Lightning talk about Perceived Performance
Real user performance - when does the user think a website is slow?
My 7min speech at webtuesday explains what perceived performance is - with the progress bar example. Then talks about the social web - how fast writes should be visible to the various users (me vs. friends vs. strangers). At last show some stuff we at local.ch do to measure perceived render time.
Slide showing the social web metric
Movie from the speech
Thanks to Hugo and Corsin for recording the session.
Links mentioned and sources of the information can be found on del.icio.us/keepthebyte/perceivedperformance
My 7min speech at webtuesday explains what perceived performance is - with the progress bar example. Then talks about the social web - how fast writes should be visible to the various users (me vs. friends vs. strangers). At last show some stuff we at local.ch do to measure perceived render time.
Slide showing the social web metric
Movie from the speech
Thanks to Hugo and Corsin for recording the session.
Links mentioned and sources of the information can be found on del.icio.us/keepthebyte/perceivedperformance
I don't see how putting Jiffy.mark all over the place can help? Is there a need to consider every single user local.ch has as a guinea pig?
My 2cents, remove every class="clear" and use more image sprites. Moving scripts at the end may not work for every one, I don't know about your case.
Posted by
Yoan |
July 14, 2008 12:27 AM
Interesting post :)
Posted by
eurodipity |
August 20, 2008 4:25 AM
This is great to see people using Jiffy even if only just parts of it. I am the dev who wrote the JS library and we would love to have others contribute to the project to make it better.
As for why you would put Jiffy.mark all over the place. Well it gives you real metrics from people looking at your site. These numbers are what your users browsers are reporting. This is valuable information that can be used to improve user experience, make adjustments and make the right decisions as to what these improvements should be. You can also back those decisions up with real user metrics not just best guesses.
Posted by
Ben Maldonado |
August 28, 2008 5:51 AM
Post a Comment