Google page speed inconsistencies, FF vs Online

was.deevey

So I was "optimizing" a sites speed earlier in the week, using a number of page speed checkers, common sense etc...

Initially the online version was giving me 90-91% depending what I was doing, firefox was giving me around 85%...

after a few hours of tinkering, google online page speed was consistently showing 60%, firefox is now at 90% constant the past 3 days.

Yslow is giving a grade "A" 95% BTW, so pretty sure FF is more accurate than the online checker :-S

Anyone else seen this ?

Trojan

The percentage figures are arbitrary on these types of tools, ignore the specific figures and just go with the trends when you're optimising.

Of course what's most important is how fast the page is being downloaded and rendering (in the geographic location of the site visitor). Apologies if that's stating the bleedin' obvious but I think sometimes we lose sight of the goals when we're neck deep in code.

I find performance optimisation fun but challenging because it requires you to have very specific knowledge in almost completely unrelated areas, from network and server hardware if you're acting on larger scale, to operating system, web server, CSS & JS compression, sprite creation, image compression, database setup - a hugely wide net to cast.

http://www.webpagetest.org/ is a super site for testing and discussing performance optimisation.

was.deevey

http://www.webpagetest.org/ is a super site for testing and discussing performance optimisation.

Thats what I was using to test the "real world" page speeds, seositecheckup.com, while normally good also seems to be pulling a cached well old version of the page without the optimization made!.

The client in this case is insisting on using google's pagespeed (online version) and complaining that the % score is so low. Pointing them to other tools like webpagetest, yslow and screenshotting FF's PS version has calmed them somewhat but they aren't totally convinced.

P.S. I know Pagespeed is also in beta, try telling a non-tech client that though grrrrrr, they all feckin swear google stuff is the sh1t and want to try nothing else.

I'd hardly say under 2s lead-time is huge (considering they have a pretty large CTA image on the homepage).

Just wondered if anyone had come up with the same inconsistancies between browser based / online versions of the tools.

Trojan

Run YSlow for them in Firebug in the V2 ruleset - this page gets a Grade C 75 here.

Then change the ruleset to Small Site or Blog and run again - this time this page gets a Grade B 81.

One size does not fit all when it comes to performance.