|Google: Your Webmaster Tools "Site Performance" Tool is Broken|
| 7:24 pm on Jan 11, 2011 (gmt 0)|
Over the past year, I moved a vbulletin site from an entry level dedicated server with no page speed optimizations in use, to an enterprise dedicated server with quad core processors and SSD drives, page speed optimizations, and use of a CDN, and Google's "Site Performance" tool shows no significant difference in the crappy page speed it says I have (slower than 70% of other websites)!
It still says my pages on average take 4.8 seconds to load, when most 3rd party tools I use average between 2-3 seconds, and browsing through the site - it is obviously faster than most sites I visit.
This is despite the fact that 3rd party page speed tools show a dramatic improvement in load time, and visitors to the site have told me it is faster as well.
Since one of the main reasons I started working on page speed was because it became an SEO ranking factor, I wonder if this broken "Site Performance" guage is what Google is using to determine that? If so, maybe they should fix it or use something else?
Does anyone have experience where this site performance tool actually seems to recognize changes you've made to your pages' load time?
I know a few other webmasters who have seen the same thing I am seeing.
| 9:07 pm on Jan 11, 2011 (gmt 0)|
A lot of the factors that impact Site Performance are about page rendering, not server response time.
In my experience, things like how many http calls need to be made by the browser (scripts, styles images) how well optimized the images are, and the use of gzip compression are extremely critical compared to a few hundred ms improvement in server response time.
Google may report this to you in one interface, but they have more than one "tool" here. Probably the most important source of information about your pages comes from real visitors who have the toolbar installed. It phones home and tells tales.
I'm not saying that the Site Performance numbers are not mysterious at times - but I am saying we need to look at the whole performance picture, and not only server factors.
| 9:52 pm on Jan 11, 2011 (gmt 0)|
@seoart i agree with tedster that it is more how the pages are rendered..download the tool from WMT that functions in firebug and analyzes page speed....here everything you can do on your pages to speed it up will get broken down for you...once you have done all and achieved as close to 100 score as you can there is then very little you can do.
I am taking it seriously too this point and then after that i ignore what is in WMT as it fluctuates too much and besides i know i have done pretty much all i can do.
| 4:57 pm on Jan 12, 2011 (gmt 0)|
Ads and specifically Adsense ads can really kill pageload time as reported by webmaster tools pageload. Also Google Analytics has a significant negative impact.
Firefox is an excellent resource (Firebug, pageload), but try using an ad blocker and your pageload times will drop dramatically for your own browser.
If you use an ad blocker, enable PageRank in the Google Toolbar, and surf your own site it is pretty easy to significantly influence the Webmaster Tools performance measure. Webmaster Tools "pageload" purposely filters out indications of slowness generated by Adsense Ads and Google Analytics. Only the Firefox Pageload tool shows whats really going on!
It comes down to how many visitors to your site are actually using the Google Toolbar with PageRank enabled, probably not many. You may be the one providing Google with data that says your site is slow!
I wouldn't be surprised if soon Google starts using its page preview functionality to measure total pageload performance. If they do I hope they speed up Adsense at the same time.
| 7:39 pm on Jan 13, 2011 (gmt 0)|
>> If you use an ad blocker, enable PageRank in the Google Toolbar, and surf your own site it is pretty easy to significantly influence the Webmaster Tools performance measure.
I agree with bumpski here! I've been away from home over the past 5 weeks and surfing my own site on some very crappy Internet connections (read: unsecured W-LAN from across the street). This in its self has caused my site performance to really take a dive last time I checked.
| 8:40 pm on Jan 13, 2011 (gmt 0)|
I don't know, my experience has been similar to that of seoArt. I made some major improvements a few months back, such as image optimization, that dramatically decreased my total payload. I also improved rendering time, number of requests, caching, and various other factors. These improvements had a noticeable impact on both the performance I personally observe and on the performance reported by various tools.
However, there was little impact on my WMT performance data, which is claimed to be very accurate. There was some impact, but not much. I'm skeptical. I think there's a reason why that section of WMT is in beta.
| 4:58 pm on Jan 14, 2011 (gmt 0)|
It is straight forward to look at what Google is measuring using Firefox, with the Firebug AddOn. Using Firebug alone look at the "Net" tab analysis and then scrolling to the bottom of the results will show you the time it takes for the "onload" event to fire. This event may occur long after your page has appeared to render. This is the time Google is using to measure your page's performance.
For example if you use Google Analytics the "onload" event will not fire until Google Analytics is done recording it's data. Visible rendering of your webpage might appear to be complete in say 2 seconds, BUT, Google Analytics may not complete for another 4 or 5 seconds, THEN, the onload event "might" fire for a total time of 6 or 7 seconds. I don't know how many times and how many sites I've seen the status bar say "Waiting for Google Analytics" (or something like that).
What it comes down to is probably one single factor is slowing your pageload drastically and you haven't found that factor yet. In many, many cases it's Adsense ads!
| 5:56 pm on Jan 14, 2011 (gmt 0)|
Stuff like Analytics and other background threads should not count in the page speed results. If the page is rendered for the user and they can interact with it then it's done IMO regardless if the stupid AdSense ads aren't finished loading or GA has recorded data.
| 5:59 pm on Jan 14, 2011 (gmt 0)|
erm , is this widget still being updated in gwmt ?
| 7:58 pm on Jan 14, 2011 (gmt 0)|
ddog, that's the way I feel about it too. If what's slowing down Google's perception of my load time is the time it takes to render Google's stuff (and yes, I do use adsense and analytics), then that's their problem. My pages are usable long before ads and analytics render, so for the user it doesn't matter and Google can fix their own speed problems if they care about them.
| 8:31 pm on Jan 14, 2011 (gmt 0)|
remove adsense and analytics for a big pagespeed boost, they are by far the slowest rendering portions of most sites right now.
| 9:03 pm on Jan 14, 2011 (gmt 0)|
Unfortunately Google search has declared they are now using this "page load" time, to a limited extent, to rank your pages in the search results. So if you use Adsense your pages will load more slowly. SO, your rankings will go down, SO your earnings will go down. Excessive corporate interdepartmental competition can destroy corporations (I've been there).
If Google is like almost any other corporate organization the various departments compete, versus cooperate. Adwords is obviously the BIG DOG because they are closest to the CASH FLOW. (The closer you are to the cash flow, the more cash flows into your pocket).
I imagine Adsense is in second place being a source of cash flow due to the content network (publishers, us).
Finally there's Google Search which actually makes no money! They are considered overhead! I believe Google Search is in charge of webmaster tools.
Publishers are simply caught up in this Google corporate conundrum! Good luck to us all!
| 10:48 pm on Jan 14, 2011 (gmt 0)|
|remove adsense and analytics for a big pagespeed boost |
LOL, yeah, because having better performance in the WMT speed performance tool is so much more important than making money...