CRO pricing is less standardised than development pricing, which makes it easier to overpay without noticing. Here is what the models are, roughly what each costs, and which arrangements put the risk in the wrong place.
The four models
| One-off auditAnalysis, prioritised findings, no implementation | $3,000 - $15,000 |
| Audit plus implementationThe findings actually shipped and measured | $8,000 - $30,000 |
| Monthly retainerContinuous testing programme | $4,000 - $20,000/mo |
| Outcome-basedYou pay on a result, not on time | Varies |
CRO work with us starts at $5K. A full audit runs $3K to $15K depending on scope.
A one-off audit is the cheapest way to find out whether there is anything worth doing. The risk is that a document is not a result — plenty of audits are read once and filed.
Audit plus implementation is the honest default for most brands. Findings are worthless until they ship, and the people who found the problem are usually fastest at fixing it.
A retainer makes sense once you have enough traffic to test continuously. Below that you are paying monthly for tests that cannot settle.
Outcome-based is rare because it requires the agency to carry risk. Ours is the single-page version: we rebuild one product page, run it as a live A/B test, and invoice only if it wins.
What decides the number
Traffic. Not because more traffic means more work, but because it decides whether testing is possible at all. Under roughly 500 sessions a month on the page in question, no test will settle and you should be buying structural fixes rather than a testing programme.
Scope. One product page is a different job from a full funnel across desktop and mobile.
Whether implementation is included. The single largest swing in any CRO quote.
Whether they run the tests. Analysis is cheap. Shipping variants, instrumenting them and reading the result is the work.
What you should refuse to pay for
A monthly retainer on a store without the traffic to conclude a test. You will receive activity reports for a year and no evidence. Ask directly: how many conversions a month does the page get, and how long would a test take to settle?
A list of best practices. If the deliverable would be identical for another store in your category, you are buying a template. A real audit is specific to your data.
Percentage-of-uplift deals without a control. If the measurement is "revenue versus last month", you will pay for seasonality, promotions and traffic mix. Uplift-based pricing is only meaningful against a properly controlled test.
Recommendations with no implementation path. "Improve trust signals" is not an instruction.
What good looks like
One primary metric agreed before anything ships. An even split against your current page. Reporting from a dashboard you can open yourself. And a plain statement when a test has not settled — roughly half of all tests fail to beat control, and an agency whose case studies are all wins is showing you a selection rather than a record.
We publish one of our own tests that proved nothing for exactly that reason, and we wrote about how to read these numbers in why most CRO case studies prove nothing.
Talk it through
Send us the page you are least happy with.
One URL and a sentence on what you think is wrong. You will get a real read within a business day, including whether your traffic can support a test at all.
Get a straight readA real read within one business day — not a sales call.
The cheapest first move
If budget is tight, buy a single-page rebuild rather than a programme. It costs a fraction, it tells you whether the agency can actually move a number, and it gives you a result you can act on instead of a document you will not.
About the author
Manpreet Singh
Manpreet Singh is the founder of Proscube, an ecommerce growth studio. He leads the studio's Shopify and Shopify Plus engineering, headless builds, CRO, and its work on AI engine optimization, and writes its guidance on how to grow a DTC brand without wasting money. He works directly with founders — no account-manager layers between you and the people doing the work — and would rather tell a client not to build something than sell them work they don't need.
More about Manpreet
