Ask ten accounting firm partners whether they survey their clients and nine will say yes. Ask to see the last batch of responses and three will produce them. Surveys are like gym memberships - bought with good intentions, rarely used, quietly resented. The reason is almost always the same: the survey was designed to look impressive to the board, not to be answered by a busy client at 7pm on a Tuesday.
This guide walks through how to design a client satisfaction survey that you will actually run, that your clients will actually complete, and that will surface the early signals of churn while you still have time to do something about it.
Decide what you are measuring before you write a question
Most firm surveys fail at step one because nobody can articulate the decision the data will inform. Are you measuring overall loyalty so you can benchmark year-on-year? Are you measuring satisfaction with a specific service line? Are you trying to identify at-risk accounts? Each of these requires a different instrument, and combining them into one Frankenstein survey is why response rates sit at 8 percent.
Pick one primary purpose per survey. If you want loyalty, run NPS quarterly. If you want service-level feedback, run a five-question post-engagement pulse. If you want early churn warning, run a brief monthly health check on key accounts. Three small instruments will outperform one giant annual survey every time.
The five-question post-engagement survey
For most firms, the highest-value survey is the one sent within 48 hours of finishing a discrete piece of work - a tax return, a set of accounts, a payroll year-end. The client's memory is fresh, the emotional state is honest, and the response rate routinely beats 40 percent.
- How satisfied were you with the outcome? (0 to 10)
- How clearly did we explain what we were doing and why? (0 to 10)
- How responsive were we when you had a question? (0 to 10)
- Would you recommend us to another business owner like you? (Yes / Maybe / No)
- One thing we should keep doing, and one thing we should change. (Free text)
NPS, CSAT, CES - pick one and stick with it
The three dominant metrics are Net Promoter Score, Customer Satisfaction and Customer Effort Score. Each measures something slightly different. NPS predicts loyalty and referral behaviour. CSAT measures transactional happiness. CES measures how easy you were to deal with - and for professional services this often correlates most tightly with retention.
It does not matter which you pick. What matters is that you pick one, ask it the same way every time, and track the trend. A firm that has measured CES the same way for three years has infinitely more useful data than a firm that switched between NPS, CSAT and a custom 17-point index because the new marketing manager wanted a fresh look.
How to write questions clients will actually answer
Three rules. Keep each question to one sentence. Avoid double-barrelled questions like "How satisfied were you with the speed and quality of our work?" - speed and quality are different things and the answer is meaningless. And never ask a question whose answer you will not act on. Clients can smell vanity questions and they punish you with non-responses to the entire survey.
Delivery: format, timing, and channel
Send via the client portal, not personal email. Portal-delivered surveys feel like part of the professional relationship rather than a mass marketing email, and response rates run roughly 2x higher in our experience. Time the send for Tuesday or Wednesday morning, never Friday afternoon or Monday before 10am. Include the partner's real name in the subject line.
Length matters more than anything else. A five-question survey gets a 35 to 45 percent response. A ten-question survey gets 15 to 20 percent. A twenty-question survey gets 5 to 8 percent and the responses skew heavily toward the most opinionated clients, which is the worst possible sample for decision-making.
Closing the loop is the entire point
The single biggest reason firms stop surveying is that nothing happens with the responses. A client who took the time to answer and then heard nothing will not answer again. Every survey response needs a defined follow-up: anything 9 or 10 gets a thank-you and a request for a public review, anything 7 or 8 gets a personal email asking what would have made it a 10, anything below 7 gets a phone call from the partner within five working days.
Track close-the-loop completion as rigorously as you track survey completion. A firm that surveys 200 clients and follows up on 12 has wasted its budget. A firm that surveys 80 clients and follows up on all 80 has built something genuinely valuable.
Where the practice-management system fits
Surveys generate noise unless they are wired into the client record. Each response should land against the client in your practice management system, visible to the partner, the manager and the bookkeeper. A low score in month nine should trigger an automatic flag for the next partner review. Accupe is the practice-management layer that tracks every client touchpoint, surfaces satisfaction trends alongside fee, AML and deadline status, and makes survey data actually actionable rather than a PDF that sits unread on someone's desktop.
Closing
A good survey programme is not impressive. It is small, regular, ruthlessly short, and religiously followed up. Five questions, sent through the portal, tracked against the client record, closed within five days. Do that for a year and you will know more about your client base than any consulting report will ever tell you - and you will have caught two or three churn risks early enough to save them.