
Audience research
Part of Accounting services audience research: a practical method
Survey methods for accounting clients, compared by question, reach, bias and data risk
Compare practical accounting service survey methods in England by research question, reach, accessibility, data risk, bias and analysis burden.
What to take away
- Name the decision, the population and the question before you compare any tool.
- Match method to questiononline forms suit satisfaction after a known event, interviews suit unfamiliar processes, transactional triggers suit filing accuracy.
- Reach and bias trade off, because a list of current clients cannot speak for prospects or lost enquiries.
- Check whether quoted prices exclude VAT, and whether the unit is per user or per account.
- Six pre-launch checkssampling frame, lawful basis, pilot, response-rate target, analysis plan and retention terms.
Define the English client population and its sampling routes
England's accounting client base is not one audience. It covers limited companies on the public register, VAT-registered traders, sole traders filing Self Assessment and employers running payroll. The frame comes before the platform.
The Companies House register is the strongest frame for limited companies, showing incorporations, sector codes and filing history. Companies House publishes those records and the filing duties behind them.
Professional bodies including ICAEW and ACCA keep member directories, which reach practices rather than clients. Use them to recruit accountants, not business owners.
Practice client lists are the quickest route and the most biased. They drop former clients and businesses that never bought accountancy services.
Match the method to the question
Decide what you need first: incidence, ranking, satisfaction, recent behaviour or explanation. A comparison of question types shows how each choice changes the design. Surveys are poor at uncovering an unfamiliar process, where interviews or observation give better early evidence.
GOV.UK's user research planning guide notes that surveys and benchmarking need far more participants than a small qualitative round. Draft neutral, single-purpose questions, set a real reference period and add "not applicable" where legitimate. Pilot with intended users.
Compare named platforms on price, hosting and data risk
Prices below are list prices or labelled typical ranges as of 2026. They exclude VAT unless stated. Confirm the billing unit, per user or per account, before you compare.
Compare named platforms
Best question fit
- SurveyMonkey
- Experience after a known event
- SmartSurvey
- Short feedback in a live journey
- Qualtrics CoreXM
- Routing, quotas and subgroup analysis
- YouGov, Toluna, Kantar or Ipsos panels
- Screened participants at speed
- Telephone questionnaire
- Inclusion where online access is hard
Reach
- SurveyMonkey
- Email link to clients
- SmartSurvey
- Email, embed, SMS, offline
- Qualtrics CoreXM
- Web, mobile, panel
- YouGov, Toluna, Kantar or Ipsos panels
- Consumer and business panels
- Telephone questionnaire
- Landline and mobile
Main bias
- SurveyMonkey
- Clients may not represent prospects
- SmartSurvey
- Misses people who never reach the page
- Qualtrics CoreXM
- Quotas can mask coverage gaps
- YouGov, Toluna, Kantar or Ipsos panels
- Recruitment shapes who appears
- Telephone questionnaire
- Interviewer effects
Data risk and controls
- SurveyMonkey
- US-owned; check EU hosting and retention. From about £39 a month, excluding VAT.
- SmartSurvey
- UK hosting; check deletion terms. From about £29 a month, excluding VAT.
- Qualtrics CoreXM
- Hosting region negotiable. Quote-based, several thousand pounds a year, excluding VAT.
- YouGov, Toluna, Kantar or Ipsos panels
- Consent sits with the supplier. Several pounds per complete, excluding VAT.
- Telephone questionnaire
- Recording and consent needed. Costs more to deliver than self-completion.
Supplier control checks
- Question, routing and validation needs
- Keyboard, screen-reader, mobile, language access
- Authentication and duplicate prevention
- Data hosting, subprocessors, access, export
- Retention, deletion, backup, incident terms
- Consent features and raw data availability
- Version history, change control, contract exit
Example: filing accuracy versus satisfaction
Say a practice wants evidence on whether its accounts are filed accurately. Accuracy depends on bookkeeping records and correction effort, so client recall is weak evidence. A transactional email triggered from Xero or QuickBooks after each filing ties the answer to a dated event.
Satisfaction is a different question, so a lighter method fits. Send a short online form after a defined service event, then keep those scores away from accuracy, accepted filings and correction effort, as metrics that lead to a decision explains.
Benchmarks for sample size and response rate
Two sums make the numbers concrete. For an estimate near 50 per cent, a 95 per cent confidence level with a five-point margin of error needs about 384 complete responses. Tightening the margin to 2.5 points raises that to roughly 1,537.
Then divide by the response rate you expect. A client email list returning 20 per cent needs 500 invitations for 100 completes.
Control recruitment, analysis and reporting
Record the sampling frame, invitation route, response rate, exclusions and unfinished submissions, then report counts with percentages, fieldwork dates and exact wording. Keep customers, prospects, accountants and staff in separate groups.
The ICO's lawful basis guidance requires organisations to identify and document a basis before processing. Consent to take part and the data-protection basis are separate decisions.
Set the numerator, denominator, treatment of partial responses and subgroup rules before fieldwork. Weighting cannot repair a population you never reached. For a response-rate target, sourcing a benchmark means treating it as a question rather than a number to copy.
Common questions
Which platform suits a short client feedback form?
SmartSurvey or Typeform suit a light UK-hosted form, and Microsoft Forms works if the practice already holds a Microsoft 365 licence. Match the tool to the question.
When should I use interviews instead of a survey?
When the process is unfamiliar and you cannot yet write good closed questions. Observation or interviews give better early evidence, and a survey can follow.
What sample size do I need?
An estimate near 50 per cent, at 95 per cent confidence with a five-point margin of error, needs about 384 completes. Subgroups need their own calculation.
What should I check before signing a contract?
Retention and deletion terms, consent features, raw data export, version history and contract exit. Confirm the billing unit and whether prices exclude VAT.



