Conversion Rate Optimization Template

Start where the absolute volume loss is greatest, which is almost always the top of the funnel. A 5pp improvement in signup completion at 3,000 visitors/month yields 150 more signups. A 5pp improvement in trial-to-paid at 100 trialists yields 5 more customers. Fix the leaky bucket where the most water is escaping.

At minimum, two full business cycles (14 days for most B2B products). Never make a decision before reaching your pre-calculated sample size. The [Product Analytics Handbook](/analytics-guide) covers statistical significance in detail. If you cannot reach significance in 4 weeks, either your traffic is too low for this test or the effect size is too small to matter.

At minimum: product analytics (Amplitude, Mixpanel, or PostHog) for funnel analysis, a session recording tool (Hotjar, FullStory) for qualitative evidence, and an A/B testing framework (LaunchDarkly, Statsig, or even a simple feature flag). You do not need a dedicated CRO platform to start.

Combine quantitative data (funnel drop-offs, cohort comparisons) with qualitative data (session recordings, exit surveys, support tickets). The best hypotheses explain both the what (users drop off at step 3) and the why (because the form asks for information they do not have yet). Avoid hypotheses that are just opinions disguised as experiments.

CRO works best as a cross-functional effort. PMs own the backlog and prioritization. Designers generate and test UX hypotheses. Engineers instrument tracking and build variants. Analytics validates results. If forced to pick one owner, the PM is the right default because CRO is fundamentally a prioritization and measurement discipline. ---