Review-based customer experience metrics: negative share, change and coverage

Why build metrics from reviews

Most customer experience metrics are scores from surveys. Reviews add something different: unprompted, public descriptions of what customers noticed. Classified consistently, they produce metrics that explain the scores, store by store and topic by topic.

1. Mention volume

The number of mentions of a topic in a period. A review that mentions three topics counts once for each. Volume shows what customers talk about; on its own it does not show whether that is good or bad.

2. Negative share

For a topic: negative mentions divided by mentions with a known sentiment. It shows how bad the topic is when it comes up. Report it with the count, and keep unclear or neutral mentions visible separately.

3. Change

The difference in negative share between this period and the last, in percentage points, for the same stores and sources. Change is what turns a measure into a signal. A topic moving from 20% to 32% negative on a steady volume is worth reading about this week.

4. Coverage

• Classified share: reviews with at least one topic / all reviews with text.
• Unclear share: reviews too short or vague to classify.
• Rating-only share: reviews with no text at all.

Coverage tells you how much of your feedback the other metrics describe. If half your reviews are rating-only, topic metrics describe the other half.

5. Reply measures

• Reply coverage: reviews with text that received a reply / reviews with text.
• Reply time: median hours from review to reply, for negative reviews.
• Workload: open reviews needing action, by owner.

Reporting them

• By topic, store and month.
• Counts beside every share.
• Small samples flagged, not ranked.
• Two or three quotes behind any highlighted change.

Pitfalls

• Mixing sources, such as store and app reviews, in one share without saying so.
• Changing topic definitions without noting the date.
• Comparing stores with very different volumes.
• Treating volume growth as worsening experience when it is just more customers.

Common questions

A worked example

A store receives 300 reviews in a month. 210 have text; 30 of those are too short to classify, so coverage is 180 of 210, or 86%. Of 120 mentions of staff, 18 are negative: staff negative share is 15%, down from 22% last month on a similar volume, a change of minus 7 points. Of 60 mentions of billing, 33 are negative: 55%, up from 40%, a change of plus 15 points. The headline for the store manager is clear: staff improved, billing worsened, and here are the reviews.

Setting targets

Fixed targets for negative share rarely suit every store. Compare each store with its peer group's typical range and with its own last period instead. Set targets for the drivers you control, such as queue times, and let the review metrics show whether customers notice.

Positive share

The mirror of negative share is worth tracking for strengths: when a topic customers usually praise starts losing positive mentions, it often signals a problem before complaints rise.

Doing this with Falcon

Falcon reports these measures for every topic, store and month: mention volume, negative share with unclear sentiment kept visible, change against the previous period with small samples flagged, and reply coverage, turnaround and workload in its CRM. Every number opens the reviews behind it.