How to Compare Conversion Rate Optimization Firms

Agencies with a verified published price

How to compare conversion optimisation firms

1. Ask how they decide what to test first. A strong answer describes research: session recordings, funnel analysis, customer interviews, support ticket themes and usability testing, producing a prioritised list with a stated hypothesis per item. A weak answer names a list of best practices, button colours and urgency banners, that could apply to any site on the internet.
2. Ask about statistical practice, and listen carefully. Ask what sample size they require, whether they fix test duration in advance, how they handle peeking at interim results, and what they do about weekly traffic cycles. A firm that stops tests when a result looks good is generating noise. This single question separates the disciplined from the decorative faster than any case study.
3. Ask to see tests that lost. Roughly speaking, most tests in a mature programme do not win, and a firm claiming otherwise is either running very few tests or reporting selectively. Ask what they learned from a losing test and what they changed as a result. The answer tells you whether they treat testing as research or as a source of screenshots.
4. Confirm traffic requirements before you sign. Split testing needs enough conversions per variant to reach a conclusion in a sensible window. If your funnel does not produce them, an honest firm will propose qualitative research and sequential redesign instead of tests it cannot read. Ask each candidate to say plainly whether your volume supports the programme they are quoting.

What the engagement actually delivers

A conversion programme has three phases and firms differ in which they emphasise. Research comes first: analytics review, funnel drop-off analysis, session replay, form analytics, user testing and customer interviews, producing a documented list of friction points with evidence attached. Experimentation follows: a prioritised backlog, a hypothesis and success metric per test, implementation, and a fixed run duration agreed before launch. Implementation closes the loop, because a winning variant is worth nothing until it is built properly into the production site rather than left running through a testing tool's overlay.

Ask which of the three your team can already do. Companies with strong designers and developers often need only research and analysis, and should not pay for build hours they do not use. Companies without engineering capacity need a firm that ships, and should confirm that development is inside the fee rather than a separate invoice that arrives after each win.

Pricing models and what they distort

Monthly retainers are the norm and are usually tiered by the number of tests or research cycles per month. Compare tiers by what a cycle contains rather than by the count, since a firm promising many tests a month on a site without the traffic to read them is selling activity. Project pricing suits a single audit or a redesign of one critical funnel and is a sensible way to trial a firm before committing to a retainer.

Performance-based pricing sounds attractive and is worth examining closely. Tying a fee to measured lift requires agreeing a baseline, an attribution window and a measurement method in advance, and disputes almost always come from those definitions rather than from the work. If a firm offers it, ask to see the measurement clause in writing before you accept the model, and check who has the final say when the two of you read the same number differently.

Questions people actually ask

How much traffic do we need for conversion testing?

Enough conversions per variant, not enough visitors, is the real threshold, and low-converting funnels need far more traffic than people expect. If a test would take several months to reach a conclusion, the programme should be qualitative research and considered redesign instead. A firm that will not tell you this is selling you unreadable tests.

Is a conversion audit worth buying on its own?

Often yes, especially as a way to evaluate a firm before a retainer. A good audit delivers a prioritised, evidenced list of problems your own team could act on. Agree upfront that the deliverable is yours to implement with anyone, so the audit is not effectively a sales document for a longer contract.

Who should implement the winning variants?

Whoever can put them in the production codebase properly. Leaving winners running through a testing tool's client-side overlay slows pages and accumulates fragile changes. Agree at the start whether the firm ships to production, hands off to your developers, or produces specifications, and price the engagement accordingly.

What should the monthly report contain?

Every test run, including losers, with the hypothesis, the pre-agreed duration and sample, the result and the decision taken. Cumulative measured effect on revenue or qualified enquiries. A backlog for next month with reasoning. A report showing only wins is a selection of results rather than an account of the work.

Browse agencies by specialty

• SEO agencies with verified facts
• PPC agencies with verified facts
• Social media agencies with verified facts
• Content marketing agencies with verified facts
• Email marketing agencies with verified facts
• Performance marketing agencies with verified facts