10 Apr 2013
If you ever get a chance to hear Hooman Beheshti speak at a conference, drop everything else and go. Hooman has a way of talking about performance that’s incredibly accessible (read: he’s really good at not making you feel bad that he’s so much smarter than you), and every time he speaks at an event he gets rave reviews that I would be jealous about if we weren’t such good friends.
Last week, Hooman spoke at Cloud Connect in Santa Clara, and I asked him if I could post his slide deck here. It’s an excellent overview of the business value of performance, mobile performance issues, measurement and tools, front-end problems and front-end optimization (FEO), the role of CDNs in performance, and some great case studies and examples.
I also wanted to call out a couple of bits of eye candy, because I know there are always a few of you out there on the prowl for new biz value graphics to use in your own presentations. It’s also really interesting to see familiar data presented in a fresh way, which can trigger new insights.
First is the oft-quoted Google research about the impact of load time slowdowns on search. You may have heard these stats before, but this simple graph really drives those numbers home.
A reduction of 0.6% may not sound like much, but when you consider that this is a result of a slowdown of less than half a second, and if you also consider the net impact across all of Google’s traffic, it’s pretty staggering.
I also really like this graphic interpretation of Bing’s well-known research, in which they segmented their traffic and served slower pages — in some cases up to 2 seconds slower — to different segments. It’s easy to see at a glance the impact on queries, clicks, revenue, and user satisfaction.
Again, you can see how little it takes to hurt metrics. Revenue dropped by more than 1% with just a 500-millisecond delay, and by more than 4% with a 2-second delay. It’s also interesting to note that while the 200-millisecond delay may not have directly affected revenue, it did have a measurable impact on user satisfaction, which is a good reminder of just how sensitive users are to even the smallest changes in load time.