Accounting services customer interviews: eight questions that get useful answers

Eight questions carry a useful customer interview about accounting or bookkeeping. The table gives each question and the evidence to keep, and the sections below explain how to ask them.

Start from the precipitating event

Start with a recent event that prompted the search, such as a deadline, notice, error or business change. Invite the participant to describe it in their own words, then name the result they wanted. This keeps the discussion tied to their intended job rather than an assumed service.

Clarify the work behind broad labels such as bookkeeping. It might involve reconciled records, payroll, accounts, a return, decision information or another need. Ask what a satisfactory result would have meant to the participant, and record that answer separately from any service idea it may suggest.

Set boundaries around the work

Establish which entity and period the participant means without offering categories for them to select. A person seeking help may distinguish personal, partnership and company affairs in their own way.

Capture how they describe the boundary, then ask where the relevant records were held and what was missing at the outset.

If examples are reviewed, use redacted material through an approved data process. Ask what records were ready, and whether anything had to be found, corrected or supplied later. Where another professional took part, establish their role in the sequence, including any prior accountant.

Trace how the provider was chosen

Map the people involved, which may include directors, employees, a spouse, a bookkeeper, a bank, a software provider or an adviser. Find out who could approve the engagement and figures. Then follow the actual search route, the factors shaping the shortlist and the evidence that changed the participant's view.

Ask whether the search began with referrals, directories and professional registers, a web search or an existing relationship. Focus on actions and evidence encountered, not a general verdict on brands. This helps retain the decision sequence and the roles of people involved without turning an interview into a sales conversation.

Reconstruct the point of friction

Invite a chronological account of where progress became difficult. It may have involved repeated questions, inaccessible documents, unclear scope, a slow reply or an unexplained cost. Ask for the incident behind the judgement, including what happened immediately before and after it, rather than accepting a broad description of the experience.

Separate provider work from tasks retained by the customer. Record deadlines, approvals, record correction and customer decisions, then ask what each person did. This can show where responsibilities were unclear without turning the discussion into advice or an assessment of fault. Do not assume who should have acted before hearing the participant's account.

Establish the meaning of finished

Ask what showed the work was complete and whether anything remained unresolved. An accepted submission, reconciled report, payslip run or documented answer may be a concrete endpoint. Record the outcome received alongside any remaining issue that affected the participant's view of the work.

Explore what they might change next time, perhaps preparing earlier, collecting clearer evidence or choosing another route. Do not immediately translate an answer into a feature request. Neutral prompts about what happened before an event, who else was involved and how a decision was made can expose the sequence behind a response.

Put limits around the conversation

Before the interview, explain its purpose, whether recording will happen, how information will be used and how withdrawal works, as set out in research ethics guidance. The GOV.UK Service Manual on user research covers planning research, recruiting suitable participants and observing real users. For information-handling planning, consult the ICO guidance on research provisions.

Make clear that the conversation is not accounting or tax advice. Do not collect full tax references, bank details, payroll files or credentials. If sensitive material appears unexpectedly, stop recording, use your approved incident route (see the guide to data protection duties) and retain only research data that is necessary.

Recruit for relevant experience

Select participants whose experience fits the research question, rather than choosing people because their answers might support a planned service. Use entity and task screening, recent experience, decision role, record condition and access need to guide selection.

Exclude staff, suppliers or professional advisers from an end-customer journey study unless they are a separate research group. Keep notes on the screening criteria and the gaps in the resulting sample, so the account of customer experience is not confused with other professional perspectives.

A balanced round could include two sole traders, two limited company directors and two people from partnerships, with at least one recent payroll task and one recent VAT return task. A typical working guideline is five to eight interviews per distinct user group in a round; look for saturation when two consecutive interviews add no new themes.

Allow time for notes and any immediate safeguarding action. Use participant codes and store contact details apart from research material. Explain any incentive before consent, without making payment depend on favourable answers. Delivery-team observers should follow confidentiality rules and note observed events before discussing their explanations.

Keep evidence distinct from interpretation

After each round, separate observations, interpretations and proposed actions. For example, notes described a sole trader facing a 31 January deadline who tried three accountants; two had not replied, and the participant chose the one that answered in a day.

Keep those details as the observation rather than turning them into a general claim about customers.

Record an interpretation about response speed separately from an action proposed to test a response-time promise. Compare participants who bought, declined and abandoned, and choose later which findings to measure with the guide to decision metrics. Treat a repeated claim as a finding only when its evidence and sample limitation are recorded, and keep contradictory observations visible until further evidence resolves them.

Prepare the discussion guide before sessions, separate personal details from research material and distinguish observers' notes about events from their explanations. The audience research method can help frame a wider research round while keeping these interviews focused on customer experience.

Using the answers when you choose an accountant

The same eight questions help a buyer as well as a researcher. Before appointing, ask a firm's existing clients how they chose it, where work stalled and what counted as finished. Compare that with what the firm promised.

Then check the firm itself on the register of a professional body such as ICAEW, ACCA, AAT or ICAS, and read the selection checklist before you sign an engagement letter.