Define the conversion and denominator
Orders divided by sessions can be a useful store measure, but it does not explain what changed. Device, traffic source and repeat customers affect the picture. Keep the definition consistent before comparing periods or different tools.
- Name the conversion event
- Choose a consistent denominator
- Separate devices and channels
- Record the measurement period
Form a hypothesis from observed friction
A useful hypothesis links a problem to a change you can test. For example, shoppers may need delivery information before selecting a product. That is more specific than saying a page needs to look more modern.
- Record the observed obstacle
- Describe the proposed change
- Choose a relevant journey metric
- Check possible negative effects
Match the method to the data
A small store may lack enough traffic for a credible split test. In that case, usability checks and cautious before-and-after measurement can still be useful. Report uncertainty and avoid treating every change in sales as proof of the design.
- Assess available traffic
- Allow for promotions and seasonality
- Keep changes bounded
- Report the evidence honestly
A common question
What does CRO stand for?
CRO stands for conversion rate optimisation. It combines evidence, a clear hypothesis, a change and measurement rather than promising uplift from a redesign.