Commerce reviews key metrics explained for evidence-led reporting

What to take away

Commerce reviews key metrics are the countable outputs of a review programme, such as review volume, average rating, response rate and conversion lift, each tied to a named source so the figure can be checked.

Pick two or three primary metrics per reporting cycle and treat the rest as diagnostics.
Every number needs a source, a date and an owner, or it is an opinion with a decimal point.
Platform dashboards, analytics tools and search data measure different things, so do not mix them in one column without labelling the origin.
Review data pulled through an API needs its own monitoring, or usage problems stay hidden until a report breaks.
Separate leading indicators from outcome metrics, and never let one column carry both.

Why the source matters as much as the metric

A review count from a marketplace dashboard and a review count from your own store will rarely match. One counts listings, the other counts orders. Neither is wrong, but a report that treats them as the same number will mislead whoever reads it.

Metric definition checklist

Define what counts as a review
State when a rating is recorded
Name the period covered
Name the tool and report
Record the date pulled

Write a one-page definitions note and keep it with the report. It should state what counts as a review, when a rating is recorded and which period the figure covers. Teams that skip this step spend meetings arguing about definitions rather than decisions.

For UK review authenticity, consult the Competition and Markets Authority (CMA) guidance and the fake-review provisions in the Digital Markets, Competition and Consumers Act 2024. If review content is used in marketing, check the Advertising Standards Authority (ASA) and Committee of Advertising Practice (CAP) Code as well.

Start by writing down where each figure comes from. Name the tool, the report and the date it was pulled. If you cannot name them, the metric is not ready for a board pack. The wider framework for this sits in the commerce reviews measurement and reporting guide, which sets out how the pieces fit together.

The metrics that carry the most weight

Volume, average rating and response rate are the three most quoted numbers, and the three most easily distorted. Volume rises when you ask more people. Average rating drifts when a single batch of unhappy customers arrives together. Response rate depends on where the prompt appears.

Core metrics and their risks

Metric

Volume: More requests sent
Average rating: Batch of unhappy customers
Response rate: Where prompt appears
Conversion lift: Hard to isolate

What distorts it

Volume: With request rate
Average rating: With period and base
Response rate: With placement noted
Conversion lift: As a range

How to report it

Volume
Average rating
Response rate
Conversion lift

Two leading indicators deserve a place on the same page: the share of orders that receive a request, and the number of days between delivery and the review landing. Both move before rating averages do, which makes them useful early warnings.

Conversion lift is the metric buyers care about most, and the hardest to isolate. Treat it as a range, not a point. Report the period, the comparison group and the change, then let the reader judge.

Define each metric before reporting

Use the same period, population and source when applying these calculations:

Review volume:count the published reviews in scope during the reporting period; state which storefronts or platforms are included.
Average rating:total the star scores and divide by the number of ratings in scope.
Request-to-review response rate:published reviews divided by successfully delivered review requests. Keep this distinct from the retailer’s response to customer reviews.
Share of orders receiving a request:eligible orders sent a review request divided by all eligible orders.
Days between delivery and review landing:the median elapsed days from delivery to a review being posted; state the period and population.
First-response time:the median elapsed time from a review being posted to the retailer’s first reply.
Conversion lift:conversion rate for a comparable group exposed to reviews minus the conversion rate for a control group. This is the absolute difference in percentage points; divide that difference by the control-group rate for relative lift. State how the groups were formed, since the difference alone does not establish causation.
Cost per review:review-programme costs in scope divided by published reviews collected in the same scope and period.
Revenue attributed to reviewed products:sum the sales attributed to those products under the stated analytics attribution rule and period. This is attributed revenue, not necessarily incremental revenue.
SLA response times:reviews answered within the stated response target divided by reviews requiring a reply. State the target and reporting period.

These definitions make the denominator and scope visible; record the calculation alongside the source and date pulled.

Using platform data without overclaiming

Review volume and ratings should come from the review provider’s own reporting or export. Examples include Trustpilot, Google Business Profile, Feefo, Reviews.io, Yotpo and Bazaarvoice. Their definitions and coverage can differ, so label the provider and scope rather than treating platform totals as interchangeable.

For conversion and revenue measures, use your analytics source, such as Google Analytics 4, and state the attribution rule used. Looker Studio can bring reported figures together in a dashboard, but it does not make different source definitions equivalent. Record the source and extraction date for each figure.

Monitoring the pipes, not just the numbers

If reviews flow into your site through an API, the metric you most need is uptime. Google's documentation on reporting and monitoring for the Places API covers the usage and quota signals available to developers. That check stops a quiet data gap becoming a wrong quarterly figure.

Set an alert for failed calls. Review it weekly. A missing batch of reviews looks identical to a fall in customer sentiment until someone checks the logs. Log the check in the same place as the metric, so the audit trail shows that the data arrived.

A decision table for choosing metrics

Choose

Launching a review programme: Review request rate and first-response time
Reporting to a finance team: Cost per review and revenue attributed to reviewed products
Comparing two platforms: Volume, rating and response rate from each, labelled separately
Reviewing supplier performance: SLA response times and API error rates
Under pressure to show growth: Change against a fixed prior period

Avoid

Launching a review programme: Conversion lift before you have a baseline
Reporting to a finance team: Star rating averages on their own
Comparing two platforms: A blended average across both
Reviewing supplier performance: Sentiment scores with no stated method
Under pressure to show growth: Percentage growth with no base figure

Common questions

How many metrics should a review report contain?

Three primary metrics and no more than five supporting ones. A longer list usually means nobody has decided what the report is for.

Can I compare platform metrics with my own analytics?

Only if you label each source and accept that definitions differ. Comparing them silently is one of the commerce reviews measurement mistakes in England that undermines otherwise sound reporting.

How often should review metrics be refreshed?

Match the cycle to the decision. Weekly for operational alerts, monthly for programme health, quarterly for commercial impact. For the monthly programme-health report, the ecommerce lead can own definitions and sign-off, with the review-platform administrator supplying review data and the analytics owner supplying conversion and revenue data. Daily dashboards rarely change a decision.

What makes a review metric defensible?

A named source, a stated period, a defined calculation and a person accountable for it. If any of the four is missing, expect the figure to be challenged.