Ask what happened in the last payroll run, and other prompts that get real answers

Ask about the last completed payroll, not an imagined ideal product. Behavioural questions expose sequence, responsibility and evidence; feature wish lists tend to collect agreeable speculation. These prompts suit employers and bureaux when the interviewer replaces generic terms with the participant's actual setting.

1. What happened in the most recent payroll run?

Ask the participant to reconstruct inputs, checks, approvals, submissions and outputs. Note who acted and which systems were open. Do not request identifiable employee records for a market-research interview.

2. Which information arrived last or needed correction?

Follow one real exception. Establish its source, detection point, consequence and repair. The answer is more valuable than a general statement that data quality is difficult.

3. How do you know HMRC received the expected report?

HMRC's payroll reporting guidance describes FPS and EPS information and timing. Ask what acknowledgement or reconciliation the participant actually retains, without turning the interview into tax advice.

4. What changes when a worker starts or leaves?

Trace the hand-offs between hiring records, payroll, pension processes, documents and access. Look for duplicate entry and unclear ownership.

5. How are workplace pension tasks completed?

The Pensions Regulator's payroll-process guidance identifies staff data and software support as practical concerns. Ask how assessment, contributions, files and rejections work with the named scheme.

6. When did you last change software or provider?

Capture the trigger, alternatives, migration work, parallel testing and final decision. Interview people who stayed put as well as those who switched.

7. Which mistake would be hardest to recover from?

Separate financial, employee, reporting, privacy and reputational consequences. Then ask what control currently reduces the risk and where evidence is stored.

8. Who can block a purchase?

Identify budget, payroll, finance, IT, security and adviser involvement. Ask what document or test each reviewer requires rather than accepting "management decides".

9. What support was needed around the last deadline?

Record channel, urgency, resolution and workaround. This helps estimate service cost as well as product need.

10. What would prove that a new approach worked?

Seek a measurable outcome such as fewer corrections, shorter approval time or a completed migration reconciliation. Avoid leading with a promised percentage saving.

The GOV.UK Service Manual's in-depth interview guidance recommends speaking with current or likely users and planning sessions around what the team needs to learn. Pilot the discussion guide, use a note-taker and analyse promptly.

Research records need care. The Service Manual's participant privacy guidance covers informed consent, data minimisation, secure handling and timely deletion. Report participant roles and recruitment method, but remove details that could expose workers or commercially sensitive payrolls.

Analyse before the memories fade

After each session, mark direct observation, participant report and researcher interpretation separately. Group notes around stages such as input, approval, submission, pension exchange and correction. Preserve counterexamples instead of forcing every interview into the initial theory.

End a round with a decision log. State which assumption gained support, which remains uncertain, what product or content choice changes and who must be interviewed next. A handful of interviews can reveal mechanisms and language, but it cannot establish England-wide prevalence. Use a suitably framed survey or operational dataset for that later question.

Finally, do not call an incentive a positive review or request public praise as a condition of taking part. Research participation and testimonial collection should remain separate activities with their own clear consent and disclosure.