I’ve been in quite some discussions lately about web performance. As a product manager it’s my concern what people would think about our web solutions. As of the beginning of a long period of testing and during some demonstrations I had the impression that our web performance was below average.
When you talk about this to technical people (funny I used to be a techy as well, not so long ago) they start waving with figures and showing you everything is fine and within the agreed KPI of 8 seconds. And those 8 seconds are a well known standard…
8 seconds or not, I’m still having the impressing things are going slow so I started some research on the internet to find out who said that a website that loads within 8 seconds is fast. I bumped into an interesting website that unfortunately isn’t updated very frequent anymore but still contains valuable information: WebPerformanceMatters.com/.
Just as I thought: those 8 seconds are not a standard, but are a threshold. During studies in the 90ies they noticed that a little more than 50% of all visitors were still attached to the website if the load time was less then 8,5 seconds. Entertaining them with animated gifs or cursors could even manage them to wait a little longer. But we are talking about the 90ies here. That time you were a geek if you had a 128K modem using a ISDN land line. The customers’ expectation was not as high as today.
Today we are all using broadband internet connection, PC’s and servers with a lot more power. And the database technology also evolved to deliver more results in less milliseconds. Opening a website and having to wait for 8 seconds is just not acceptable anymore. Will 4 seconds be? Yes and no. It doesn’t matter how fast things are. If the perception is that it’s slow, the performance is bad. If you can manage to let your customer think it’s reacting fast you’re ok.
Also you’ll have to make a distinction between your website as a product or a support tool. If you are running a webshop or a newssite and your competitor is having a product that is 50% more performant, you’ll lose money. If you give your internal customers a support tool to check their own usage reports e.g. they’ll wait anyway as they really want to get their data. However, you can’t hostage your customers everlasting as this will impact your customer’s satisfaction quota.
Bottom line: you can’t prove performance with a figure. It’s a perception. If your customer isn’t happy, you have a problem. How do you know if your customer is happy: you’ll have to ask, but more important: you’ll have to listen!