Reading VoC data honestly — behind the headline

Voice of Customer · ~7 minute read

A rising NPS that hides a vulnerable-customer collapse. The disciplines — distribution, segment, drift, verbatim, complaint — that turn the headline into honest insight.

Distributions, not just averages

A headline NPS of +28 can be a tight distribution around +28, or a bimodal split between promoters and detractors, or a score dragged up by one segment and dragged down by another. The aggregate hides shape.

Read the distribution; compare shape over time; watch for drift the headline conceals; report shape alongside the score where the audience can handle it.

The average-customer trap

There is no average customer. Treating the aggregate as a real customer leads to design and policy decisions that fit no one well. Segment routinely; persona-ground when working through implications; watch the tails; resist the headline.

"Our customers love us" said about an aggregate where 25% are detractors is dishonest.

Verbatims and complaints, read at scale

Theme verbatims at scale (AI-aided where it fits); read a sample directly; read across the score range; read by segment. Complaints — every escalated complaint, not the summary; theme and trend; connect to other VoC posts.

A weekly sample of 20-30 verbatims keeps the analyst grounded; aggregated themes lose nuance the raw text preserves.

Worked example

A VoC report: monthly NPS +22 to +28 over six months. Undisciplined read: "VoC programme working." Disciplined read: response rate fell 18% to 12%; vulnerable NPS fell +9 to +2; complaints up 15%; speech-analytics theme on "claims delay" up 40%; verbatim "I gave up trying to complain."

The headline is misleading. The disciplined analyst surfaces the gap rather than reporting the headline.

The closing principle

The headline alone misleads. Read distribution, segment, drift, drivers, verbatim, complaint — the gap between aggregate and segment is often where the discipline starts.