Google Incorporates Site Speed Into PageRank Calculation

lee1 writes "Google is now taking into account how fast a page loads in calculating its PageRank. In their own words: '[W]e're including a new signal in our search ranking algorithms: site speed. Site speed reflects how quickly a website responds to web requests. ... our users place a lot of value in speed — that's why we've decided to take site speed into account in our search rankings. ... While site speed is a new signal, it doesn't carry as much weight as the relevance of a page. Currently, fewer than 1% of search queries are affected by the site speed signal in our implementation and the signal for site speed only applies for visitors searching in English on Google.com at this point.' Considering the increasing dilution of high-ranking results by endless series of plagiarizing 'blogs,' brainless forums, and outright scam sites, anything that further reduces the influence of the quality of the content is something I would rather not have. Not that Google asked me."

Comments

Asking

[Anonymous Coward]

Not that Google asked me.

Well, now they know that you're an influential Slashdot contributor I'm sure they'll sit up and take notice.

Slashdot

[SimonTheSoundMan]

So when a site gets slashdotted and blown to oblivion, Google also ranks it lower. Awesome!

Re:Slashdot

[DNS-and-BIND]

Google Bot is always indexing your site. I push 8-10Gb of traffic a month (yeah I know it's not a lot, thanks for informing me) and of that, 1Gb is Google. I don't know why Google constantly loads my pages even though they don't change that much, but Google does it.

Re:Slashdot

[Stan Vassilev]

One would think only if the Google Bot happens to be indexing your site at that exact moment; one would additionally think they'll revisit to see if it's structural or not?

If you use Google Webmaster Central you may notice that, while Google's algorithm is smart, it's also very overestimated in some areas, and involves plenty of manual tweaking by the Google employees for it to work properly.

Site Speed is not calculated solely from the times the Google bot takes to crawl the page, it's calculated from Google toolbars that have the pagerank feature enabled (that feature calls home which sites you visit, and how fast the page got loaded).

Whether Google can detect clusters of frequent accesses such as from "slashdotting" is entirely under question, since most slashdot users may not have google toolbar with pagerank on, but for the *few* users that do, the site will just appear slow in general.

Additionally, if a site targets a demographic that has worse latency (low income people, areas with dial-up and so on), then, again, that site will appear to be slower, while actually the visitors have slower internet in general.

Additionally yet, often the reason a site is slow is somewhere along the route, nowhere close to either the visitor ISP, not the site server, and it's not for all users either. So if you have bad luck or due to your content you pick up users that happen to often be routed through the bad route, you'll lose page rank.

Re:Slashdot

[loufoque]

Maybe if you correctly used Last-Modified and Etag headers with a 304 Not Modified response, you could avoid a significant part your bandwidth usage.

Re:Slashdot

[amorsen]

You can associate your site with a Google account and override their heuristic.

Re:Slashdot

[FireFury03]

If a server can't handle much load, it's probably not that important

Or it is a very informative hobbyist site with lots of useful info on it, which is comparatively slow compared to a well funded commercial site that has nothing but marketing-speak.

TFA says they are looking at "server response times", but I can't see this being at all useful unless they look at the total page load time (including all the ads that come off slow servers).

Slashdotting, power failure, tsumani, cleaning lady tripping over the network cables, poor server-side scripting, badly configured web server... What's the difference anyway?

The difference is that some of these problems are transitory and some are more permanent. You probably don't want transitory problems to affect the ranking (here's hoping they average it over several crawls).

Re:Slashdot

[metamatic]

Yes, but if only there was some open standard protocol he could use to tell Google how often to index which parts of his site, which Google implemented...

Re:Slashdot

[DNS-and-BIND]

So, a guy with a 8Gb per month website is now required to have a server with root access and his own copy of apache, which he will then tune like a whistle without even having to read the documentation? Come on, man. Shared host. Non-expert admin that doesn't even like to mess with his Drupal install now that it's working, for fear of breaking it in some subtle yet damaging way. But thanks for shitting all over a novice with that nasty tone, though.

so, spammers just need servers...

[FuckingNickName]

...close to and prioritising Google. Gotcha.

Really, am I the only one to find Google a fairly poor *find* engine? I mean, for anything which might remotely come close to sounding like it's a product, you've got Wikipedia right at the top, followed by 1000 review/comparison/pricing sites. For a tech question, you have expert-sexchange and 1000 crappy forums with responses from the downright wrong to the gratuitously abusive. I barely use Google (or any search engine much) for their generic WWW search - I'm more likely to be +site: searching a specific newsgroup/support forum/journal/enthusiast site I already know has intelligence. I don't need Google using yet another algorithm to fail at finding useful information - just employ 100 people spending 8 hours a day tagging the clone/spam/pricecheck/etc sites if you actually want to make a difference.

Re:so, spammers just need servers...

[GIL_Dude]

You hit the nail on the head with that one. I, too, find myself using queries containing "site:msdn.microsoft.com (rest of search)" (for say Windows API information) or using "-" in the searches to suppress certain results. Like you say, otherwise you get basically "a bunch of crap" - mainly from people who have no idea what they are doing. Just today I had a problem with elbyvcdshell.dll (from Slysoft's Virtual Clone Drive) causing Windows Explorer to hang for 5 minutes each time I renamed a folder. I tried searching that on Google - hell half of the hits were stupid posts of every file on a system at malware check sites, or bleepingcomputer.com, or other "is this malware" posts. Did I say half? Shoot - I just checked again and I think I meant 85%. The results for most tech searches are indeed useless unless you already know what site you want and include that information in your search. The internet is just filled with crap sites that make it into the indexes and get high relevance.

Re:so, spammers just need servers...

[rockNme2349]

Half the people I heard from said that if they scroll all the way to the bottom they can read the answers for free, and the other half say this doesn't work. This confused me for the longest time until I finally figured out the answer.

Expertsexchange allows you to scroll down to the bottom to get a free answer the first time you visit their page, then gives your browser a cookie saying that you have gotten your free answer, and won't show you any more. So if you want to ensure that you can always scroll to the bottom, you simply have to block cookies from them and you are good to go.

Re:so, spammers just need servers...

[TheLink]

I thought it checked the "http-referer" - so if you clicked via google, you'll get the answer at the bottom.

But if you copy the URL and paste it on a browser, you don't get to see the answer at the bottom.

Personally what annoys me more than expertsexchange are the journal sites. For those I don't get the answer at the bottom or anywhere, even though it shows up in the Google search results.

Used to be Google policy that a site is not allowed to show different content to Google from what it shows to users - they smacked BMW Germany down for that. But now I see lots of sites getting away with that, and no, those journal sites don't get fooled by the user agent thing.

Perhaps they pay Google to be allowed to do it.

Slowbotted

[Naatach]

Yay! I can DDoS my competitors and have Google endorse it!

Re:Where is the 'speed' measured from?

[caffeinemessiah]

Geez, will you at least RTFS?

Currently, fewer than 1% of search queries are affected by the site speed signal in our implementation and the signal for site speed only applies for visitors searching in English on Google.com at this point.

The main site serves visitors from the US. Thus, measuring speeds from multiple locations around the US is probably the best thing to do. They're presumably measuring speed from all their datacenters (their crawlers are likely to be distributed across the country (and world), so recording the average speed over multiple crawls would be a good approximation when you're dealing with the scale of Google and the Web).

Re:Sweet

[complacence]

tl;dr

That sounds reasonable.... so far

[erroneus]

Does this help do battle against spam/scam sites? Yes.

Does this help hosts of original content? Maybe... maybe not.

Does this serve as an indirect or otherwise passive-aggressive push for network neutrality? I suspect it might be.

After all, those seeking to act against Google's interests by lowering speed and throughput to and from Google would automatically get a lower rank. Think about some of the newspapers out there who can't get over their aging business model. Think about other sources of information who might also be a competitor of Google in other markets? At the moment, Google is the primary source for lots of people.

I must admit, I am having some difficulty coming up with arguments against this idea but I can't help but get a slightly uneasy feeling about this just the same.

Re:That sounds reasonable.... so far

[rockNme2349]

Does this serve as an indirect or otherwise passive-aggressive push for network neutrality? I suspect it might be.

It sounds to me like a push completely against net-neutrality. The websites that are served up faster get a higher rank. The websites that are throttled get a lower rank. Net neutrality isn't about how website owners filter their bandwidth for their visitors; they've always been free to do what they want. Net neutrality is about the ISPs and other backbone entities of the internet throttling traffic. If there was an ISP between google and two webpages it could direcly influence their ranks by throttling the site it wants knocked down and prioritizing the site it wants to give a higher rank to.

Re:Sweet

[K. S. Kyosuke]

Mod parent up! Google is catching on to something us slashdotters have known for a long time. The person who posts fastest usually has the most insightful things to say!

I have an idea: Slashdot could easily incorporate average commenting speed into its UserRank and serve pages to excessively-first poster slowly, giving chance to other, more insightful readers, such as the humble me.

Isn't Google missing the point?

[fluffernutter]

I occasionally put websites together for small businesses and it seems increasingly hard to get these kinds of websites known. Google seems to be more and more indexing websites with lots of content and now with speedier response which will completely slant their rankings towards large companies with huge resources.

For example, I did a website for a lady that sells garden and landscaping lighting local to where I am from. Her business focus is not one that needs a large web page, she just wants her catalog to display basically but she does want people to find her with Google. I've done all the things like making sure the title is accurate and headers are relevant, etc. However, it seems to me that much of it is futile. Unless she is the type of business that focuses on inviting people to add content to her site (in other words an internet/web business) the sad truth is that she will basically get ignored by Google.

Re:As a user of Google

[thetoadwarrior]

Good thing the quality of content is still king. However if you have two sites that rank at 100% and one is significantly faster then it comes up first. I can't see how that is bad thing.

Google says this affects a tiny fraction of sites and let's face it, it will be irrelevant when comparing two text only sites. But with the growing web app trend then yes speed does make a difference. If you want to use an online Office replacement, like offline software, you don't want to sit there waiting for things to happen. Online games, like offline, can succeed or fail based on their speed.

The guys at Google aren't dumb. I doubt we'll see pages punished for loading in 100 ms rather than 50ms. However if you take 2 minutes compared to 50 ms then you may be if your content isn't the best and quite frankly I'm happy with that.

What ever happened to "remove this result"

[beakerMeep]

Recently there used to be a feature to "ban" a result (like experts exchange) but they removed it in favor of only being able to "star" results you like. I'll have to say this seemed the single best feature they had ever added to search results. It was very useful to be able to identify (for myself) who was gaming the results. But apparently google thinks I'm better off with the safety of little pretty stars.