The CRO loop
(1) Measure: instrument funnel drop-off to find where visitors leak. (2) Investigate: heatmaps, session recordings, user interviews to understand why. (3) Hypothesize: form a specific testable claim about the cause. (4) Test: A/B test the proposed fix. (5) Decide: ship the winner, document the loss. Repeat.
Where most CRO programs go wrong
Skipping step 2. Teams jump from "conversion is low" to "let's test a new headline" without ever asking why visitors aren't converting. Test results bounce around, no learnings compound, and the team concludes "CRO doesn't work." It does — but only if the test follows a real hypothesis grounded in real data.
Where to start
If you don't have a CRO practice yet, start with one page (your highest-traffic landing page) and one tool (a free heatmap or session recorder). Diagnose, test one thing, document, repeat. Run a CRO audit to find the leakiest part of your funnel before testing.