How UK marketers check review-led advertising against the ASA CAP Code

What to take away

What to take away

How the CAP Code treats testimonials and review claims

The CAP Code is the rulebook the Advertising Standards Authority enforces for non-broadcast ads in the UK. It sits alongside the broadcast codes, and the Committee of Advertising Practice writes it. The consolidated set is published in one place, which is where a compliance check should start: Advertising codes - ASA | CAP.

Rule 3.45 is the heart of it. Marketing communications must not display a testimonial, review or endorsement unless the advertiser holds evidence that the person really made it and that it relates to the product being advertised. A quote lifted from an old version of the product, or from a customer who never used it, fails that test.

Rule 3.46 adds the qualifier most review-led campaigns miss. If a testimonial or review is not typical of the results a consumer can expect, the ad must say so.

One glowing review from a power user is not typical of a base of ten thousand users, and the ad has to make that clear rather than let the quote do the work.

Rule 3.47 covers the reverse case: a review or endorsement that is presented as typical when the evidence shows it is not. The distinction between 3.46 and 3.47 matters because it decides whether you add a disclaimer or pull the claim.

Relevant reviews also have to be attributed. If a review is used to promote a product, the ad should identify where it came from, so a consumer can weigh it. An unattributed pull quote on a landing page is weaker evidence than a named reviewer with a visible source.

Section 3 as a whole is the misleading advertising section, and it is the one most review claims are judged under. The full text is at 03 Misleading advertising - ASA | CAP.

The four tests a review claim has to pass

The ASA remit and what it does not cover in review marketing

The ASA regulates ads in paid space: display, search, social, influencer content, email, print and out-of-home. It also covers ads on a company's own website when they are promoting goods or services, which catches most review-led landing pages.

The remit stops short of editorial content. A review written by a journalist, or a customer review left on a third-party platform, is not an ad. The moment you quote that review in your own marketing, it becomes part of your ad and the CAP Code applies to your use of it.

Organic social posts by a brand are usually outside the remit unless they are part of a promotional campaign. Paid amplification brings them in. This is the line that catches agencies who treat a boosted post as content rather than advertising.

The ASA is not a consumer redress body. It cannot order a refund, award compensation or settle a contract dispute. Its job is to rule on whether an ad broke the Code and to require a change.

The Competition and Markets Authority handles the underlying consumer law, including fake reviews and misleading commercial practices. The Digital Markets, Competition and Consumers Act 2024 strengthened the CMA's direct enforcement powers in this area, so a fake-review problem can now attract a regulator with fining power rather than a ruling alone.

Data protection sits with the Information Commissioner's Office. Publishing a review with a customer's name, photo or order detail is processing personal data, and consent or another lawful basis is needed. That is a separate check from the CAP Code.

For a wider set of pre-publication checks, including the disclosure side, see our notes on review-led advertising rules.

A step-by-step check before publishing a review-led claim

This is the sequence we would run before any review-led ad goes live. It takes an afternoon for a single campaign and it is worth doing once properly, then repeating per asset.

Step seven is the one that saves money later. Review counts and averages change weekly. A claim that was accurate in March can be misleading by June without anyone editing the page.

Worked example: a 4.9 star claim on a services page

A UK agency publishes "Rated 4.9 out of 5 by our clients" on its homepage. The figure comes from a review platform with 38 reviews, of which 31 are five star.

The claim is accurate on its face, but the ad does not say the sample is 38 clients out of several hundred served. Under rule 3.47 that can mislead about how typical the rating is.

The fix is either to state the number of reviews next to the rating or to remove the rating and quote a named client with the result they got.

If the agency also runs "our clients see a 40 per cent uplift in enquiries", that is a results claim, not a review claim, and it needs its own evidence base. Mixing the two in one sentence is where most complaints land.

Keep a written record of the correction path for any claim you withdraw, because a versioned disclosure and correction record is what you show a regulator when a complaint arrives months later.

Misleading advertising rules applied to star ratings and counts

A star rating is a claim. It must be accurate at the time it runs, and it must not exaggerate the weight of the evidence behind it.

Three things go wrong most often.

First, the platform is not named. A 4.8 on Trustpilot is not the same as a 4.8 on Google Reviews, Feefo or Reviews.io, and consumers read an unlabelled rating as a general one.

Second, the sample is hidden. A rating from 12 reviews carries different weight from one built on 4,000, and the ad should let the reader tell the difference.

Third, the date is stale. Ratings move. A figure captured a year ago and left on a page is a claim you can no longer support.

Review counts have the same problem. "Over 10,000 five star reviews" is a claim about a number, and the number has to be verifiable and current. If the count includes reviews from a discontinued product, say so or exclude them.

Award badges and "number one" claims sit in the same family. If a badge comes from a paid scheme, the ad should not imply independent editorial selection. If a ranking is based on a survey, the survey's sample and date belong in the small print, not nowhere.

Comparison claims against named competitors need evidence that is current and like for like. A rating comparison between your product and a rival's is a head-to-head claim, and the ASA expects the basis to be clear.

Finally, check the aggregate against the source before every campaign flight. Rating drift is the most common cause of a claim that was compliant when written and is not when it runs. Our piece on starting with the displayed rating covers the measurement side of that drift.

Naming the platform, the sample and the monitoring record

UK marketers rarely work with a single rating. Trustpilot, Google Reviews, Feefo and Reviews.io each hold a separate score, built on a different sample of reviewers and governed by that platform's own review policy. A reputation-management dashboard that pulls those feeds together is useful for spotting a rating that has moved, but the claim that goes on the page still has to name the source it came from.

An unlabelled 4.8 reads as a rating across the whole market. A 4.8 on Trustpilot reads as a Trustpilot score. When you quote one, record the platform, the number of reviews behind the figure, the date you exported it and where the screenshot or export is filed. That record is what makes the claim defensible months later, and it is the trigger to recheck the figure before the next flight.

How reputation-management software supports the check

The judgement is a marketer's, but the evidence behind it comes out of software. Three jobs matter for a review-led claim: watching the rating drift, exporting the figure with its sample and date, and versioning the wording that goes on the page.

Monitoring ratings across platforms

The Google figure is visible in Google Business Profile, and Trustpilot shows its own score, both without a paid reputation suite. Multi-platform tools such as Birdeye, ReviewTrackers and Yext pull the scores from several platforms into one view, and they are paid subscriptions. A fall from 4.9 to 4.6 shows up there first, which is the trigger to recheck the claim before the page is republished. An alert is a prompt to look at the position, never evidence that the claim still holds.

Exporting the evidence

When you quote a rating, export the page that carries it: the platform name, the review count, the date and the URL. A dated export or report filed with the creative it supports is what lets the same figure be produced months later when a complaint arrives, and it is the file the step-by-step check above asks for. A screenshot with no date proves very little.

Versioning the claim

Copy changes more often than evidence does. Keep a claims register that pairs each asset with the wording, the platform and sample behind it, and the date of the last check. Reputation-management software holds the rating history; the register holds the decision. When a rating moves, or a product version stops being sold, the register shows which pages need an edit before the next flight.

None of this decides whether a claim breaks the Code. That stays with the marketer and, where the evidence is thin, with Copy Advice. The software supplies the record that makes the answer defensible.

Copy Advice and pre-clearance for review-led advertising

The ASA offers Copy Advice, a free and confidential service that checks ads before publication. You send the draft, they tell you whether it is likely to break the Code, and you get an answer before the money is spent. The service is described at Copy Advice - ASA | CAP.

It is not a clearance. Copy Advice is non-binding, and the ASA can still rule against an ad it previously advised on if the facts change or the final creative differs from the draft. Treat it as a strong signal, not a guarantee.

Turnaround is usually a few working days for standard requests. Plan for it in the campaign calendar rather than treating it as an emergency step the week before launch.

What to send: the final copy, the evidence for each claim, the platform and date for any rating, and a note on what the ad is trying to say. Vague submissions get vague answers.

When to use it: any claim where the evidence is thin, any rating used without a named platform, any testimonial from a customer whose result is unusual, and any comparison against a named competitor.

What it is not for: it will not settle a contract dispute, and it does not cover data protection questions, which belong with the ICO.

A short pre-flight checklist

Sanctions, non-compliant advertiser lists and Trading Standards referrals

The ASA has no fining power. Its sanctions escalate, and the published list is the one that matters commercially. The current set of sanctions is set out at Sanctions - ASA | CAP.

The Trading Standards referral route is the one with teeth. Trading Standards sits inside local authority regulatory services. A referral can lead to action under consumer protection law, including the Consumer Protection from Unfair Trading Regulations 2008 and the newer powers in the Digital Markets, Competition and Consumers Act 2024.

Persistent non-compliance also affects media buying. Ad Alerts tell publishers and platforms that an advertiser is non-compliant, and some will refuse the booking. That is a commercial sanction even without a fine.

For agencies, the risk is reputational as much as regulatory. A client named on the sanctions list creates a paper trail that procurement teams find during tenders.

The practical defence is boring: keep the evidence file, date every claim, and recheck ratings before each flight. Most sanctions follow repeat complaints about the same claim, not a single mistake.

Reading ASA rulings as precedent for your own claims

Rulings are the closest thing to case law in UK ad regulation. They show how the ASA has applied the Code to specific facts, and they are the best evidence screen available before you publish. The archive is at Rulings - ASA | CAP.

The rulings to study are the ASA's decisions on testimonials claims where an advertiser used customer statements without holding evidence that the customers existed or had used the service. In those cases the ASA upheld the complaint under the testimonial rules, and the ad had to be withdrawn.

The lesson is not the individual case; it is that the burden of proof sits with you, not the complainant.

Another recurring pattern involves ratings presented without a platform or sample. Rulings in this family usually turn on whether the ad gave a misleading impression of how representative the rating was, not on whether the number itself was wrong.

Search the archive by sector and by keyword before you write copy. If a competitor has already been ruled against for a similar claim, that is a signal about how your version will be read.

Keep a short internal note of the rulings you relied on, with the claim they relate to. If a complaint arrives, that note shows the claim was made on a considered basis rather than copied from a competitor's page.

Common questions