Three Prompts to Interrogate a Member Survey Result
SENIOR INTERNET RESOURCE
The solution members ask for is rarely the problem they have.
Every membership organisation eventually sends a survey out — an annual satisfaction check, a renewal feedback form, a one-off pulse survey after a change. The results usually come back with one item members mention more than anything else
Your survey came back with a clear top request. Before that becomes a project, these three prompts help you find out what's actually underneath it — run in order, in one AI conversation.
Run these in whichever AI tool you already have; each one feeds the next, so keep the whole conversation in a single thread rather than starting fresh each time.
Interrogate the request
This is the one that matters most. It stops leadership acting on the survey until they've articulated the problem underneath it. Most organisations never do this step.
I run a membership organisation called ... Our member survey returned this as a top request: "[paste request verbatim]". Before we act on it, I want to understand what's actually behind it. Ask me ten questions, one at a time, to separate the solution members proposed from the problem they're actually experiencing. Don't accept my first answers — push where I'm vague. Do not offer solutions.
Answer honestly rather than defensively. If you find yourself unable to answer a question, that gap is in itself a finding.
Find the evidence
More often than not, survey designs are structured to tell you what members feel, not what they do. This prompt sends you looking for the behavioural evidence that almost certainly already exists somewhere in a system you own.
Based on what we've established, list every place inside my organisation where behavioural evidence for this problem might already exist — systems, logs, records, anything routinely captured but not analysed. For each, tell me the specific question to ask of that data and what pattern would confirm or contradict the survey result. Assume no budget for new tooling and no analyst on staff.
Typical places worth checking
- Website form submissions and contact-page enquiries
- CRM records and membership status changes
- Call logs and helpdesk or inbox ticket categories
- Login and portal access data
- Event and training registrations, and drop-off points
- Renewal and lapse timing
Get to root cause
Once you have the data, the temptation is to act. Don't — not until you've asked why the behaviour is happening, and then asked why again.
Here's what the data shows: [paste findings]. Run a five-whys analysis on this behaviour. At each level, tell me what evidence would confirm that level versus what I'm assuming. Then tell me where the root cause sits — process, technology, or communication — and what the members' original request would and wouldn't have fixed.
That last clause is the important one. It tells you the cost of having simply built what the survey asked for.
What these prompts can't do
An AI doesn't know your organisation. It doesn't know which systems you own, which data has been quietly accumulating for years, or which patterns you've seen before elsewhere.
What it can do is force a disciplined interrogation of a survey result — which, if you don't have an insight team, is a genuinely useful place to start.
Treat this as the start of the diagnosis, not the diagnosis.