Some colleagues were taking the measure of a new flash site built for one of our customers. The analytics numbers seemed to show people were spending more time on it–and since it’s one of their goals to have people spend more time on the site, this seemed a good thing.
Until they noticed that the new site was also taking quite a bit longer to load. The way measurement was set up, there was no way to tell whether the longer visits were a result of longer wait-time or not.
We’ve built a tool called Dynamic Response that specializes in measuring time-to-serve; we’re probably going to customize it so it can deliver a measure that equals “time spent on visit minus time spent waiting for page to load”.
Since visit length is such an important KPI for so many companies, I thought it would be of interest that: a) visit length is not always what it seems and b) there is a way to tease out the truth with the right kind of measurement tools and approaches.
