4/4+150 XP

Running a pre-launch compliance check before a campaign goes live

Ten working days out, a Type 2 diabetes HCP campaign exists as four separate things: eight LinkedIn sponsored-content variants, a landing page, an annotated reference pack, and a targeting sheet nobody has read closely. The creative is the part everyone reviews. The targeting sheet is the part that gets campaigns pulled mid-flight.

This lesson follows one campaign through that gate. Take the sponsor to be a company with the kind of US Type 2 diabetes portfolio AstraZeneca or Eli Lilly carries: several branded products, one shared prescriber audience, an in-house medical affairs function, and a media agency doing the buying.

Who signs, and in what order

MLR (Medical, Legal, Regulatory) is a routing sequence, not a single meeting. Each function guards a different thing:

Medical checks that every claim matches the approved label and the cited trial data, and that the reference pack actually supports what the headline says.
Regulatory checks the indication, label consistency, internal SOPs (Standard Operating Procedures), and puts an expiry date on the approved piece.
Legal looks at liability, competitor claims, and the contracts sitting under the asset (agency, platform, any third-party voice).

Sequence matters more than most marketers expect. Send a finished eight-variant carousel into review before the claim set has cleared medical and a single failed claim multiplies into eight rejections, eight resubmissions and eight new version numbers. Clear the claims first, then build.

Most large sponsors route this in Veeva Vault PromoMats (Veeva sells the review and promotional-asset software described here). Every approved piece leaves with a job code and an expiry date. When the label is updated or a cited reference expires, everything downstream of it has to be recertified or withdrawn, including live ads.

The FDA's OPDP enforcement letter archive is a reasonable calibration tool for what your reviewers are guarding against. Note the volume: enforcement letters run to a handful a year across the whole industry. The binding constraint on your launch date is your own MLR queue, not the agency.

What the reference pack has to prove

Prominence rules for risk information are the ones the patient-facing lesson sets out. At the gate the question is operational: which file carries that information, and does the platform render the file intact?

Two things reviewers push back on hardest:

Substantiation depth. Every factual statement traces to a page and a table. A twelve-word banner headline routinely sits on a reference pack of dozens of pages, because there is no room in the asset to carry the qualifying context that a printed detail aid can. Digital assets carry more references per word than almost anything else in the portfolio.

Source quality. A claim resting on "data on file" rather than a published trial is a weaker footing, and many reviewers will not accept it for paid external media at all. Indirect comparisons are the other reliable rejection: if the copy says "more effective than" a named competitor, medical wants a head-to-head trial, not two separate studies read side by side.

Staying inside the approved label is the non-negotiable one. Several billion-dollar US settlements have come out of off-label promotion allegations, which is why an image implying use in adolescents, when the indication is adults, gets escalated even when the copy is clean.

Walking the gate, step by step

Step 1: Claim-by-claim mapping. The headline promises an average HbA1c reduction. Medical asks for the exact trial table, the population, the timepoint and the comparator arm, and logs the citation against the asset in Vault.

Step 2: Indication and population. Copy, images and any implied patient in the visual have to sit inside the approved indication. Stock photography is a frequent offender.

Step 3: Rendering, not just layout. LinkedIn truncates sponsored-content intro text behind "see more" on mobile. If the safety information lives below that fold, the asset as designed and the asset as served are two different pieces. Fix is either to carry the information in the image itself or above the truncation point, and to make the link to full prescribing information a single click with no interstitial.

Step 4: Targeting settings as a regulated surface. This is the step that skips review most often, because it lives in the media plan rather than the creative file:

Job title and function targeting is self-reported. A "physician" audience on LinkedIn includes medical writers, students, industry staff and retired members. If the asset is only defensible in front of prescribers, gate it behind a credential check on the landing page rather than trusting the platform audience.
Audience expansion and lookalike features quietly widen delivery beyond the audience you had approved. Off.
Off-platform placements (the LinkedIn Audience Network) put an approved asset next to context you did not review. Off, unless someone signed for it.
Geo-targeting is inference-based. Exclude countries explicitly rather than only including one. Which markets permit which kind of promotion is the geography lesson's subject; the job here is making the platform settings match the fork the brand plan already chose.
Comments. Enabling them on a promoted post creates a pharmacovigilance surface: any mention of a suspected side effect has to be routed into the safety database, and serious unexpected events feed a regulatory reporting clock (15 days in the US for alert reports). Either disable comments or staff daily monitoring under a written SOP with named owners and cover for holidays.

Step 5: Archive and expiry. Legal confirms the final served files, not the design files, are archived per retention rules, and that the expiry date is in the media calendar so the flight ends before the approval does.

Where campaigns actually get stuck

The recurring objection categories are stable year to year: overstated efficacy, minimised or omitted risk information, unsubstantiated superiority, and an implied patient population broader than the label.

The operational failures are different, and more expensive:

Version drift. The agency resizes the approved 1200 x 627 single-image file into a square placement and the safety text reflows or crops. The resized file is a new piece. It needs its own approval, and it usually does not have one.
A broken chain. The landing page is recertified, moved or expires while the ads are still running, so live creative points at a page that no longer carries the safety information. Link checks belong in the weekly flight review, not only at launch.
Cost of a mid-flight pull. Committed media spend is forfeited, the re-review of eight variants absorbs more internal time than the original build, and the brand loses the launch window it was buying.

A pattern worth internalising: if a claim feels like the strongest possible version of the truth, it is the version MLR flags first. The safest claims are literal restatements of trial data.

A simple pre-submission gut check

Before an asset reaches the formal queue, self-screen with four questions:

Can I point to the exact page and table for every claim, including the ones in the image?
Does the asset as the platform renders it, truncated and on mobile, still carry what the approved layout carries?
Would this claim survive being read aloud next to the approved label with no additional context?
Do the targeting settings match the jurisdiction and audience we actually got signed off, with expansion features off and exclusions set?

Building the gate into production

Teams that hit launch dates treat MLR as a filter inside creative development rather than an inspection at the end.

Pre-clear claim language with medical affairs before production starts. Redesigning a finished asset because the headline fails substantiation burns budget and calendar at the same time.
Maintain a claims library of cleared statements creative can draw from, so the same underlying fact is not re-argued every quarter.
Get modular kits approved: locked risk-information blocks and pre-cleared crops for each placement, so a resize does not create an unapproved piece.
Run a mock review internally for first-of-kind formats, where no precedent exists in the library, and budget review rounds in weeks rather than days when you do.

🎬 [VIDEO: "FDA Regulation of Prescription Drug Advertising" - youtube.com - search for FDA OPDP or industry compliance channel explainers on how drug ad claims are reviewed and enforced, useful for seeing real Warning Letter examples discussed]

Key Takeaways

MLR review is a routing sequence with an order: clear the claim set with medical before creative production, or every rejection multiplies by the number of variants.
Approved pieces carry a job code and an expiry date, and the flight has to end before the approval does; label updates invalidate everything downstream.
Targeting settings are part of the compliance package. Audience expansion, off-platform placements and loose geo-targeting can push an approved asset outside the audience it was approved for.
The asset the platform serves, truncated on mobile, is the asset that gets judged, not the layout that was signed.
Enabling comments on promoted posts creates an adverse-event reporting obligation with a regulatory clock attached; decide who monitors it before the campaign goes live.

Fair-balance and consumer-protection rules in patient-facing content