UK ad creative software: a CAP risk review workflow

What to take away

How CAP Code risk areas shape a creative review

For UK ad creative software, the CAP Code is the rulebook to check against. It is written by the Committee of Advertising Practice and enforced by the Advertising Standards Authority across non-broadcast media, including online display, social, influencer, email, print and outdoor.

Broadcast ads follow the BCAP Code, cleared separately by Clearcast for television and Radiocentre for radio. A creative tool works on the CAP side, which is where most UK brand spend now sits.

The Advertising codes - ASA | CAP page sets out the sections a tool should map to. Those include misleading advertising, comparisons, food and soft drink, gambling and alcohol. Children, health and beauty, and privacy sit alongside them. Each section has its own tests, and the tests differ.

A claim about a savings rate is judged on substantiation. A gambling ad is judged on targeting and imagery. A soft drink ad is judged on nutrient profiling.

That split matters for software design. A single "compliance score" hides which rule is at risk. A better model runs separate checks per rule cluster and reports them separately, so a creative lead in Manchester or Glasgow can see whether the problem is a claim, a placement or a model's age.

Misleading claims sit at the centre because they cut across every sector. The CAP Code requires that claims be truthful, substantiated and not exaggerate what a product can do. Comparative claims need like-for-like evidence. Environmental claims need a full life-cycle basis. Price claims need the reference price to be genuine and recent.

The other three clusters are sector rules layered on top. HFSS food rules restrict where and how less healthy food and drink can be advertised to children. Gambling rules restrict content, targeting and the use of certain imagery and offers.

Alcohol rules restrict where ads can appear, who they can reach, and what they can imply about success, confidence or social acceptance.

A UK creative team should treat these as four separate libraries in the tool, not one. The UK advertising, privacy and copyright rules map shows where each library sits alongside data protection and clearance duties.

What a creative tool can flag before submission, and what it cannot

A tool can flag text, images, audio transcripts, targeting parameters and landing page content. It can catch a missing qualifier, an absolute claim, a price without a was/now basis, an HFSS product shown to a child-skewed audience, and a gambling ad missing its responsible gambling message. Alcohol ads with strong appeal to under-18s are caught too.

It cannot make a final legal judgement. It cannot verify that a substantiation file exists in a brand's evidence locker. It cannot see the full media plan unless the plan is fed in. It cannot clear a broadcast ad, and it cannot decide whether an influencer's audience is genuinely 75% adult.

That boundary is the honest selling point. Software narrows the review queue and catches the repeat offenders. A human reviewer still signs off, and for TV and radio the clearance bodies still rule. The checks that decide whether it runs are a useful framing for where the tool stops and the reviewer starts.

A practical setup uses three layers. First, automated flags on the creative asset. Second, a human review against the flagged rules. Third, a pre-submission check against Copy Advice for anything novel or borderline. The tool's job is to make layer two fast and layer three rare.

Some checks are cheap and reliable. Keyword and phrase detection for absolute claims ("best", "cures", "risk-free") is mature. Logo and disclaimer detection is mature. Age-gate and targeting checks are mature if the platform data is available. Nutrient profiling is harder because it needs product data, not just the ad.

Other checks are weak. Sarcasm, implied claims and visual exaggeration are hard to detect. A tool may miss a claim made by a presenter's tone or by a background prop. It may also over-flag legitimate puffery, which trains reviewers to ignore warnings. Tune thresholds and log false positives.

A sensible rule for buyers: ask the vendor to show the rule library, the update cadence and the audit trail. If the library is not versioned against CAP Code sections, it will drift. If the audit trail does not record who overrode a flag and why, it will not survive a challenge.

Illustrative risk patterns and possible software flags

The Rulings - ASA | CAP database is the best training set a UK creative team has. It is free, searchable and indexed by sector, so it can be used to build prompts and test cases. The table below groups recurring themes with the kind of flag a tool could raise; it does not summarise named rulings.

These are recurring patterns, not one-off slips. A tool that scans for the right-hand column will catch most of them. A tool that only scans for banned words will miss the price and nutrition cases, which depend on context.

Take the savings pattern. The ad shows a was/now price. The tool should check that the was price is a real price the product was sold at, for a reasonable period, and that the saving is calculated correctly. If the brand cannot evidence the was price, the claim fails regardless of how the ad looks.

Take the health halo pattern. A cereal bar described as "wholesome" or "natural energy" may still be HFSS. The CAP Code's food rules look at the product, not the copy. The tool needs product data to flag this, so integrate it with the brand's product information file.

Take the gambling bonus pattern. A sign-up offer is not banned, but the significant conditions must be prominent. The tool should check that the terms appear in the ad, not only on a landing page, and that the text is legible at the size it will run.

A good use of the rulings database is to build a test suite. Take twenty rulings per sector, run the creatives through the tool, and measure how many flags fire. That gives a defensible accuracy figure and a list of gaps to fix.

HFSS food and drink rules: where the creative tool needs a hard stop

HFSS rules are the clearest case for a hard stop because the restriction is about the product, not the wording. Food and drink high in fat, salt or sugar face restrictions on advertising to children across non-broadcast media. The tool cannot soften that. If the product is HFSS and the audience is child-skewed, the ad should not run.

The first job is classification. Is the product HFSS under the UK nutrient profiling model? That is a technical test based on the product's composition per 100g or 100ml, not on the marketing copy. The tool should pull the classification from the brand's product data, not guess from the ad.

The second job is audience. Where will the ad run, and who will see it? A tool can flag a mismatch between an HFSS product and a placement whose audience skews under 16. It can also flag creative treatments that appeal to children: cartoon characters, toys, school settings, child voiceovers.

The third job is the health halo. Words like "natural", "wholesome", "balanced" and "energy" can imply a health benefit that the product does not have. The CAP Code treats that as misleading. The tool should flag health halo terms on HFSS products and require a reviewer to justify them.

A hard stop is better than a warning here. A warning gets overridden under deadline pressure. A hard stop forces the reviewer to record a reason and route the ad to a senior sign-off. That is the difference between a tool that changes behaviour and one that adds clicks.

There is a fourth job: the boundary between HFSS and non-HFSS variants. Brands often run a healthier variant alongside the core product. The tool must keep the product data separate, or it will clear the wrong SKU. This is a data hygiene problem, not a creative one.

For agencies in the West Midlands or Yorkshire working on multiple food brands, the product data feed is the integration that matters most. Without it, HFSS checks are guesswork. With it, they are the most reliable flags in the whole system.

The AdviceOnline Library - Advice for Marketers - ASA | CAP carries the detailed guidance on HFSS, and it should be the source the tool's prompts cite. If a vendor's rule library does not reference AdviceOnline, ask why.

Gambling and alcohol creatives: the extra checks the CAP Code demands

Gambling and alcohol ads carry extra duties because the harm is specific and the audience rules are strict. Both sectors have their own CAP Code sections, and both attract a high volume of ASA rulings.

For gambling, the core checks are targeting, content and terms. The ad must not be targeted at under-18s or at vulnerable groups. It must not suggest that gambling can solve financial problems or be a way to make a living.

It must not link gambling to sexual success or to enhanced personal qualities. Sign-up offers must show significant conditions prominently.

A creative tool can flag most of this. It can check age-gating on the placement, detect problem phrases, and check that bonus terms are present and legible. It can flag imagery that implies a lavish lifestyle funded by winnings. What it cannot do is judge whether a specific audience is vulnerable, which needs human oversight and platform data.

For alcohol, the checks are about audience, content and context. The ad must not be aimed at under-18s, must not encourage excessive drinking, and must not imply that drinking brings confidence, popularity or success. It must not link alcohol to driving, sport or risky behaviour. It must not suggest that alcohol is a necessary part of social occasions.

A tool can flag success and confidence language, detect drinking games and rapid consumption imagery, and check placement against child-skewed audiences. It can also check that the ad does not imply that a drink improves performance or mood in a way that breaches the rules.

Both sectors need a reviewer who understands the rules, not just a reviewer who can read a flag. The tool should route gambling and alcohol creatives to a named specialist in the team. That specialist should have the ruling history for the brand and the sector to hand.

The advertising rules and ethics guide is a useful companion for teams building that specialist review. It sets out the wider duties that sit alongside the CAP Code sections.

One more check applies to both: the responsible gambling message and the alcohol responsibility message must be present where required, and must be legible. A tool can verify presence and size. It cannot verify that the message is understood, but presence is the baseline the ASA checks.

Sanctions and the non-compliant online advertisers list as a feedback loop

The Sanctions - ASA | CAP page sets out what happens after a ruling. Most cases end with a requirement to withdraw or amend the ad, plus a published ruling on the ASA site. That published ruling is itself the sanction for many brands, because it is indexed and permanent.

For persistent offenders, the sanctions escalate. The ASA can place the advertiser on the Non-compliant online advertisers - ASA | CAP list. That list names advertisers whose ads remain non-compliant, and it is used by platforms and networks to take action. It can also lead to paid search ads being pulled.

The list is a feedback loop for software. A tool can monitor it and warn a brand if a competitor or partner appears on it. More usefully, a brand can check its own name and its agency's clients. If a name appears, the creative process has failed somewhere upstream.

The sanctions regime also includes referral to other regulators where relevant, such as the Competition and Markets Authority for certain online advertising issues, or the Information Commissioner's Office where data protection is involved. A tool should flag when an ad touches those areas, even if the CAP Code check passes.

For agency creative leads, the commercial risk is real. A ruling is public, it can be cited by competitors, and it can affect platform access. The cost of a pre-submission check is trivial next to the cost of a campaign being pulled mid-flight.

A practical feedback loop has three steps. First, log every ruling that names your brand or your clients. Second, map the ruling to the rule cluster and the creative element that failed. Third, update the tool's rule library or prompts so the same failure is flagged next time. Without step three, the loop is just a record.

Encoding Copy Advice and AdviceOnline into ad creative software prompts

Copy Advice is the ASA's free pre-publication service for non-broadcast ads. It gives an opinion on a specific ad before it runs. It is not a clearance, and it does not bind the ASA, but it is the closest thing to a safe harbour a UK creative team has.

Encoding this into prompts takes work. The prompts need the rule, the test the ASA applies, the evidence the brand must hold, and the failure examples from rulings. A prompt that says "check for misleading claims" is useless. A prompt that says "flag absolute savings claims unless a genuine prior price is evidenced for a reasonable period" is usable.

A worked example shows the shape. Suppose a soft drink brand wants to run a social ad with the line "the healthy way to refuel". The tool should: classify the product; if HFSS, hard stop on child-skewed placements; flag "healthy" as a health halo term; require product data; and route to a specialist if the claim is retained.

The reviewer then decides whether to substantiate, amend or drop the line.

That example also shows the limit. The tool cannot know whether the product is HFSS until the data is loaded. It cannot know whether the placement skews to children until the media plan is loaded. The prompts are only as good as the inputs, which is why integration matters more than the model.

For teams choosing a tool, the vet ad creative compliance checklist is the place to start. It asks the questions that separate a UK-ready tool from a generic one: rule library versioning, AdviceOnline sourcing, product data integration, audit trail, and specialist routing.

A final point on maintenance. The CAP Code and AdviceOnline change. Rulings accumulate. A rule library that is not updated quarterly will fall behind, and the fall will show up as false negatives. Ask the vendor for an update log, and test it against recent rulings each quarter.

Common questions