Once the review data of several hundred stores lands on a single screen, a new problem appears: everyone looks at the same number and reads something different. To a regional manager "negative rate" means one- and two-star reviews; to a customer experience director it means reviews the AI classified as negative, and the two quote different figures in the same meeting. This dashboard metrics guide explains the definition, scope and operational reading of every indicator in BranchVoice.
Why dashboard metrics need a single definition
A chain's reviews arrive from five platforms with different natures. Google Maps and Yandex Maps reviews carry a star rating; reviews from Şikayetvar (Turkey's consumer complaint site), Ekşi Sözlük (a long-running Turkish collaborative forum) and X have no stars, only text. A star-based metric excludes the unrated sources; a content-based metric cannot count the text-free, stars-only reviews. BranchVoice therefore states the scope of every metric and uses the same definition on the dashboard, in reports, in alert rules and in the weekly digest.
The filter bar: context for every number
Every figure on the Overview page follows the filter bar at the top: period (last 7, 30 or 90 days, this year or a custom range), comparison (previous period, same period last year or off), platform, channel (physical store, e-commerce, general) and region, city or store. Filters live in the URL, so a shared link opens the same view.
"Change versus the previous period" always means the preceding period of equal length: the previous 30 days in a 30-day view, the previous 7 days in a 7-day view.
The scorecard: five cards and a coverage strip
The first row of the page is a five-card scorecard. Each card shows the value, the change against the previous period and a small trend line, and clicking it opens the matching review list.
- Review volume: reviews collected in the period, with a platform breakdown underneath.
- Experience score: a 0 to 100 score computed from the content of reviews the AI analysed; 50 is neutral, below is negative-leaning. Only reviews with text that were analysed count.
- Star rating: the average of rated reviews, map sources only. Reviews with high stars but negative content are counted separately.
- Loyalty balance: the share of reviews recommending the brand minus the share saying they will not return; below zero is a net churn signal.
- To-do: unanswered reviews needing same-day attention plus reviews with critical risk, broken down into urgent, critical, unanswered negative, open anomalies and reviews awaiting store matching.
The coverage strip beneath the scorecard tells you how much to trust the numbers: total reviews, reviews with text, reviews analysed, reviews found irrelevant and when the latest data arrived.
Metric glossary: stars and content
Star-based metrics
Alert rules, store comparison and reports share a family of star-based metrics. They are computed from rated reviews only:
| Metric | Definition |
|---|---|
| Review count | Total reviews in the window |
| 1-star count | One-star reviews in the window |
| Negative count and rate | One- and two-star reviews; rate as a percentage |
| Positive rate | Percentage of four- and five-star reviews |
| Average rating | Mean rating, 1 to 5, of reviews in the window |
| Google store rating | The current rating on the store's Google listing |
| Unanswered negative age | Age of the oldest unanswered one- or two-star review in the last 30 days |
Average rating and Google store rating should not be confused. The first is the mean of reviews in the selected period and moves daily; the second is the all-time figure Google shows on the store page. The period average can drop while the Google rating holds steady for a while, and that difference decides which one an alert rule should watch.
Content-based metrics
AI analysis gives every review a sentiment class, a 0 to 4 risk score, a response urgency, churn and recommendation probabilities and topic labels. The metrics derived from these cover analysed reviews: experience score, negative content rate, high-risk and critical review counts, urgent unanswered count and churn rate. Each can also drive an alert rule; a rule such as "critical reviews at a store exceed 2 in the last 7 days" works on content rather than stars.
The contradiction between rating and content is its own indicator: reviews that give four or five stars yet describe a bad experience, or one or two stars alongside praise. When that count is high, the star rating alone is misleading.
Topic distribution and the impact matrix
Topic distribution shows which subjects reviews touch, but volume and negativity are not the same thing. Staff attitude may be the most discussed topic while mixing praise and complaint; returns and exchanges may be rarely mentioned yet almost entirely negative. The Overview therefore plots topics on an impact matrix: the horizontal axis is the topic's share of reviews, the vertical axis its negative rate, and the dot size its number of high-risk reviews. The quadrants are labelled "priority" (top right), "strength" (bottom right), "watch" (top left) and "minor issue" (bottom left).
Next to the matrix, the rising and falling list ranks the topics whose share grew or shrank most against the previous period. The Topic Analysis page adds each topic's trend, platform and store breakdown and sample reviews; your own topics are defined through analysis groups and topic taxonomy.
Stores, operations and customer voice
The stores row lists the best and worst five stores in the selected region or city by experience score, stars, review count, high-risk count and unanswered negatives. For a deeper comparison, use the store comparison screen.
The operations block computes response rate only across sources where replies are possible, and adds median response time and the ageing of unanswered negative reviews. The customer voice block gathers suggestion categories, emotion distribution, employee mentions and moderation counts. The e-mail version of the same numbers is described in the weekly digest email post.
Next steps
Metrics with a single definition end the data argument in a chain and start the action discussion. Once your store list is connected, the scorecard and impact matrix fill up with the first analyses. To see them on your own data, request a demo.