Building a Voice of Customer programme
Who to ask, when to ask, and how to turn answers into action. A practical guide to designing a VoC programme that changes decisions, not just dashboards.
A VoC programme is the system an organisation uses to collect customer feedback across channels, understand it and act on it, then tell customers what changed. Surveys are one part; the programme is the loop around them.
At a glance
A Voice of Customer (VoC) programme is the system an organisation uses to hear its customers on purpose: it collects feedback across channels, turns it into insight and makes sure something happens as a result. A survey tool on its own is not a programme. The programme is the loop around it: listen, understand, act, and tell customers what changed.
Most organisations already collect plenty of feedback. What they often lack is a design: clear questions, the right moments, owners for each kind of signal and a rhythm for turning what they hear into decisions. This guide walks through those choices.
Customers speak in many places, and only some of them are surveys. A good programme combines three kinds of listening post.
| Kind | Examples | Strength |
|---|---|---|
| Direct | Relational NPS, transactional CSAT and CES, open questions | You choose the question and the moment |
| Indirect | Contact-centre calls and chats, complaints, online reviews, social media | Unprompted, in the customer’s own words |
| Inferred | Behaviour and operational data: repeat contacts, returns, churn, waiting times | Shows what people do, not only what they say |
For relational measurement, survey a representative sample of active customers, so the score reflects the whole base rather than whoever happened to call this week. For transactional measurement, invite customers who have just completed the interaction you want to improve. Do not forget the people who are easy to miss: customers who left, prospects who did not buy and customers with an open complaint, whose feedback you should collect through the case rather than a survey.
Decide in advance which segments you need to report separately, such as region, customer value or channel, because each segment needs enough responses of its own.
Ask transactional questions as close to the moment as possible, ideally the same day. Run relational surveys on a fixed schedule, quarterly or twice a year, and keep the timing the same between waves. Then protect customers from over-surveying with a contact rule, for example no more than one survey to the same person within a set number of weeks, and no survey while a complaint is still open.
Keep it short: one metric question, one open “why” question and, only if you will use them, a few driver questions. A survey that takes more than a couple of minutes on a phone loses people. Keep the wording identical over time so the trend means something.
How satisfied were you with [the experience]? What is the main reason for your score?
How many responses do you need for a result you can trust, and how many invitations does that take?
Feedback only creates value when it changes something. Act on two levels.
Inner loop
The frontline follows up with individual customers, especially those who gave low scores; the closing-the-loop guide covers this.
Outer loop
The organisation looks across all feedback for the issues that keep coming back and fixes their root causes.
For the outer loop, tag open comments by topic and sentiment, find which topics move the score with a driver model such as SEM, and use MaxDiff when you need customers to rank competing priorities. Then tell customers and employees what changed; it is the strongest reason for them to keep answering.
Give the programme a single owner, usually a CX team, and a senior sponsor who can unblock budget and decisions. Each listening post needs a named owner, and each recurring topic needs someone who is accountable for fixing it. A simple rhythm keeps it moving:
Weekly
Operational review of alerts and open cases.
Monthly
CX review of topics and actions.
Quarterly
Leadership review of trends and investments.
Track the health of the programme, not only the scores: response rate, the share of low scores followed up within target, the number of actions completed and the trend of the main metrics.
- Pick two or three critical journeys where you know customers struggle or where most revenue sits.
- Set a baseline from the data you already have: complaints, reviews, operational figures.
- Launch short surveys and alerts at those moments, with owners for every low score.
- Hold the first action review after a month and agree two or three fixes.
- Share the first wins with customers, employees and leadership, then add the next journey.
As few as possible: one metric question and one open question cover most needs. Add driver questions only if someone will analyse and act on them.