My first thought when hearing that Google are making site speed a ranking factor, was how this would affect International SEO.
Hosting infrastructure in the US is large-scale, reliable and often cheaper than local providers in developing countries. The Planet states that 42% of their customers are located in international markets. On the other hand local servers are often faster for local users and have less chance of network interruption.
Will Google’s move towards site speed as a ranking factor push international SEOs to ask for local hosting? Patricio Robles over at eConsultancy seems to agree,
The location of your host could also be a consideration. An offshore host, for instance, might be at a disadvantage here, especially if it doesn’t have good peering.
However Richard Hearne at RedCardinal makes some good points against this argument on Twitter.
RedCardinal: @nickwilsdon You’re assuming they’ll test locally, but that’s not how this will work from my understanding. (status link)
RedCardinal: @nickwilsdon I think 2 things worth looking at: Page Speed (likely they’ll use these metrics), and Adwords QS page speed. (status link)
RedCardinal: @nickwilsdon server location will probably be normalised – otherwise this would never work (status link)
I think Richard’s right and he’s backed up by the notes Google has released on how site speed is tested in regard to Adwords Quality Score.
We evaluate your load time relative to the average in your server’s geographic region. If your website is hosted on a server in India, for example, your landing page load time will be compared to the average load time in that region of India. This is true even if your website is intended for an audience in the United States.
There is still an advantage in good local hosting if you can get hold of this, especially if it out-performs the average in the region. As Matt Cutts suggests, users don’t want slow websites. However the emphasis for this new ranking factor will likely be on the page speed metric rather than loading time and the physical location of the server.
** High Speed train by lrargerich one of the artists who make their work under a Creative Commons license at Flickr – thank you!
Popularity: 6% [?]
Nick Wilsdon is the CTO of 
6 responses so far ↓
1 Gab Goldenberg // Nov 17, 2009 at 3:00 am
Excuse my ignorance – what’s the difference between Page Speed and load time?
2 Tom Coady // Nov 17, 2009 at 8:03 am
@gab it looks to me like “Page Speed” is synonymous with the firefox plugin from google that measures load time. I’m guessing that google use slightly different metrics to evaluate the site’s “speed” but the question is which given that “Page Speed” appears to measure about 20 different tests.
Their Page Speed recommendations are sound enough but not specially convenient for certain types of dynamic content and in some cases may bork otherwise sound browsers like IE6. I’ve yet to see a site that ticks all the page speed boxes so I can’t see how this can be used as an effective, even if objective, measure of good end user experience.
3 Nick Wilsdon // Nov 17, 2009 at 9:40 am
@Gab
As Tom said – sorry I didn’t explain it better.
The way I understand it, page speed will be a measurement of the time the different elements on the page take to load. They will try to factor out the time it takes to reach the page itself. Maybe they can do this with their bot, by starting the count from the time they successfully connect with the page.
Load time would be a combined measure of the the page speed + the page retrieval time. Load time would also be subjective to the location of the user who is retrieving the information, so they will use an average for the servers in that region.
The reason people are saying the hosting needs to be good is that the CPU/RAM on the server will affect the time the page elements take to load. Shared servers with a heavy load will have less CPU/RAM resources to give your site.
If this is wrong, feel free to correct me! I’d like to get this right as it looks to be an important factor in 2010. Maybe we should try and get JohnMu to comment here? :)
4 Richard Hearne // Nov 17, 2009 at 11:56 am
I think latency will be factored in, but this will most likely be normalised also.
There are still a lot of unknowns about how this will work, and another interesting issue is flash pages which are a preloader for larger flash content (loading 1%…2%…). No idea how they will score this type of thing. The use of CDNs are also another interesting area which should help load quality (multiple hosts, faster closer servers) and the pricing on CDN networks is now coming down to levels any webmaster can afford.
I have a feeling John and the Googlers will be quiet on this until it’s officially launched. But he’ll probably show up right about now to prove me wrong ;)
5 Paul Richardson // Dec 23, 2009 at 5:41 pm
I think a speed metric within Google’s indexing algorithm is a very positive thing.
It will encourage companies to think about speed and discourage them from using excessive amounts of 3rd party media such as poor performing adverts and flash based content, etc.
I’m sure the speed metric will only count as a small plus towards page ranking and content will always be the primary measure.
At http://www.getmecooking.com we are implementing features to make the site as fast as possible – much faster than other large cooking sites. See how fast the recipe page dynamically loads 20 recipes at a time. Other sites such as http://haystack.com also do this.
6 affordable web page design // Dec 31, 2009 at 12:53 am
I’d like to thank you Nick for this one, now I have better understanding about the load speed and page speed.
Also for the people who share their thought on this one.
Leave a Comment