Research methods

Customer research is useful when the method matches the question. Moderated observation can reveal where a task breaks.¹ A small qualitative round does not establish how common the problem is.²

This module teaches that evidence boundary. Use direct observation when the question is how a person understands and completes a task. Use a survey or poll when the question requires structured stated responses from a defined population. Use behavioral data with known coverage when the decision depends on measured behavior or prevalence. Use a feasible controlled experiment when it depends on causal effect.³ The verb-to-method mapping and recommendation to combine methods are Atlas synthesis.

Module diagnostic

Write the current research question as one sentence. Underline the verb:

• Find or understand: observe likely users completing a realistic task.
• Count or compare stated responses: use a carefully sampled survey or poll.
• Measure behavior at scale: use behavioral data with known definitions and coverage.
• Estimate causal effect: use a feasible controlled experiment.

If the verb and method do not match, change the method or narrow the claim.

Assignment

Choose one storefront decision that is currently being argued from opinion. Write what the team needs to learn, which shoppers or records can answer it, and what evidence would change the decision. Select the lightest method that can answer that question.

For a usability round, prepare one realistic task, recruitment criteria, the representative storefront state, a neutral prompt, and an observation sheet. For work delegated to a researcher or agency, include those same inputs plus the commercial decision the findings will inform. Ask for raw notes or a recoverable artifact, not only a slide with conclusions.

Keep three columns in the final record: observation, interpretation, and next check. This prevents a participant comment from turning into a prevalence claim and makes the handoff usable by design, development, and ecommerce owners.

Evidence and further reading

1. [1]"Using moderated usability testing," GOV.UK Service Manual, https://www.gov.uk/service-manual/user-research/using-moderated-usability-testing. Supports observing actual or likely users completing specific tasks with neutral facilitation. Accessed August 23, 2026.
2. [2]"Beyond the five-user assumption: Benefits of increased sample sizes in usability testing," Behavior Research Methods, Instruments, & Computers, https://link.springer.com/article/10.3758/BF03195514. Supports that small qualitative samples do not establish reliable problem prevalence. Accessed August 23, 2026.
3. [3]"Controlled experiments on the web: survey and practical guide," Data Mining and Knowledge Discovery, https://link.springer.com/article/10.1007/s10618-008-0114-1. Supports controlled random assignment as a method for estimating causal treatment effects. Accessed August 23, 2026.