How to Test Customer Reaction Before Changing Price or Product Terms

How to Test Customer Reaction Before Changing a Product, Service or Pricing Model

If you are preparing to reprice, restructure terms, withdraw a feature, or migrate customers to a new proposition, the question is not whether to test reaction. It is how to test in a way that produces evidence a board, a risk committee and the FCA would all accept as a sound basis for the decision. Done well, customer reaction testing sharpens pricing judgement, surfaces fair value concerns early, and gives you a defensible record of why you proceeded. Done badly, it confirms what you already believed and leaves you exposed when complaints arrive.

Key Executive Takeaways

• Reaction testing should be designed to disprove your proposed change, not validate it. If the test cannot produce a result that would stop the change, it is not a test.
• Combine stated-preference research with behavioural evidence from live pilots or analogues. Neither alone is reliable for pricing or term changes.
• Treat vulnerable customer segments as a separate analytical cohort from the start, not a sensitivity check at the end.

Start with the decision, not the research method

Before any fieldwork, write down the specific decision the test must inform and the thresholds that would change it. For a price increase, that might be: expected attrition, movement in net promoter among retained customers, complaint volume per thousand, and the proportion of customers in vulnerable circumstances showing material detriment. If you cannot articulate the numbers that would cause you to pause, delay, or redesign, the research will drift toward confirmation.

This is also the point to involve risk, compliance and the Consumer Duty champion. Their input shapes what counts as a meaningful adverse signal, particularly on fair value and consumer understanding.

Choose methods that match the risk of being wrong

For low-reversibility changes (repricing an in-force book, withdrawing a feature customers rely on, migrating closed-book customers), you need converging evidence from at least three sources:

• Quantitative survey work with sufficient sample size to read vulnerable cohorts separately. Use conjoint or discrete choice for pricing to avoid the known weakness of direct willingness-to-pay questions.
• Qualitative depth interviews with customers who would be most affected, including those likely to lose out. Avoid relying on your most engaged customers, who are rarely representative.
• Behavioural evidence: a controlled pilot on a sub-segment, an A/B test on digital journeys, or analogue data from a comparable change you or a peer firm has made.

For a reversible digital change, lighter-weight testing is defensible. For contractual or pricing changes affecting existing customers, it is not.

Design the test to find the people who lose out

The most common failure in reaction testing is averaging. A change that is neutral or positive on aggregate can inflict real harm on identifiable sub-groups: long-tenure customers on legacy rates, customers with low financial resilience, customers who chose the product precisely for the feature you are removing. Segment the analysis before you run it, not after. Pre-specify which cohorts you will read separately and what detriment thresholds would trigger a redesign.

This matters commercially as much as it does for Consumer Duty. Concentrated harm in a small cohort produces disproportionate complaint volumes, remediation costs and reputational damage.

Interpret signals honestly

Stated intention to switch overstates actual switching, often by a factor of two or three. Satisfaction scores understate quiet dissatisfaction among customers who feel locked in. Pilot results flatter the main rollout because pilots usually get better communication and support. Build these known biases into how you weight the evidence, and say so explicitly in the board paper.

Where the signals conflict, the behavioural evidence from pilots or analogues should generally outweigh the stated-preference work.

Connect findings to the governance decision

The output should not be a research report. It should be a decision paper that states: the proposed change, the evidence of likely customer reaction by segment, the identified pockets of detriment, the mitigations (grandfathering, phased implementation, enhanced support, opt-outs), and the residual risk the board is being asked to accept. Where the evidence shows material harm to an identifiable group, the paper should set out what has been changed in the proposal as a result. If nothing has changed, explain why.

Next step

Before the next pricing or proposition committee, write the one-page test design: the decision, the thresholds that would stop it, the segments you will read separately, and the methods. If you cannot complete that page, you are not ready to test.

Frequently Asked Questions

How long should reaction testing take?

For a material pricing or terms change, assume six to twelve weeks from design to board-ready findings, including a pilot where feasible. Compressing below that usually means skipping the behavioural evidence or the vulnerable customer analysis, both of which are the parts that matter most.

Can we rely on existing complaints and NPS data instead?

No. Those tell you about the current product. They do not tell you how customers will react to a specific proposed change. They are useful context, not a substitute for forward-looking testing.

What if the test shows the change is unviable?

That is the test working. The cost of a research programme that stops a bad change is trivial compared to the remediation, complaints and regulatory engagement that follow a change customers reject. Build the governance route for a negative finding before you run the test, so the result cannot be quietly reinterpreted.

How do we test reaction for closed-book or legacy customers?

With particular care. These cohorts are often older, less digitally engaged, and more likely to include customers in vulnerable circumstances. Use outbound telephone research and face-to-face interviews rather than online panels, and over-sample the longest-tenure customers.

Who should own the test internally?

The business owner proposing the change should sponsor it, but the design and interpretation should sit with a function that has no stake in the outcome, typically customer insight or a dedicated research function, with risk and the Consumer Duty champion as standing reviewers.