Scope: what this audit covers and what it does not
The question this audit answers is whether the business earns a steady, honest flow of reviews where customers look, answers them well and acts on what they say, and what should change first. It covers reviews and ratings wherever customers read them: how and when customers are asked, which platforms the business is judged on and watches, replies and escalation, how reviews are shown on the business's own pages, the integrity of what is collected and published, and how themes reach the people who can fix them.
It does not re-teach the work. Asking and replying on a local profile belong to the local reviews and reputation management guide, collecting and displaying product reviews in a store to the e-commerce reviews and social proof guide, and app store ratings to the tutorial on app ratings and review replies. A review requestThe message or prompt that asks a customer to leave a review, sent once the product has been used and offered to every customer on the same terms. programme about to start or change is checked with the review request launch checklist, which this scorecard does not repeat.
Comments and replies on social accounts are scored by the social media presence audit, and store listing ratings are rated within the app marketing audit; this scorecard takes app replies only as part of the same review routine. In an online shop, product page reviews are one row of the online store audit; this scorecard rates them alongside every other platform the business is judged on. Every scorecard sits in the audit library.
Before you score: access and evidence
Ask for owner or manager access to each review profile, including the Google Business Profile (GBPLocal listing on Google Search and Maps.) for every location, any review platform the business uses on its own site, and the app store consoles if there is an app. Without access you can read public reviews but cannot see alerts, reported reviews or who replies, so the coverage and reply rows stay provisional. Fix an evidence window long enough to include ordinary weeks and any recent spike, and keep the same window for every rescore.
Collect:
- the list of platforms where the business has reviews, with access to each and its recent reviews;
- the review request messages, when they are sent and to whom;
- a sample of replies over the evidence window, positive and negative, with response times;
- the business's own pages that show reviews, with any review markup;
- any incentives, filters or moderation applied to reviews before they are published;
- records of review themes passed to operations, product or service teams.
Ask the people who send requests, not only the person who wrote the policy, how they decide whom to ask. The gap between the written process and what happens at the counter or in the follow-up email is often the main finding.
Criteria and scoring
Rate each row missing (absent or broken), partial (present but incomplete, inconsistent or out of date) or solid (matches the "what good looks like" column). The scale is your judgement against that column, not a number. Rows marked gate decide the band on their own when they score missing, because a rating built on filtered asking or invented reviews makes every other row describe a picture that is not true.
Asking
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Asking within the rules (gate) | Every customer can be asked, with no filtering by expected sentiment and no rewards the rules forbid | Missing: only happy-seeming customers asked, a gating form, or rewards for reviews where the platform forbids them / Partial: policy is sound but some staff or sites still filter / Solid: everyone asked the same way, any incentive allowed and disclosed |
| Request timing | Requests go when the customer can judge the product or service, through the channel they used | Missing: no request, or one sent at the moment of purchase / Partial: sensible timing but the wrong channel or a hard path to the form / Solid: timed to first use, sent in the customer's channel, with a direct link |
| Asking at every location | Where there are several sites, each asks in the same compliant way | Missing: each site does its own thing / Partial: a shared process that some sites ignore / Solid: one process, followed and checked at every site (or a single site) |
The gate follows the section on asking within the rules in the local reviews and reputation management guide, which states from Google Business Profile Help that selective asking and incentivised reviews breach the platform's policies, and that the CMA and the FTC treat undisclosed incentives and misleading review practices as consumer protection matters. The e-commerce reviews guide adds that regulators' guidance generally allows an incentive for leaving a review only when it does not depend on sentiment and is clearly identified. Treat both as general direction, never legal advice. Timing is scored against the same e-commerce guide, and multi-site asking against the guide to managing reviews across many locations.
Coverage
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Platforms that matter | The business knows where customers read reviews of it and keeps those profiles accurate | Missing: only the platform the owner uses is known / Partial: platforms known but some profiles unclaimed or out of date / Solid: a list of platforms, each claimed, accurate and owned |
| Monitoring and alerts | New reviews on every platform reach a named person quickly, with one view across sites | Missing: reviews found by chance / Partial: alerts on the main profile only / Solid: alerts on every platform, a named owner and one view |
For a local business the first platform is usually the profile shown beside the map in the Local PackThe map and short list of business listings shown for searches with local intent., where the local reviews guide explains that the rating sorts the options before anyone reads a word; the same guide covers platforms beyond the search listing and monitoring in their own sections. A group with several sites scores monitoring against the multi-location guide's single view.
Replies
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Reply practice | Replies to positive and negative reviews are prompt, specific and never argumentative, and follow a guideline rather than a script | Missing: no replies, or defensive ones / Partial: replies copied word for word, or only to negative reviews / Solid: a guideline, specific replies, escalation for serious complaints |
| App store replies | App reviews are answered and routed to the product team as part of the same routine | Missing: app reviews unread / Partial: answered but never passed on / Solid: answered, tagged and routed (or no app) |
| Sudden waves handled | A sudden run of negative or suspicious reviews is investigated and reported through the platform's process | Missing: no plan, or panicked public arguments / Partial: reported but the cause never investigated / Solid: cause found, fakes reported with a record, real failures fixed and answered |
Reply practice and escalation are scored against the guide to managing reviews across many locations, app replies against the tutorial on app ratings and review replies, and waves against the diagnosis of a sudden wave of negative reviews.
Display
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Reviews near decisions | Ratings and reviews sit near the price, the form or the button, attributed and dated | Missing: reviews on a separate page only, or none / Partial: shown, but far from the decision or unattributed / Solid: near the decision, attributed, dated, critical ones included |
| Review markup accurate | Review markup describes only reviews shown on the page, collected in the way the rules allow | Missing: markup for reviews not on the page, or self-serving ratings / Partial: accurate but incomplete or untested / Solid: matches the visible reviews and is tested (or no markup) |
| References for business buyers | Where buyers ask for references, customers have agreed what they will do and are not worn out | Missing: the same few customers called on without asking / Partial: a list but no record of use / Solid: a pool with agreed roles and a record of each use (or no such buyers) |
Placement is scored against the trust and proof elements guide, which treats reviews as Social ProofEvidence that other people have chosen, used or approved something, used to reduce a buyer's doubt. that works best close to the decision. Markup follows the section on review markup and rich results in the e-commerce reviews guide, which explains when Schema MarkupStructured data vocabulary. for reviews is eligible. References are scored against the guide to customer references and advocacy. Where a move of reviews closer to the button is tested, the conversion rate calculator helps compare the pages before and after on the same basis.
Integrity
| Criterion | What good looks like | Score guidance |
|---|---|---|
| No fake or selective reviews (gate) | No bought, invented or selectively published reviews, with any incentive disclosed where it is allowed | Missing: staff or bought reviews, generated reviews, or negative ones held back / Partial: no fakes, but moderation removes critical reviews or incentives go undisclosed / Solid: genuine reviews only, moderation for abuse, spam and personal data, incentives disclosed |
| Review flow watched | The flow of new reviews is tracked over time, and a stall is investigated before it shows in ratings | Missing: nobody watches the flow / Partial: noticed only when the rating moves / Solid: tracked on a rhythm, stalls traced to their cause |
The integrity gate is scored against the guide to AI claims and consumer protection, which treats a generated review as a fake review, and the incentives and moderation rules in the e-commerce reviews guide, which the site states from the FTC's business guidance. A stalled flow is diagnosed with the troubleshooting page for customers not leaving product reviews.
Acting on reviews
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Themes reach the people who can fix them | Recurring themes in reviews are passed to the teams that own them and the fix is recorded | Missing: reviews read only by whoever replies / Partial: themes noted but never passed on, or passed on with no record of the fix / Solid: a regular theme report, owners named, fixes recorded |
| Lost reviews understood | Reviews that disappear are traced to the platform's rules before anyone appeals | Missing: appeals sent without checking why / Partial: causes guessed / Solid: each loss traced to its likely cause, practice changed where it was the business's own |
The theme row follows the section on turning reviews into content and fixes in the local reviews and reputation management guide, and lost reviews are traced with the diagnosis of Google reviews disappearing.
How to score and prioritise
Score the two gates first. If customers are filtered before they are asked, rewarded where the rules forbid it or reviews are published selectively, the rating misleads prospects and is exposed to removal and to regulators, so the coverage, reply and display rows describe a picture that is not true. Mark them provisional.
Then rank the remaining weaknesses in this order:
- The platforms customers actually read, with monitoring, because unseen reviews cannot be answered.
- Replies and how reviews are shown near decisions, because both change what prospects conclude.
- Learning from themes and the remaining refinements next.
- Markup and display polish last, unless they hide a problem the other rows need to see.
Within each group, rank by how many prospects the weakness reaches. The profile that most customers read before choosing outranks a minor directory, and the busiest site in a group outranks the quietest.
Reading the result: what to do with each band
Red means a gate scored missing. Stop the filtered asking or the incentive at once, remove anything invented or bought, publish held-back reviews under a moderation policy limited to abuse, spam and personal data, and restart requests to every customer. Rescore once the new process has run for a full cycle of customers.
Amber means the gates hold at least partially but several other rows are partial or missing. This is the usual outcome. Claim and monitor the platforms that matter, write a reply guideline, and work through display and learning in order.
Green means the gates are solid and most rows are too. Move effort into using reviews: theme reports that change operations, tests of placement near decisions, and references for business buyers. Rescore each quarter or after any spike.
From findings to a prioritised fix list
Write each partial or missing row as a finding in plain words: what you saw, the evidence, and why it matters. Then turn the findings into one fix list with a line per fix: the finding, the fix, the owner, the guide that explains it and the rescore date. Order the list by the priorities above, not by how easy each fix is.
Review fixes often sit with people outside marketing. Asking usually happens at the front desk, in the delivery or in the follow-up message, so the owner of each customer-facing step owns its part of the asking fix. Replies sit with whoever holds the profiles, with escalation to the service or site lead; display and markup with whoever owns the website; and themes with the operations, product or service teams they describe. Present the band, the gate findings and the first few fixes to the person accountable for customer experience, and keep the full scorecard as the working copy for the rescore.
Illustrative example
A dental practice group audits its reviews after one site's rating fell
Skibbaro Dental, an invented group of four dental practices, runs the audit after one site's rating on its Google Business Profile fell from 4.7 to 4.1 over six months. Asking within the rules scores missing: reception staff ask only patients who seem happy, and one site hands out a card that sends unhappy patients to a private feedback form instead. Asking at every location scores partial, because each site words its request differently. Monitoring scores partial: alerts reach one practice manager for one site only. Reply practice scores missing, because the same three sentences are copied word for word under every review at every site. Themes score missing: 23 reviews in the window mention waiting times, and none of them reached the practice managers. Integrity scores solid, with no bought or invented reviews. The band is red on the asking gate. The fix list puts a compliant request for every patient, sent by message after the appointment with the private card withdrawn, first; a short reply guideline with escalation for clinical complaints second; and a monthly theme report to each practice manager third.
Common mistakes
- Chasing the rating before checking how it was earned. A high rating built on filtered asking is a liability. Score the gates first.
- Treating a gating form as harmless feedback. Routing unhappy customers away from the platform produces a misleading picture. Score it as filtering, not as service.
- Scoring replies on speed alone. A fast reply copied word for word under every review reads as indifference. Score specificity and tone as well.
- Watching one platform. Customers read reviews where they choose, not where the business looks. Score coverage against where customers actually read.
- Hiding critical reviews. A page with only top ratings reads as filtered and invites scrutiny. Score display on honesty as well as placement.
- Answering complaints and never fixing their cause. The same theme returns in the next review. Score the theme row on recorded fixes, not on reports sent.
