WebPagetest Forums
Times versus user experience. - Printable Version

+- WebPagetest Forums (https://www.webpagetest.org/forums)
+-- Forum: Web Performance (/forumdisplay.php?fid=3)
+--- Forum: Discuss Test Results (/forumdisplay.php?fid=4)
+--- Thread: Times versus user experience. (/showthread.php?tid=511)



Times versus user experience. - JoegoTaak - 01-27-2011 12:12 AM

I've got a question about the reported load times, start rendering time, document completed time and fully loaded time related to user experience.

Our norm for "page load time" is 4 seconds meaning that a user should see the page within 4 seconds.

A lot of our sites are reported as much slower in web page test. Here is an example: http://www.webpagetest.org/result/110124_88_1DBB/

When we actually load these pages on our computers the pages seem to be really fast (try it yourself).

So I am assuming that reported times in webpagetest are not accurate with regard to "user percieve load time". Am I right?

Any comments?


RE: Times versus user experience. - pmeenan - 01-27-2011 12:45 AM

What kind of connectivity and which browser are you using on your desktop? Looking at the waterfall, IE8, Firefox and Chrome should all be SIGNIFICANTLY faster than IE7 because of all of the css and js files referenced in the head.

IE7 can only request 2 items at a time from a given domain (and with the base page taking a while, only one other request is being served at a time). Additionally, loading of javascript is a blocking operation for IE7. IE8 bumps the number of parallel requests to 6 so I fully expect it to be close to 3x as fast in your use case (Chrome and Firefox are at least 6, and in some cases even higher).

One of the cool things with WebPagetest is the video recording so you can see what the actual user perceived performance was like. I re-ran the tests with video capture.

Here is the "filmstrip view": http://www.webpagetest.org/video/compare.php?tests=110126_QP_1QEZ-r:5-c:0-l:First+View,110126_QP_1QEZ-r:4-c:1-l:Repeat+View

And a side-by-side video: http://www.webpagetest.org/video/view.php?id=110126_1b00402edba76fbe58c5f8bd103a9e52a02a9caa

Looks like the bulk of the content popped to the screen at around 4 seconds with the "new" tags filling in on the menu at around 5.5 seconds.


RE: Times versus user experience. - pmeenan - 01-27-2011 01:05 AM

FWIW, 10 external css files and 13 external javascript files is pretty extreme. If you merged them together into one of each the page would load at least 2 seconds faster, even in IE7. Combine that with image spriting and you could get the load times under 2 seconds pretty easily.


RE: Times versus user experience. - JoegoTaak - 01-27-2011 02:00 AM

Thanks Patrick. But I think your missing the point here. I use IE7 loading the site at my office PC as well as at home. A lot of customers are using IE7 as well and there are even quite a lot of them using IE6. They are experiencing http://www.xperthr.co.uk as very fast. When I hand clock load time at the office or at home it loads in 1,5 seconds. Even with a cold cache.

So I was wondering whether the absolute numbers webpagetest is reporting are accurate.


RE: Times versus user experience. - pmeenan - 01-27-2011 04:37 AM

What are your ping times from your home and office connections to the server? Not missing the point, the numbers are accurate from where they are being measured on the system they are being measured on. The video capture is literally what was displayed in the browser when it was loading.

It looks like the ping times from the test machine to your server was ~70ms and looking at the waterfall that's likely the limiting faster. If you have faster ping times you'll get faster page loads. You can tune the default connectivity settings if you want to measure with lower latency than the default profile (in the advanced settings).


RE: Times versus user experience. - JoegoTaak - 01-28-2011 04:09 AM

Ping times from my location is much faster indeed. It's ~27 ms. So that's probably the reason why we experience a faster loadtime in real life. Thanks for explaining.