Last year we experimented with a new way of measuring user experience. The “CXi”, as we called it, combined a few different objective, measurable aspects of performance and, with some weighting, rolled them into a single number. We gave it a test drive by measuring the top 64 eCommerce companies and setting up a “tournament” that compared the performance of different merchant categories.
The equation worked, but we weren’t totally satisfied. For one, there were sites that we felt, based on our own subjective experiences, didn’t score right. This informal “sniff test” only found a few such misses, but it was still more frequent than we were comfortable with. We also found that aspects of the equation were not easily repeatable, which would have made it an onerous task to replicate at larger scale for future projects.
A new year, a new index
For the reprised version, we went back to the drawing board. Same principles, but a slightly different approach. We still use multiple measures of page load speed from both mobile and desktop simulations as the basis of the score, but adjusted how they’re compiled and weighted. We also introduce a new aspect we’re calling “performance power” to reward or demerit sites.
Most importantly, the equation addresses both previous issues: its rankings match up well with our own subjective experiences (that is, it passed the sniff test) and it’s easily repeatable for any site or population of sites. We even changed the name slightly: CEXi. (Yes, internally we pronounce it exactly how you think we do).
To (re)introduce our index, we have an introductory whitepaper
The introduction to CEXi is complete with results from the initial proof-of-concept study we performed with a sample of 175 well-known retail sites. It includes a detailed breakdown of the ingredients that go into a CEXi score, and some of the data analysis (with charts) we performed to validate the score. Think of it as a teaser for some of the deeper analyses to come soon!
Read more below to find out what retailers scored highest and how metrics like page weight correlate with CEXi score (there may be some surprises).