I’ve been watching the frontend optimization video’s on Yahoo.
They have some startling statistics and case studies which I felt were worth sharing,
The first is from Steve Souders talk 18 months ago:
On average 10% of user experience load time is due to backend page serving, this means that 90% of the load time experienced by users is controlled by the frontend. So for example if you optimize your backend by 50% at most you are optimizing the overall user experience by 5%, however if you optimize the frontend by 50% this would result in a 45% overall saving in user experience load time.
This then goes on to be even more pertinent when you consider the information from the Nicole Sullivan’s talk:
Amazon did two tests where they artificially increased the load times of their pages, it gave the following results:
100ms – 1% drop in sales
400ms – 5-9% full page traffic drop off
Google had similar results:
500ms – 20% fewer searches
I think this highlights just how important load times and by proxy frontend sitedevs are, when you consider that in a big build you will have an approximate ratio of say 1 sitedev to 10 backend devs, it really doesn’t make sense and we should be looking to sell in as many sitedevs as possible into projects for optimization purposes as this is what will really increase the load times and hence drive profit for the client. These are some powerful statistics, that I will be adopting in the future to help back up my case, but this in turn raises some interesting questions. Like how do we measure performance increase to show metrics of increase justifying the cost and how should this be implemented -should an extra optimization effort occur post build, how do we bake front-end performance into continuous integration environments? How do we replicate what back end devs have achieved which is to get clients to pay for the extra work?
somebody pointed out to me that it might be nice to have some links to relevant source material:
Some other helpful links courtesy of Per Swantesson: