Information checked: 21 July 2026.
Useful user testing does not require a laboratory or a large sample. A small business can learn a great deal by watching a few representative people attempt realistic tasks without coaching, then fixing repeated points of confusion.
In brief: Test tasks, not opinions, and record observed behaviour before discussing preferences.
Prepare a focused round
- Choose three to five high-value tasks such as finding a service, checking eligibility, booking or contacting.
- Recruit participants who are unfamiliar with the site and reasonably match the intended audience.
- Use a phone and desktop where both matter.
- Ask participants to think aloud without teaching them the interface.
- Record the path, hesitation, error and outcome.
- Prioritise issues by frequency and business impact.
Use neutral prompts
| Leading prompt | Neutral task |
|---|---|
| Can you see the booking button? | You want an appointment next Tuesday. Show me what you would do. |
| Do you like this page? | What do you think this business offers, and what would you do next? |
| Is the menu easy? | Find the returns information. |
| Would you trust us? | What evidence would you use to decide whether to contact this business? |
Combine with accessibility checks
Include keyboard-only use, zoom, text resizing and at least one participant who uses assistive technology when the service and budget allow. Automated scans can identify some issues but do not replace observation of real tasks.
How to check the result
After fixes, repeat the same tasks with new participants. Improvement means fewer errors and clearer completion, not merely positive comments.
Where businesses get caught out
Budget tests fail when colleagues are used as participants and the facilitator explains the interface as soon as someone hesitates. The session becomes a demonstration, producing compliments rather than evidence. Asking what people “like” also misses task failures.
Keep these records
- Research questions and task script
- Participant profile and consent
- Observation notes separated from interpretation
- Issue severity and decision log
- Retest results after changes
Your next action
Run one unmoderated rehearsal with a person who has never seen the site. Count every intervention the facilitator wants to make; each is evidence that the task or prompt needs improvement.
Sources and date checked
This practical guidance was checked against the following primary sources. Date checked: 21 July 2026.
Keep the decision under your control
Retain the relevant accounts, source material, supplier terms and recovery information. Recheck changing prices, interfaces and rules before acting.