Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Marketing Operations › Troubleshooting

Legal Keeps Rejecting Marketing Work

Why legal or compliance keeps sending finished marketing back: claims not flagged at intake, no claims library, edited wording, endorsements and AI-made copy.
Troubleshootingintermediate12 min readUpdated

Take the last few pieces the reviewer sent back and record two things for each. The first is the reason. The campaign workflow guide asks each gate to tag every return with one reason from an agreed short list (the guide's examples are off-brief, an inaccurate claim, a brand standard, a legal point or a technical fault), which is what makes rework countable; for legal returns, split the reason further into a new or changed claim, a price or offer, an endorsement or testimonial, generated content and a translated line. The second is the stage at which the reviewer first saw the piece: at intake, at the brief, during build or only when it was finished. If the returns carry no reasons, that is the first finding, because a gate that returns work without them teaches the team nothing.

If most returns were first seen after build, the problem is routing: claims reach the reviewer too late to shape them. If they were seen early and still came back, the wording, the evidence or the disclosure arrived incomplete.

The theory of constraints explains why this matters. Legal review is often the constrained step in a marketing route, and the framework's own example of getting more from it without new spending is a pre-approved claims library, so the reviewer looks only at what is new. Every return spends that scarce time twice.

This page does not judge whether a reviewer is right, and nothing here is legal advice. A single asset that needs sign-off today goes through the creative approval SOP. An ad turned down by a platform rather than by your own reviewer is a different diagnosis, covered for generated creative by AI-generated ad creative being rejected. And if the complaint is waiting rather than returns, with work queued for the same reviewers, the diagnosis for campaigns that keep launching late examines the queue.

Diagnosis table

Diagnosis

SymptomCheckLikely causeFix
Claims, prices or testimonials first reach the reviewer in built workWhether the intake form asks about claims, prices, endorsements and regulated productsRisky content not flagged at intakeAdd the questions; a yes routes the copy to legal before build
The reviewed version is not the one about to launchWhether build started from a signed brief, and what changed after reviewBuild before the offer and message were agreedNo build without a signed brief; changes go back through the signature
The same claim is argued afresh each time, in slightly different wordsWhether a substantiation file or claims library exists and builders use itNo claims libraryBuild the library and take wording from it, never from memory
A claim approved before comes back rejectedIts wording, market and evidence date against the approvalApproved wording edited, moved to another market or past its recheck dateRecord conditions, markets and recheck dates; treat changes as new claims
Reviews, testimonials or creator posts come back for disclosureWhether the brief carried the disclosure requirementEndorsement work briefed without disclosureWrite the disclosure into every brief and check it at review
Generated copy carries claims, reviews or people nobody can sourceGenerated claims against the library, and quotes against their sourceAI drafting adds unsupported contentCheck generated copy against the file before it reaches the reviewer
A translated piece comes back for a claim, price or guaranteeWhether claim lines were back-translated before reviewLocalised copy shifted a claim's meaningBack-translate every claim, price, offer and legal line first

Fix, verify and prevent each cause

New claims, prices and testimonials are not flagged at intake

The approvals and compliance guide names reviewing finished work as a common mistake, because legal then meets claims only once the design exists and the work has to be made again. Its remedy is an approval matrix applied at intake: the form asks four questions (any new claim? any price or promotion? any testimonial or endorsement? any regulated product?), and each yes sends the copy to legal while it is still words, before anyone designs around it. The principle behind those rows is that marketing must not mislead and claims must be capable of substantiation before they are made; the approvals guide points to the Advertising Standards Authority for the UK and to the Federal Trade Commission's business guidance for the US.

Fix: add the four questions to the intake form and route every yes to the reviewer with the copy and its evidence before build starts.

Verify: returns over the next period show the reviewer first seeing flagged work at the brief, not after build.

Prevent: keep each question tied to something observable in the work, so requesters can answer it without a judgement call.

Build starts before the offer and message are agreed

Some returns come from work that changed after the reviewer saw it. The offer moved, a line was strengthened in production, and the reviewed version is no longer the one about to go live. The campaign workflow guide lets build begin only once both the requester and the campaign owner have signed the brief, treats any later change as needing those signatures again, and puts the offer or claim on the intake form so it is visible from the start.

Fix: let nothing reach design until the brief carrying the offer and the reviewed claim wording has both signatures, and send any later edit to a claim to the reviewer as well as back for re-signature.

Verify: the version on the approval record matches the version at the launch check for the next few pieces.

Prevent: keep a "changed after approval" reason on the return list and look at its count at each quarterly read.

No substantiation file or claims library

When every claim is argued from nothing, the reviewer rebuilds the evidence each time and builders write the same benefit in a new way for every asset. The tutorial on building a claims library keeps every approved claim word for word beside its evidence, its conditions, the markets it may run in and who approved it; routine work borrows those words and skips a fresh review, and legal sees only wording that is new or bolder. A claims library also stops the quiet drift the tutorial describes, where builders paraphrase approved claims from memory.

Fix: collect the claims already live, find and date the evidence for each objective claim, write each entry in its exact wording and have the named reviewer approve the entries in batches by product or theme.

Verify: sample recent work; every claim should match an entry or have gone to the reviewer as new.

Prevent: give builders read access and link the library from the copy and creative templates, so the approved words are the easiest ones to use.

Approved wording is edited, moved to another market or kept after its evidence lapsed

A claim the reviewer once approved can still come back. The claims library tutorial treats any change to wording as a new claim, approves each claim for named markets and channels, and gives its evidence a recheck date, because comparisons, statistics and certifications age. Without those fields, a lightly edited line, a claim carried into a new market or one kept after its evidence lapsed looks approved and is not.

Fix: add conditions, markets and a recheck date to every entry, and when a claim stops holding, retire it and take it down everywhere it runs with the tutorial on withdrawing an asset from every channel. Partners often keep quoting retired claims, and the tutorial on checking affiliate content checks their pages against the library.

Verify: no entry in live use is past its recheck date, and the next returns no longer cite a claim the library already holds.

Prevent: run the claims audit the approvals guide names, using the recheck dates as its schedule.

Reviews, testimonials and creator posts are briefed without the disclosure requirement

Endorsement work comes back when the disclosure was never asked for. The approvals guide puts the disclosure requirement into the brief of any creator or testimonial work and has the reviewer look for it. The influencer contracts and disclosure guide sets out the general direction regulators enforce, without giving legal advice: where a creator has received something of value from the brand, the audience should be able to tell the content is advertising. The FTC's endorsement guidance, as that guide summarises it, asks that any material connection linking the person endorsing to the brand is made plain to the audience, and that endorsements are honest. Quoted customers get the same care in the claims library tutorial: each testimonial's entry notes where it came from, any relationship behind it and the disclosure it needs.

Fix: write the required disclosure into every creator brief and every brief that quotes a customer, record the source and the relationship behind each testimonial in the library, and require the agreed wording in creator contracts.

Verify: the next endorsement pieces reach the reviewer with the disclosure already in place.

Prevent: add disclosure to the brief template itself, so nobody has to remember it.

AI-drafted copy or images add claims, reviews or synthetic people

Generated drafts produce confident superlatives, outcome claims and customer-sounding quotes in seconds, and none arrives with evidence. The guide to AI claims and consumer protection is clear that switching tools does not lower the standard: claims need evidence before publication whoever wrote them, a generated review is a fake review, and a synthetic "customer" describing results is a fabricated endorsement.

Fix: check generated copy against the claims library before it reaches the reviewer; matching claims pass, new or stronger ones go back for evidence or are softened, quotes are replaced with genuine reviews used with permission, and synthetic people are labelled and never presented as customers or experts.

Verify: returns for generated content fall, and the editor's claim check is recorded on each piece.

Prevent: put the off-limits practices, such as polished reviews and composite testimonials, in the team's AI use policy.

Localised copy shifts a claim's meaning in a language the approver cannot read

A translated page can promise more than its source without anyone in the central team noticing. The tutorial on reviewing localised copy in a language you do not speak tests meaning through a back-translation of every claim, price, offer, guarantee and legal line, made by a translator who never saw the source, and compares each with the source line: does it promise the same thing? Differences go to whoever owns compliance in that market and are checked against the claims library.

Fix: add the back-translation of risky lines to the localisation workflow before any localised piece reaches its compliance owner.

Verify: the next localised returns, if any, concern market rules rather than lines that drifted from the source.

Prevent: record approved translations of each claim in the library against the market they were approved for.

Prevention: a claims question at intake, a library entry for every approved claim and a quarterly read of returns

The causes above share a pattern: the reviewer decides too late, or decides the same thing again. Keep the claims question on the intake form and route every yes before build. Make every approval add to the library: when the reviewer approves new wording, it becomes an entry with its evidence, conditions, markets and recheck date, so the next piece that needs it skips review. Over time the work reaching the reviewer shrinks to what is genuinely new.

Read the returns together each quarter, as a starting rhythm, with the reviewer in the room. Count them by reason and by the stage at which the reviewer first saw the work, and compare with the last reading. A reason that keeps returning names a missing habit; a stage that keeps appearing names a routing gap. The approvals guide suggests reporting return rates by gate beside how much work carries a complete approval record, as signs of governance health rather than a measure that rewards reviewers for speed. When the same kind of return comes back after it was fixed, the diagnosis for marketing mistakes that keep happening traces why the fix did not hold.

Between readings, two audits keep the arrangements honest. The claims audit, run from the library's recheck dates, finds evidence that has aged before a reviewer or a customer does. The approval audit samples live work and checks that it either used library wording or went to review, which shows whether the shortcut is being used as intended.

Finally, give the reviewer their part. Agree the list of return reasons with them, ask that every return carries one with a note of what would make the piece acceptable, and tell them at kick-off which pieces are coming and what they carry. A reviewer who sees new claims early, with evidence attached, has less to return. If returns are only one of several operations problems, such as missed launch dates or errors that come back, start from the marketing operations troubleshooting hub, which lists every diagnosis in this collection.

A mattress maker cuts the work its reviewer sends back (illustrative)

Illustrative example — not a real result

Vemblorick Mattresses

Vemblorick Mattresses, an invented mattress maker selling online, has a marketing team of 5 and one compliance reviewer shared with the product team. In one quarter the reviewer returned 34 of the 61 campaign pages, ads and emails sent for review. The operations lead sorted the 34 by the reason given and by the point at which each one first reached the reviewer.

Three patterns covered 29 of them. 18 carried comfort and sleep claims that each designer had written afresh, such as "deeper sleep" on one ad and "sleep better every night" on another, none taken from approved wording. 10 quoted customer reviews collected with a free pillow offered for each review, with no mention of the pillow. 1 translated product page offered a "lifetime guarantee" where the source promised 10 years. And 29 of the 34 had reached the reviewer only after build, because the intake form never asked whether a piece carried a claim.

The maker added the claims question to intake, built a library of 41 approved claims with their evidence and recheck dates, wrote the disclosure line into its review-request emails and creator briefs, and had every claim, price and guarantee line in localised copy back-translated before review. The next quarter the reviewer returned 7 of 58 pieces, 5 of them new claims seen at the brief stage rather than in finished work.

Common mistakes

  • Sending finished work as the first sight of a claim. The build is redone when the claim changes; route claims at intake.
  • Treating the reviewer as the problem. Returns usually reflect what the reviewer was sent, and when.
  • Rewording approved lines by hand. Even a one-word edit needs the reviewer again; copy from the library.
  • Carrying a claim into a new market unchanged. Approval is for named markets; check before reuse.
  • Quoting reviews without the relationship behind them. Record the source and any incentive, and disclose it.
  • Letting generated copy reach the reviewer unchecked. Check it against the library first.
  • Approving translations by eye. Back-translate the claim lines; fluent copy hides drift.
  • Returning work without reasons. Nobody can fix a pattern they cannot see.

Frequently asked questions

Isn't it legal's job to make our copy compliant?

The reviewer decides whether a claim, price or endorsement can run; that decision stays theirs. What the marketing team owns is bringing work that is ready to be decided: flagged at intake, built from an agreed brief, using approved wording or carrying the evidence for anything new, and with the disclosure written into the brief. When the reviewer has to rewrite copy, rebuild evidence or discover a testimonial in finished work, review turns into production, and both sides lose time. Arranging the work so the decision is quick is an operations job, not a legal one.

Can work that only uses approved claims skip legal review?

Where the approval matrix says so, yes. Routine work that copies a library entry word for word, used only where the entry allows, can pass with brand review and the launch check alone. Changing the wording at all makes it a new claim that returns to the reviewer, as does a claim carried into a market it was not approved for. Prices, promotions, testimonials and regulated products follow their own rows in the matrix whether or not the claims are already approved.

Our reviewer sends work back without saying why. What can we do?

Agree a short list of return reasons with the reviewer and ask that every return carries one, alongside a note of what would make the piece acceptable. Returns without reasons cannot be sorted, so the team keeps repeating whatever caused them. Read the reasons together from time to time: when one reason dominates, it usually points at a missing habit on the marketing side, such as a claim that never reached intake or a disclosure missing from the brief, rather than at the reviewer's judgement.

Terms used on this page

Frameworks

Sources

  1. Advertising Standards Authority — ASA
  2. FTC business guidance — Federal Trade Commission
  3. FTC Endorsement Guides: What People Are Asking — Federal Trade Commission

Related