Ok maybe this is just me being cynical, but when Google said that page load times in Google would be a factor in organics I decided to make my site faster. In fact we got it down to 700ms pretty quick I thought. Then I thought hmmm that test was a server near where my server is located, wonder what would happen if I picked a server in London or San Francisco ?? results are in :
Manchester load times 723ms :
London load times 1.1s still not bad ( I just want to be around 2 seconds) :
San Francisco load times :
poo! that’s all I can say
Dave
24 Comments
-
- 2
smush.it reckons you could still save 4.64kb on your main logo
- 3
They will probably use that in their algo i.e. the if crawl = fast rank high if not so fast dampen SERPS a bit, and so on.
CDN will help.
- 4
I would have thought google would crawl from places other than just their HQ? I presumed they would have data centres all over the world.
I’m in the UK and adding caching to my site so it is essentially just serving up static html has really improved my ranking. Well, i think it was that that did it…
- 5
Hmm, doesn’t the page load time depend on many other things as well? Different browsers might give different results? What about the time of the day?
Is there a site where this can be tested properly?
Thanks
- 6
Yes ofcourse they have data centres all over the world.
- 7
This works like the “Qaulity Score” Adwords, Google takes the average download time for sites that are hosted in places close to your host.
If your site is in Spain, Google will make the average download time for all sites that are hosted in Spain
more info in my blog in Spanish: http://edgargranados.es/tiempo-de-descarga-factor-seo-ranking-de-google/16/
- 8
Interesting! What service did you use to check response time?
- 9
Anyone know what tool/website Dave used to carry out the speed tests, seems like a handy site to add to the collection
- 10
You would assume that it would base the ping on your location or something like that.
I doubt the load speed will be too important in terms of SEO though
- 11
I’m already looking into having a dedicated server on each continent. It’ll be a pain to manage, but I hope it pays off.
- 12
Right now I have a server located in Germany, which has seperate domains set up for 16 countries, all around the world. The site is ranking well in Germany and all countries around it, but the ranking decreases as the distance increases.
I think the load time already is a factor in the ranking algorithm.
- 13
It will be very tricky for Google to implement this as David’s results show the speed of download is not a constant. Perhaps they will have ceiling figure if you site is slower than that then you might be pushed down a few places.
- 14
[...] latest release certainly confirms the general feeling from SEO’s that page load will become a ranking factor in 2010. Google certainly [...]
- 15
I have to agree with TractorFan – load times already play a part in rankings.
But, Dave please tell us the tool you’re using to test load times.
- 16
Can’t work out exactly which speed test dave is using here.. Few good examples of speed test sites can be found here though
http://seogadget.co.uk/tools-to-speed-up-your-site/
Especially like http://www.webpagetest.org/
- 17
This seems to confirm why Matt Cutts suggests having a server local to where your target audience is.
I think the various cloud services out there are about to get a big boost in business.
- 18
reduce the number of HTTP requests, since round-trip time is the biggest factor in this particular instance.
move your static content (images, js) to multiple hostnames (IE6 is 2 connections per hostname, so if you have 10 images, use 5 hostnames – but there are ways to benchmark this).
use a CDN (amazon cloudfront is fast, cheap, and fairly easy to set up), so that images and javascript (which create most of your load times) are loaded from a server near to the user.
http://code.google.com/speed/page-speed/ has some great resources, also, and maybe check out webpagetest.org (provided by AOL)
- 19
Page load time not only affects the rankings but also the visitors of the site. If a website loads slowly, visitors tend to search for other sites where they can get the information they need quickly.
- 20
Actually this post came just in time after Google announced its’ public DNS. Quite a coincidence.
- 21
Same load differences I have. Time to move hosting provider for several of my domains to fit the local audience.
Better do it now than later, as this clearly will have a direct impact on your rankings - 22
Surely a couple of milliseconds here and there will not effect the ranking of a site and its more relating to sites what take 15 – 30 seconds or longer to load
- 23
[...] Page Speed Load Times in Google – David Naylor’s (mini)case study on load times depending on server location. When every second counts, should people make the shift? [...]
- 24
[...] latest release certainly confirms the general feeling from SEO’s that page load will become a ranking factor in 2010. Google certainly [...]
So if Google starts to use page load times in its algorithm but still crawls from the US, surely only US sites will get a bit more of a boost in the SERPs?
Maybe it’ll help knock out the Australian results from the UK SERPs…