Define the target consistently
Make sure the target and the result use the same definition of revenue (for example, whether tax is included).
Then ask whether the immediate problem is too few suitable visitors or a journey that prevents existing visitors from buying.
Use segments to find questions
Compare relevant device, browser and channel groups. A weak segment can point you towards a problem worth reproducing, but a different conversion rate is not proof of a technical fault. Those groups may have different intentions and sample sizes.
The original article noticed differences between mobile devices without knowing the cause. Keep that uncertainty visible. Test the actual journey before claiming to have found the explanation.
Verify the change
Record what changed and compare suitable periods. A few days of improvement is an early observation, not proof of a lasting uplift. Check other changes, stock, promotions and the traffic mix.
The same applies to infrastructure: adding a performance service is not proof that the site became faster. Measure the experience and investigate the configuration.
See our CRO work and performance and customer-experience guide.