When comparing CRO agency proposals, you want to know what work your shop will receive for the budget. Looking at the first specific engagement helps: which question will the agency investigate, what will it hand over and who will implement a change? Those details let you establish whether the prices cover the same scope at all.
If your shop lacks reliable measurement, the provider needs to explain how they will handle that. An ongoing testing programme is a different engagement from investigating a problematic shopping journey.
Ask for a defined first engagement
A short description of that first task makes the proposals easier to assess. Knowing the pages involved, the open question, required access and deliverable helps you judge the work and its value for your team. “We optimise your conversion” alone leaves those details unanswered.
A possible first engagement: investigate variant selection and the subsequent cart, interpret existing exit data and deliver specific changes tied to their locations.
Your planning also depends on what happens after the findings. An annotated report supports different work from a design, an implemented flow or an evaluated experiment. If the proposal only includes analysis, your team or another partner will need time and budget for the remaining tasks.
Compare the same components
A common table makes it visible whether the providers offer the same work. Marking an unexplained component as “unresolved” prevents it from looking like an included service. Those gaps then show you what to ask about before comparing prices.
| Component | Question for the provider |
|---|---|
| Investigation | Which journey and data fall within the scope? |
| Design | Which change will be specified clearly enough to build? |
| Implementation | Who develops, checks and publishes the variation? |
| Evaluation | Which outcome and decision rule will be agreed in advance? |
| Operation | Who monitors faults and can roll back a change? |
| Handover | Which documents and results will we keep? |
If implementation is left open, ask for a separate quote. Otherwise you are comparing the price of an investigation with the price of a change built into the shop.
Useful questions can be uncomfortable
“Would you advise us against an A/B test?” is worth asking. Have the provider explain which prerequisites might be missing and what investigation would help with the decision instead. NN/g explains how UX research supports more informed test variations. A test of arbitrary designs only answers a question about those designs.
References also need context. What did the provider contribute? What was studied, and what was the comparison? A reported uplift alone does not explain the work or its relevance to your shop. You do not need confidential customer data; ask for an intelligible account of the approach.
Before you commission the work
The proposal should name the first task, the handover and the person responsible for the next decision. Resolve omissions before signing an ongoing contract. If nobody internally can develop the changes after the analysis, include an implementation quote in this decision.
For a journey investigation, compare the scope of our shop diagnosis. Our CRO service page sets out the subsequent design and implementation options.
