When does Google Site Speed consider a page to be loaded?

[Uber_SEO]

I'm having some discussions with our developers at the moment with regards to improving page load times to help improve the site from a usability point of view, and also for SEO.

The dev team are asking at what point Google is defining a page load, as they wish to place some Javascript after the page load event (you'll have to excuse me - this is reaching the limit of my knowledge).

So if I've got this correct, they want the page to continue to do some funky Javascript once the progress bar has completed.

So here's the question - when Google analyses page load data for SEO reasons, at what point is a page load determined. Is it when server side actions are complete, is it when the browser hits page load status, or is it once everything is complete (i.e. after the funky Javascript has finished)?

[Sally Stitts]

There is always a substantial difference reported between the Google speed tool, and other tools, such as Alexa.

I believe that the "funky javascript" is the reason. There has to be SOME reason, and I think you have found it.

The funny thing is that Google reports much slower times, based upon THEIR OWN code - AdSense, AdSense for search, and Google Analytics.

We are told that our load times are slow, but the reason appears to be Google. And "Too many domain lookups" - let's see, one for the page, and 3 for Google. Which Google recommends we REMOVE, to speed up load time! I don't think that all of the departments at Google talk to each other. Obviously, Google does not want us to remove AdSense code, AdSense for Search code, or Google Analytics code.

Also, I can't count the number of times I have looked down at my status bar, only to see "Waiting for ads.doubleclick.whatever".

[edited by: Sally_Stitts at 10:26 pm (utc) on Sep 27, 2010]

[Future]

Sally Stitts answers the question well.
Goog reports much slower times due to there own code Adsense, adsense for search and/or analytics.

to check the page load for your site, you can consider adding the javascript code when browser hits page load status or end of </html> code, make sure you test the same with and without any adsense, adsense for search, analytics and/or any other 3rd party script on our page.

I will also wait for some more replies to this topic.

[tedster]

I know I read a very specific blog post about when Google technnically considers the page to be "loaded" and what events can occur after that - but I've tried a good bit and I just can't find it now.

I'm pretty sure it's when the onload event fires - and that further JS methods that are fired BY the onload event are not counted. There are pages that wait as much as 30 seconds to pop up a poll or something like that, and I'm pretty sure those don't get counted as part of the rendering time.

However, most if not all adservers are called before the onload event, so those times definitely drag you down.

I'm going to keep looking for that official reference, however - it will mean a lot to have that on hand.

[Robert Charlton]

However, most if not all adservers are called before the onload event, so those times definitely drag you down.

To get into a major-player specific here, not to out them but to understand what's going on... many New York Times pages seem to load their text quickly, but then there's a deadly pause... and by pause I mean that you might want to go and make yourself a cup of coffee... while the ads load. Rich media and Flash ads are the worst.

While they load, your computer is useless. The ads completely tie you up. You can't even read the page, because you can't scroll it. This, IMO, is what Google should be going after. But by this measure, many major online sites... not just the NYT... would be thrown out of the index (of course, that will never happen). But, is Google not seeing this? That's where the major disruptions come.

Candidly, I don't care much about the extra fraction of a second that many are shaving off their load times by streamlining code. What really hurts are back end times on dynamic pages, ad servers, and very occasionally a 2-Gb jpeg that someone has loaded onto one of their pages.

[dstiles]

Tedster - can I assume by onload you expect there to be an actual javascript event?

If that is so then none of my sites will show as loaded because there ain't no such event. Added to which, I visit almost all sites with javascript off anyway (not that I use any reporting toolbar but I'm sure there are those who foolishly do and also run NoScript).

If this is your meaning then could that be why so many people have been saying their site load time as reported by google is rubbish?

[tedster]

My understanding is that a browser fires an onload event for every object (including the body), whether or not there is a script running that is listening for that event. In other words, the browser itself always notes the fact that the object is loaded.

The greatest majority of site visitors will be generating information that Google can use - and they're comparing "apples to apples" across site visitors at any rate, so I don't see this as being a source of bad data.

[tangor]

Isn't site speed just one of 200 variables for G? I would (and do) consider that 1/200th of issues.

More telling is why the js junk in the first place? Me and a few other 'illionth users are NoScript or JavaOFF. That's what I'd ask the dev folks: Why have this?

[dstiles]

Thanks for the elucidation, Tedster. I had a nasty suspicion that might be the answer but if it's true there should be no "penalty".

[Uber_SEO]

thanks Tedster - your help is much appreciated as always

[Robert Charlton]

I'd love to get some thoughts on the adservers. The load times are so slow that you'd think the discussion would be centering on them, but I'm not seeing much attention drawn to how bad they are. Is that ad load time factored into Google's Site Speed?

[sublime1]

Data from several of my sites of which one has some rather significant performance issues:

1) typical page load times as reported by Page Speed Activity record all of the events (regardless of whether before or after onReady has fired).

2) PageSpeed times correlate very closely to page load times as reported by Google Webmaster Tools > Labs > Site Performance, which one might think is consistent with Google's view (just a guess :-) This report sometimes points out issues with specific pages and timings for them -- it is these times I have cross-checked with PageSpeed.

3) These poor-performing pages include a number of references to external resources, including Google Analytics, Verisign, and several others. They are being counted.

Tom

[enigma1]

Put aside the js, I was thinking the load times maybe an average subject to a target location a site is after. So if a site targets a worldwide audience (eg: .com) I would expect google to check the load times from various continents, while if it's country specific to check the speed domestically. It's of course an assumption.

[renzo008]

Of course Google is filtering out Adsense, Search etc etc.

I'm almost sure they also filter out Analytics code. They just measure it by the Google toolbar.

They can even make sophisticated systems with analytics and filter it out later while calculating.

Just be sure your page render as fast as possible.

[Uber_SEO]

@enigma1 Site speed as viewed by Google is based on aggregate toolbar date for visitors that hit your website, so Google will view your performance based on the countries where your visitors come from - i.e. if you're audience is 100% French located in France, then your site speed will be determined by that audience. Unfortunately I don't think any website audience I've ever seen would be as strictly defined as that - you'll always get visitors from other countries