Experiment
Conversion rate optimisation for e-commerce
The problem with most testing programmes is not the tests, it is the sample. A retailer running a button-colour test on a page with four hundred weekly sessions will spend six months learning nothing. We start by working out where you have enough traffic to learn anything at all.
What the work involves
- Analytics review and funnel instrumentation — confirming the data can support a decision
- Qualitative research: session replay, on-site surveys, support and returns data
- Prioritised hypothesis backlog with the expected value and required sample stated
- Test design, build and statistical analysis, including when to stop
- Checkout, PDP and category journey work for sites without the traffic to test
- A documented record of what was learned, including the losing tests