Yes. Performance-based CRO pricing ties part of the agency fee to the lift its work produces, usually as a revenue share, a share of incremental profit or a success fee per winning test. It works when every change is measured against a control group that runs at the same time and when the baseline is written down before the work starts. Without those two things, nobody can say which part of a revenue change the agency caused, and the deal turns into an argument about attribution.
What does performance-based pricing mean for a CRO agency?
In a classic retainer, a store pays a fixed monthly fee for research, test design, build and analysis, whatever the tests show. In a pay for performance model, some or all of the fee depends on the measured result. The agency shares the risk, and it earns more when its changes add revenue.
Most performance deals are hybrids: a lower fixed fee that covers the work, plus a variable part tied to results. A fully variable fee is rare, because research and failed tests cost the same whether a winner follows or not.
Which performance pricing models are common?
- Revenue share: a percentage of the extra revenue a winning change produces, for an agreed number of months after it goes live.
- Share of incremental profit: the same idea on profit instead of revenue, which protects the store when a change lifts sales of low-margin products.
- Success fee or pay for uplift: a fixed amount for each test that beats the control by an agreed margin at the agreed confidence level.
- Hybrid retainer: a fixed base for the work plus one of the variable parts above.
Each model needs the same three definitions in writing before the first test: which metric counts, how the lift is calculated and for how long the variable part runs.
Why does performance pricing work for CRO but rarely for SEO or ads?
Because an A/B test has a control group. Half the visitors see the current page and half see the change, in the same weeks, from the same campaigns, during the same season. The difference between the two groups is the effect of the change, measured rather than estimated. Ron Kohavi, Diane Tang and Ya Xu describe this as the core strength of online controlled experiments in Trustworthy Online Controlled Experiments.
Search rankings, ad campaigns and brand work have no such control. When revenue rises, it could be the work, a competitor dropping out, a price change or the season. That is why a fee tied to their results is hard to settle fairly.
What makes a performance deal go wrong?
Disputes are almost never about the work. They are about the baseline. If a store raises ad spend, runs a sale, goes out of stock, changes prices or ships its own site changes while the deal runs, a before-and-after revenue comparison mixes all of it with the agency’s changes.
- Pay only on lift measured inside a controlled test, never on total store revenue.
- Write the baseline, the metric and the calculation into the agreement before work starts.
- Agree how long a winner counts, because the effect of a change can fade.
- Agree what happens when the store changes something that touches a running test.
Which stores can use performance-based pricing?
Stores with enough traffic for tests to reach a clear answer in a few weeks. On low-traffic pages, most changes are based on research and usability principles rather than tests, and there is no control group to measure them against. The guide When is an A/B test the wrong tool? explains where that line sits.
For the parts of the work that cannot be tested, a fixed fee is the honest model. A good agreement often uses both: a fixed fee for research and untestable changes, and a variable part for tested winners.
How does Volition set its fee?
The fee is agreed after the first call, once the store’s traffic and data are known. Where a store has the traffic for controlled tests, part of the fee can be tied to the lift measured in those tests, with the baseline, the metric and the calculation written down first. Where it does not, Volition works on a fixed fee. Book a call to talk it through.