, at least wrt the algo. I don't think they have much to do with url discovery (see footnote below about this)....

Regarding the user engagement signals picked up for the algo, though, take a look at this thread in April 2011, where Brett posted thoughts about possible data sources. He does include Google's toolbar and browser in his list...

Panda Metric : Google Usage of User Engagement Metrics
http://www.webmasterworld.com/google/4302140.htm [webmasterworld.com]

Many of Google's tests that can be readily observed involve user behavior in serps, and don't necessarily involve the specific browser itself, which is what I think we're discussing here.

Both Google and Bing have search histories which can be disabled. These "search histories" are likely treated differently from the browser histories of their respective company browsers.

Bing's official "Search History" is cookie-driven, browser specific, and can be disabled... but it's on by default unless you turn it off for that browser. They're not tracking an account... they're tracking a Bing searcher on a browser.

Google is IMO using both cookie based data as well as account-based data... and if you want to, say, set preferences or disable "Web History" on Google, you in fact need to stay signed-in to your account to do that. That has given me what appear to be the most neutral results I've seen.

I'm guessing that if you stayed signed in and did not disable Web History, the personalized search data that Google collected might (in the aggregate if signals were strong enough over time) also eventually affect everybody's results... say as in the case of searchers returning to the serps and showing signs of satisfaction or dissatisfaction in subsequent behavior. The point of my earlier question was that this is not browser-specific... it's account-specific... so Chrome itself wouldn't be the tool collecting the data.

Just guessing here... it sounds from what Matt said to bhartzer that Google would not be using Chrome's browser history. That's great, as it's a drag to constantly flush these things. I don't know, though, exactly what Duane has said about whether Bing might be using IE's browser history.

I feel that Google in general is so unclear in its privacy documentation about what's going on that they're probably creating much more paranoia than is really necessary. It looks like Google is beginning to redo some of this material, but these help files are generally self-serving and vague.

In light of Matt's comments, I feel there's more reason to worry about the privacy implications of Double Click cookies than about Chrome... though I really don't know whether these are used in the organic algo.


Regarding url discovery, see the latest discussion about this here...

How has Google found a page with no links to it?
http://www.webmasterworld.com/google/4477784.htm [webmasterworld.com]

[indyank]

Google has been repeatedly asking people to focus on user engagement. Browsers and toolbars are ways by which they can measure UE data. This includes clicking the Back button of the browser, which microsoft have admitted to use.

What Matt might have meant by saying they aren't using any chrome (or any other browser) data is they don't directly use the browser search history or any other data stored as part of Chrome's database Sqlite.

But they might still collect all UE data through cookies and other means (irrespective of browsers) and send back the essential UE data elements they measure, to their remote db used by the organic search algorithms.

[indyank]

I feel there's more reason to worry about the privacy implications of Double Click cookies

Double click cookies might be used only for serving relevant ads while organic search might be using their own cookies.

[indyank]

and send back the essential UE data elements they measure, to their remote db used by the organic search algorithms

this remote db might also be the Google accounts database or linked with it for logged in searches, as that would make it easier for them to associate all the UE data with individuals.

[ergophobe]

Robert Charlton - it depends on whether or not you believe Google reps like Cutts. If you do...

Matt's comments from September 2, 2008.

Preventing paranoia: when does Google Chrome talk to Google.com?
[mattcutts.com...]

Google Chrome Privacy Policy
[google.com...]

Some interesting excerpts....

by default Google receives standard log information including your system’s IP address and one or more cookies.

If you use Chrome to access other Google services, such as using the search engine on the Google homepage or checking Gmail, the fact that you are using Chrome does not cause Google to receive any special or additional personally identifying information about you.

If you navigate to a URL that does not exist, Chrome may send the URL

If you use Chrome’s location feature... Chrome will send local network information to Google Location Services to get an estimated location... The local network information may include... information about the wifi routers closest to you, cell IDs of the cell towers closest to you, the strength of your wifi or cell signal, and the IP address that is currently assigned to your device. We use the information to process the location request and to operate, support, and improve the overall quality of Chrome and Google Location Services. The collected information described above will be anonymized and aggregated before being used by Google to develop new features or products and services, or to improve the overall quality of any of Google’s other products and services.

(I read that one to mean "we can do anything we want with that info, but it will be anonymized").

Information that Google receives when you use Chrome is processed in order to operate and improve Chrome and other Google services.

I'm going to assume that one has be honest, because
1. it's very easy to check what the browser is sending to Google.
2. the legal liability there is substantial if they lie.

So we "know" what Google captures and receives (not your Chrome history unless you are synchronizing), but what they don't say is which services they are improving. So Matt's comment that started the thread is saying that the information may or may not be used to improve organic search in some way (as allowed by the privacy policy), but is not being used directly in the algo.



What Lifehacker has to say
What data does Chrome send to Google about Me
[lifehacker.com...]

[edited by: ergophobe at 4:14 am (utc) on Aug 24, 2012]

[indyank]

Robert Charlton - it depends on whether or not you believe Google reps like Cutts. If you do...

I guess Robert Charlton is also saying that Google organic search algo. might not be using chrome data but data collected thro. "both cookie based data as well as account-based data" and stored in a remote db on their servers.

[ergophobe]

Understood - I was thinking mostly of his browser history question.

Also, an interesting bit of reading is
Logging policies for Chrome Instant
[support.google.com...]

This details how they log your choices when using address bar prediction and Chrome Instant.

[claaarky]

How I wish I'd seen Brett's Panda Metric thread a year ago. Mind you, it probably wouldn't have made much sense to me back then!

I've been thinking about what Google are up to in the mobile market and I wonder if that might tell us something about whether user metrics are important (and maybe how they get them).

Mobile results on Google can differ slightly but not as much as I would expect. I don't have a mobile version of my site and consequently it is slow on the iPhone, for example, but my rankings are more or less the same, despite the fact that some competitors do have quick and easy to use mobile sites. That might suggest that Google deosn't have sufficient user metric data for mobile users yet for it to affect the results.

When I search on an iPhone I'm using Safari, I never log into my Google account and I'm using a mobile signal, occasionally wi-fi but rarely. I imagine this creates quite a barrier to obtaining on-site user metrics.

Google now have the Nexus, Chrome Mobile and Motorola. What's the main motivation behind these moves? Is it purely to push out Apple and Safari, all part of a general desire for world domination, or driven by their core business of search, and producing the best search results?

If user metrics are a key part of the algo these days, Google would be keen to get that mobile data. Especially as the share of mobile search is getting bigger and bigger. I never looked at my mobile stats until recently and I discovered 25% of my Google visitors are coming via mobile devices (mainly the Ipad). I can see that getting much bigger and bigger, and possibly there will come a time when the majority of Google searches will be made on mobile devices. If Google's algo has user metrics as its' heart, they need to become a major player in the mobile search market and quickly, which it appears they are trying to do.

[nomis5]

I agree with Claaarky's comments - mobiles and tablets are going to be more ans more popular and that's where G should be doing much more if it wants to get metrics.

The G search results for mobiles are a joke, yes, they are almost exactly the same as the normal SERPS. But G has loads more time to acquire mobile metrics because there simply isn't any competition. Bing is even more of a joke with their mobile SERPS, and Bing's main thrust is still riveted to laptop and PC searchers.

[claaarky]

Correction to my last sentence, it should read "they need to become a major player in the mobile market" (obviously, they're already pretty major in mobile search!).

[nettulf]

And think of all the data they will collect with their very cheap fiber project. The privacy options just states that

Other information from the use of Google Fiber Internet (such as URLs of websites visited or content of communications) will not be associated with the Google Account you use for Fiber...

[londrum]

they must use chrome and toolbar data for some things.
what about page speed? google said ages ago that page speed could be used in a "tie-break" situation, to move one site ahead of another in the serps.

surely the only way they can get data on page speed is from chrome and the toolbar, because not every site has analytics installed

[aristotle]

Good point, londrum
Google definitely has been using the tool bar to measure page speed, which seems to contradict Matt Cutts apparent statement that the algorithm doesn't use any tool bar data. However, they may have recently stopped measuring page speed, and if so, Cutts' statement could be correct, at least for the moment.

[jimbeetle]

which seems to contradict Matt Cutts apparent statement that the algorithm doesn't use any tool bar data

I can't find where Matt said that. Do you have a source?

[zeus]

He could also lie, thought about that, its not the first time for google, remember when they took photos of the streets, but also took passwords from computers ups, they diddent know, but had a patent for that collecting system.

[aristotle]

which seems to contradict Matt Cutts apparent statement that the algorithm doesn't use any tool bar data

I can't find where Matt said that. Do you have a source?

It's in bhartzer's original post that started this thread:

We can now put this all to rest. Matt told me, in person, that Google's organic algorithm does not use any Google Chrome data. The same goes for the Google Toobar, as well.

[seoskunk]

Page load time is the total time from the moment the user clicks on a link to your page until the time the entire page is loaded and displayed in a browser. It is collected directly from users who have installed the Google Toolbar and have enabled the optional PageRank feature.

Source [support.google.com...]

So Google does use toolbar information for that part of the algo it seems

[aristotle]

@seoskunk
Except that they may have recently stopped using it a few days before Matt's talk with bhartzer. In which case, Matt may have been correct.

[londrum]

would anyone really mind if they collected data in this way? i always assumed that they did anyway.

[Leosghost]

Maybe Matt "misspoke" ( again )..G reps and their "top brass" have a habit of doing it ;)

[edited by: Leosghost at 7:31 pm (utc) on Aug 25, 2012]

[seoskunk]

You could be right aristotle I went to site speed in WMT and it says no data