A conversion rate optimisation agency sells a compounding claim: the traffic you already pay for can produce more revenue if the pages are tested into shape. The claim is true and the industry around it is uneven, split between research-led programs that move revenue and button-color theater that moves dashboards. The difference is visible before signing: how the agency researches, how honestly it treats statistics, and whether its case studies survive three follow-up questions.
Research is the product; tests are the delivery
Strong CRO starts with evidence about why visitors do not convert: session recordings and heatmaps read with judgement, user testing on the actual funnel, form analytics, survey and support-ticket mining, and a prioritized hypothesis list that names the expected mechanism for each test. Ask any candidate to walk the research that produced a past win and the hypothesis it tested. An agency that leads with a testing tool and a template library is selling activity; the research narrative is what separates a program from a subscription.
Statistical honesty is the character test
Testing math punishes impatience: small samples produce fluke winners, peeking at results inflates false positives, and a program reporting weekly wins on modest traffic is describing noise. The honest agency tells you upfront whether your traffic supports meaningful testing at all, sizes tests to reach significance, reports losers and inconclusives alongside winners, and measures lift in revenue terms against a holdback where feasible. Ask how many of their tests lose; the credible answer is most of them, said without embarrassment, because losing tests are the cost of the winners being real.
What a real engagement delivers, month by month
The visible cadence: research findings documented, hypotheses prioritized by expected impact and effort, tests designed and shipped with clean implementation, results reported with the statistics shown, and winning variants actually implemented in the site rather than living forever in the testing tool's overlay. The compounding asset is the learning library: documented results that make every subsequent test smarter and survive the agency's departure. Contract for that documentation explicitly, keep the testing and analytics accounts in your name, and judge the program quarterly on revenue lift against the fee, which is a comparison an honest agency proposes itself.
Questions people ask about conversion rate optimisation agency
How much traffic do we need for CRO to work?
Enough conversions monthly for tests to reach significance in weeks, not quarters. Below that, buy research-led fixes (the agency applies findings without split-testing each) and honest agencies will say so.
What does a conversion rate optimisation agency cost?
Retainers scale with research depth and testing velocity. Judge against measured revenue lift, and be wary of pricing tied to reported win rates, which rewards exactly the statistical dishonesty you are screening out.
CRO agency or in-house experimentation team?
Agencies suit getting a rigorous program started and transferring the method; mature high-traffic businesses usually internalize it. Contract for documentation and training so the capability stays when the retainer ends.