How Much Consumer Research Do You Need?

You've narrowed a campaign down to three headlines.

Do you commission a research project?

For many teams, probably not. By the time you write a brief, scope the project, agree on a methodology, set everything up and wait for the findings, the process can feel out of proportion to the decision.

So the team chooses internally. Or asks a few colleagues. Or runs a quick poll.

That is an understandable response to a practical problem. But making the research smaller does not necessarily have to mean accepting weaker evidence.

Use the smallest sound research approach that can support the decision. Reduce unnecessary scope before reducing the quality of the evidence.

Marketing is full of choices that do not look much like research projects.

Which claim should appear on the pack? Which headline should run in the campaign? Which names should make the shortlist? Should an element of the design change?

A five-word claim is a small piece of copy. It may also appear on hundreds of thousands of packages.

So the visible size of the thing being tested is not a very good way to decide how much evidence you need.

Consider the consequences instead.

How widely will the decision be applied? How uncertain are you about the answer? What happens if you get it wrong? Could you easily reverse the decision later? Do you already have evidence that helps answer the question?

A seemingly small decision may deserve careful research if it has substantial reach or is difficult to reverse. A much larger piece of work may need little new research if the uncertainty is low and good evidence already exists.

Suppose the question is: Which of these three claims most clearly communicates the benefit we want consumers to take away?

You could build a large communications study around it.

You could explore category attitudes, purchase drivers, appeal, differentiation, relevance and any number of other things that would be interesting to know.

But if those findings would not influence this decision, they may not belong in this study.

A focused comparison of the three claims against the criteria that matter may be enough.

This is one of the simplest ways to make research proportionate: remove the questions the decision does not need.

There are many parts of a research project that can be made smaller.

You can narrow the business question. Test fewer alternatives. Use an established method instead of designing something bespoke. Keep the analysis focused. Produce a shorter report.

Those changes reduce the scope of the project.

Other things are harder to compromise without weakening the evidence itself.

The participants still need to be appropriate for the question. The alternatives need to be presented fairly. The measures need to support the kind of conclusion the decision requires. Poor-quality responses still need to be identified and handled appropriately.

And a narrow question does not necessarily mean a tiny sample. The amount of data you need depends on the comparisons you want to make and the conclusions you expect the research to support.

A short study can be rigorous. A large one can be badly designed.

Proportionate research also means being realistic about what the evidence can resolve.

Imagine two claims perform almost identically.

The fact that the team needs to choose one does not mean the research has to manufacture a difference between them. Sometimes the useful finding is that the study did not establish a meaningful difference.

That can still help with the decision. It may mean the team can choose based on another consideration, such as strategy, creative fit, cost or ease of execution, without pretending consumer evidence favoured one option.

A smaller study should not be expected to provide more certainty than its design and data can support.

There is another perfectly reasonable option: don't run a study.

Perhaps you already have relevant evidence. Perhaps the uncertainty is small. Perhaps the decision is inexpensive and easy to reverse. Or perhaps no plausible result would actually change what the team plans to do.

In those cases, judgement may be enough.

Research should have a job.

The useful shift is to move away from treating the options as a substantial bespoke project or no consumer evidence at all. There is a lot of room between them.

For a consequential and uncertain decision, proportionate research may still be substantial. For a narrow question, it may be a focused test using an established approach.

Remove what the decision does not need. Keep what makes the evidence worth using.