The survey your clients will answer.
The scenario:
Last year’s client survey had twelve questions, a rating matrix, and a 9 percent response rate. The year before, same.
Long surveys measure one thing: who has time for long surveys. The clients you most need to hear from, the drifting ones, are exactly the ones who delete it.
The prompts:
In your AI tool of choice.
STEP 1 - Design the task
You are designing a client feedback system for an MSP.
Context:
Past survey attempts and response rates: [honest history]
What we most need to know: [renewal risk, service quality, unmet needs]
Client contact points where an ask lands naturally: [ticket close, monthly report, QBR, invoice]
Design the two-question core: one score question and one open question, worded to invite honesty rather than politeness. Then the moment map: which client touchpoint carries the ask, at what frequency per client, with the annual cap so nobody is over-surveyed.
-
STEP 2 - Build the discovery
Write the asks: the 40-word email version, the one-line in-signature version, and the spoken version an account manager uses at the end of a review. Then the response handling: the thank-you for every response, the 48-hour rule for responding to any score below par, and who owns that response.
-
STEP 3 - Close the loop
Build the visible-change loop: the quarterly “you said, we did” note, under 100 words, sent to every respondent, and the internal review: 30 minutes per quarter, themes not individual scores, one service change per quarter attributed publicly to client feedback. Include the measure: response rate trend and repeat-response rate, because people answer again when the first answer visibly mattered.