Site speed as reported by Google, what's the distribution?

[g1smd]

In Google WebmasterTools, there's a site speed report for each of your registered sites, graphed for the previous 90 days.

It's under

Labs

>

Site Performance

Performance overview

On average, pages in your site take nnn seconds to load (updated on dd ddd, dddd). This is faster than nnn of sites. These estimates are of www accuracy (nnn data points). The chart below shows how your site's average page load time has changed over the last few months. For your reference, it also shows the 20th percentile value across all sites, separating slow and fast load times.

I'm wondering what the wider distribution of site speeds is.

I have only a few noted here:

2.1 seconds to load. This is faster than 66% of sites.
1.9 seconds to load. This is faster than 71% of sites.

1.0 seconds to load. This is faster than 90% of sites.
0.9 seconds to load. This is faster than 92% of sites.
0.8 seconds to load. This is faster than 93% of sites.

Perhaps a few members could fill more details in?

[deadsea]

14.4 seconds to load. This is slower than 98% of sites.

3.7 seconds to load. This is slower than 61% of sites

3.0 seconds to load. This is slower than 51% of sites.

2.7 seconds to load. This is faster than 54% of sites.

2.6 seconds to load. This is faster than 57% of sites.

2.5 seconds to load. This is faster than 58% of sites.

2.4 seconds to load. This is faster than 59% of sites.

2.2 seconds to load. This is faster than 64% of sites.

1.9 seconds to load. This is faster than 71% of sites.

1.5 seconds to load. This is faster than 80% of sites.

[deadsea]

You could also get more data points by authorizing a bunch of sub-domains or directories.

[g1smd]

Thanks for the input. I'll pick up a few more as it varies over the next week or two.

If anyone has a really fast site under 0.8 seconds or can fill in any detail between 4 and 14 seconds, please do.

[FranticFish]

0.3 seconds to load. This is faster than 99% of sites.
0.4 seconds to load. This is faster than 99% of sites.
0.7 seconds to load. This is faster than 95% of sites
0.7 seconds to load. This is faster than 96% of sites

[g1smd]

Sweet!

I do realise there will be some odd out of sequence figures if the report is not for some time in the last few weeks.

I guess Google has been tracking these figures for a long time. It would be interesting to see what the distribution was say 2 years ago. Is the web getting faster or slower? But that's for another time.

This thread will also be useful to make a comparison at some point in the future.

[sanjuu]

5.8 seconds to load. This is slower than 81% of sites.

[sanjuu]

1.8 seconds to load. This is faster than 73% of sites

[Andy Langton]

I also have a 2.7 seconds, but GWT tells me that's faster than 53% (rather than 54%). This is listed as "updated on Mar 27, 2012". It's possible there might be a geographic element?

[deadsea]

2.0 seconds to load. This is faster than 67% of sites.
2.9 seconds to load. This is faster than 51% of sites.

I may get some more on the slow side if I keep checking. My site is translated into many languages. The Korean site was the 14.4 reading, but it gets "low accuracy" because it doesn't have many users and the graph bounces around all over the place. The same with the Russian translation of the site.

[deadsea]

One percent differences could also be caused by rounding in the displayed numbers.

It could be
2.65 seconds is 53%
2.74 seconds is 54%
but because of the rounding, Google is showing both as 2.7 seconds.

[Andy Langton]

Good point, deadsea.

So, from the data above, the average load speed is between 2.9 and 3 seconds. Interesting!

[deadsea]

I graphed the data we have so far and uploaded a screen shot: [i.imgur.com...]

Here is the data in sorted column format:
Load Seconds Percentile
0.3 99
0.4 99
0.7 96
0.7 95
0.8 93
0.9 92
1 90
1.5 80
1.8 73
1.9 71
1.9 71
2 67
2.1 66
2.2 64
2.4 59
2.5 58
2.6 57
2.7 54
2.7 53
2.9 51
3 49
3.7 46
5.8 19
14.4 2

[g1smd]

The other reason for a difference could be if one of the reports is several months old.

It may be that several months ago, 53% of the web was faster than 2.7 seconds and now it is 54% (or vice versa).

[Chrispcritters]

8.7 - slower than 92%
4.4 - slower than 69%
4.2 - slower than 67%

[g1smd]

Thanks! I think I now have about 75% of what I need.

WebmasterWorld posters are always helpful. :)

[Sgt_Kickaxe]

Keep in mind that only 1% of your pageviews will be assigned a speed check and that tends to occur on your busiest pages.

[g1smd]

Yeah. I see quite a spread in the results. For a site that today is reported as 1.8 seconds overall, there are actually 9 entries from 1.2 to 2.4 seconds.

[deadsea]

3.6 seconds to load. This is slower than 60% of sites.

[EmptyRoom]

Does this REALLY affect the SERPs?

[deadsea]

Yes, it has a dramatic effect on the SERPs.

Google says they explicitly penalizes the slowest small percentage of websites.

Users also react very favorably to faster sites. User satisfaction metrics such as bounce back rates seem to factor heavily into Google's algorithm. There have been many reports here of webmasters getting dramatic rank improvements when improving site performance. Especially by getting load speed under 3 seconds.

The tool that Google uses is somewhat questionable though. It appears to measure time to the onload event for users with the Google toolbar installed. Many pages are very usable even before the onload event, so working with this metric may or may not improve user experience (and therefore rankings) depending on your situation.

[EmptyRoom]

Is there a valid way of measuring site speed? A tool that gives correct data?

[tedster]

Yes, there are several such tools. We normally don't discuss specific tools, but I'm willing to make the exception to our guidelines for this thread.

I'd say the most accurate way is to measure what your actual users are experiencing. There is a bit of freely available JavaScript called boomerang.js that will do the job extremely well across your entire site, for all user with js enabled in their browser.

In addition there are free services that use both Yahoo's Y!Slow tool and Google's PageSpeed tool. In addition, you can install either of those tools on your desktop and run them yourself - but then you're only testing site speed from your own location.

Warning - don't jump to the idea that you need to upgrade your server. In almost every case it's your pages and images that need attention most of all.

[setzer]

The Page Speed metric is pretty flawed. Google says 80% of sites are faster than us, but in reality, our site is lighting quick - faster than 80% I would say. We do all the optimizations, gzip JS + CSS, serve content from a high performance CDN, etc. The thing is, our ads can take awhile to load as they rely on third party networks. I load them asynchronously so the user experience is not impacted at all but Google's page speed algo still counts the ads against us.

[g1smd]

Warning - don't jump to the idea that you need to upgrade your server. In almost every case it's your pages and images that need attention most of all.

Also the server configuration, especially if you use URL rewriting; there's a lot of very inefficient code out there. The structure of your database tables and the types of queries made to your database to serve your content are also important.

[robzilla]

I load them asynchronously so the user experience is not impacted at all but Google's page speed algo still counts the ads against us.

Don't worry, it's not likely to affect your rankings. This thing in Webmaster Tools is just a statistic, you're not going to get a magical raise in the SERPs if you go from faster than 80% to faster than 90%. As long as your users experience acceptable load times, you're good.

Warning - don't jump to the idea that you need to upgrade your server.

And don't buy into the idea that using certain (paid) optimization services is going to work wonders for your rankings. Many companies promote such services saying that faster sites rank better (and Google, wanting a faster web, probably doesn't mind them saying so), but I believe the reality is quite different.

And if you have the time, energy and interest, it's generally more beneficial to optimize your pages manually rather than using such an automatic, one-size-fits-all service.

In short, while I love the fact that site speed is getting more attention, I still believe such statements to be inaccurate:

Yes, it has a dramatic effect on the SERPs.

[Habtom]

Yes, it has a dramatic effect on the SERPs.

Faster pages make for a better user experience, and as for effects on SERPs - unless it is slower than >95% I wouldn't worry.

[levo]

I load them asynchronously so the user experience is not impacted at all but Google's page speed algo still counts the ads against us.

There is a misconception about asynchronous loading, it doesn't block loading, but if it creates an object (for example an iframe for ad/button) it delays the onload event until those 3rd party resources are loaded too.

Google is using 'loadEventEnd' from 'Navigation Timing' [dvcs.w3.org...]

[Trueman]

This is what Google says about my site

On average, pages in your site take 37.3 seconds to load (updated on abc). This is slower than 100% of sites.

This stat is ridiculous. Not valid for my website. I have no idea how they calculate that rate. My pages are downloaded and rendered within 0.5-5 seconds, all data included.

Diagnostics->Crawl stats:

Time spent downloading a page (in milliseconds): 282 (High)

[aakk9999]

0.1 seconds to load. This is faster than 99% of sites.
2.6 seconds to load. This is faster than 56% of sites.
3.8 seconds to load. This is slower than 63% of sites.
4.5 seconds to load. This is slower than 70% of sites.
5.2 seconds to load. This is slower than 77% of sites.