Dynamic Targeting and Dynamic Creative in Pharma: What Is Allowed

Dynamic creative optimization (DCO) came from retail, where a platform can assemble an ad from a product feed and test thousands of variants. Pharma cannot work that way. Each claim is reviewed, each version carries risk information, and OPDP judges the ad the viewer actually saw. That still leaves a lot of room. Most of the value of dynamic approaches in pharma comes from timing and relevance, which can be built from a small set of approved pieces.

Dynamic targeting pharma teams actually use

Dynamic targeting means the bid decision responds to a signal that changes in real time. Common pharma signals:

• Environmental data: pollen counts, air quality, temperature, humidity. Used by allergy, asthma, migraine, and dermatology brands.
• Geography: DMA or ZIP code, often to match formulary wins, field coverage, or regional disease patterns.
• Time: daypart, day of week, season, or days since a known event such as a formulary change.
• Context: page content or video category, used to pick a message that fits the surrounding content.
• Audience segment: HCP specialty or a DTC segment, used to pick a message suited to that group.

None of these change the claims in the ad. They change bid intensity and which approved version is shown. That distinction is what makes the approach workable under MLR. For how each signal fits into a broader targeting plan, see the targeting options comparison.

How dynamic creative works with pre-approved modular content

A pharma DCO setup is a closed system. The brand builds a library of approved modules: headlines, images, calls to action, backgrounds, and the ISI treatment for each ad size. A rules engine in the creative platform picks a combination based on the signal. Nothing is written or generated at serve time.

A workable structure for a display campaign might be three headlines, two images, two calls to action, and one ISI treatment per size. That is 12 combinations per size (3 x 2 x 2). Across four sizes, 48 renders. That is reviewable. Ten headlines, six images, and four CTAs is 240 combinations per size, and review will stall.

What MLR needs to approve combinations

1. The module library, each piece with its own approval reference.
2. The decision rules: which signal triggers which combination, written in plain language.
3. Rendered output for every combination and size, or a matrix that shows all of them.
4. The ISI and fair balance treatment for each size, including how it behaves in small formats.
5. The fallback creative served when the signal is missing or the feed fails.
6. A change process: what happens when one module is updated.

Teams that send only the modules usually get the submission bounced, because reviewers need to judge the ad as served. More on speeding up review in MLR review for programmatic media.

A hypothetical allergy example

Imagine a branded prescription allergy product (hypothetical) running display and online video in 40 DMAs during spring. The team sets up a pollen feed and three states per market: low, moderate, and high pollen.

Each version carries the same approved efficacy claim and the full ISI. The pollen signal changes timing and emphasis, not what the drug does. If the brand had a boxed warning, reminder-style versions would not be an option, and every version would need the full risk treatment. For measurement, the cleanest read is to hold out a set of matched DMAs from the dynamic rules (they run the static base creative at the same budget) and compare outcomes. That tests whether the dynamic layer added anything beyond spending more on high-pollen days.

Guardrails that keep dynamic creative compliant

• No implied claims from the trigger. "High pollen today? Ask about relief" is fine if approved. A version that implies the drug works faster on high pollen days is a new claim.
• Consistent fair balance across versions. Risk information cannot shrink in the versions with more aggressive headlines.
• No sensitive personal triggers. Weather and geography are low risk. Triggers built on inferred diagnosis or health behavior raise state privacy and platform policy questions.
• No generated copy. Generative AI tools that write or rewrite headlines at serve time do not fit a review model built on approving final ads.
• Logged decisions. Keep a record of which combination served where and when, in case a question comes later.

The fair balance in digital pharma ads article covers how risk presentation holds up in small formats, which is where DCO setups most often break.

Where dynamic approaches fit, and where they do not

Dynamic targeting earns its setup cost when the signal really does change the likelihood that the viewer needs the product right now: seasonal allergies, flu, migraine triggers, regional access changes. For chronic conditions without a timing signal, a simpler plan with two or three static versions and audience-based rotation usually does as well with far less review overhead. I would not build DCO for a brand that cannot name the signal in one sentence. The broader context is in the guide to programmatic advertising in healthcare.

Practical takeaway

Before proposing DCO, write the decision table: signal, threshold, bid change, and creative version, with the total combination count at the bottom. If the count is under about 50 renders and every row maps to an approved claim, take it to MLR as a single submission with the fallback creative included. If not, cut modules until it is.

Editorial note. Analysis and frameworks are the author's own and do not represent Acxiom or any current or former employer, client, or named platform. Examples labeled hypothetical or illustrative are not results from real campaigns. Nothing here is legal, regulatory, or medical advice.