Why the Traditional Healthcare Agency Model Is Breaking
Why the old agency model is reaching the end of its useful life in life sciences content production, and what replaces it. For fifty years, healthcare agencies have sold the same thing: people. They bill hours, mark them up and grow by adding headcount. That model built some of the largest marketing networks in the world. It still dominates life sciences, where the top 100 North American healthcare agencies earned more than US$11 billion in fees in 2023. But for regulated content production, the model is reaching the end of its useful life.

The Agency After Headcount
The old model sells time
The traditional agency estimate is a stack of hours. Junior staff might be billed at US$100 to 150 an hour, while principals can be billed at US$300 to 500. Revenue grows with headcount, and the margin comes from the difference between what staff cost and what the client is charged.
Regulated content makes this model particularly profitable. Every promotional asset must pass medical, legal and regulatory review. Each review round creates more agency hours. Localisation, adaptation and channel variants create more again.
This model still provides strategic thinking, brand stewardship, creative judgement and trusted client relationships. Those are real services. The narrower problem is the production of the content itself, which accounts for much of the time and much of the invoice.
The scarce input has changed
Clients need more content than ever. Content volume on the leading life sciences platform doubled between 2019 and 2022.
But the labour sold by the agency is no longer the scarce input.
AI production tools can now help with drafting, layout, versioning, reference checking and the first compliance pass. The scarce input is becoming senior judgement applied at the right moments, together with brand knowledge encoded well enough for the tools to use it.
This is already changing client expectations. Novo Nordisk reported increasing annual content production from 5,000 to 31,000 assets while cutting asset creation time from 104 days to 23. Pfizer built an internal generative platform with the aim of multiplying its content output. Offshore production companies are growing, and platforms such as Veeva are moving further into automated review.
Clients have also started to question the value of producing so much material. Published field data covering more than 600 million healthcare professional interactions suggests that about 80 per cent of approved pharmaceutical content is rarely or never used.
Once a commercial leader has seen that figure, the question changes from “What does this cost?” to “Why are we making it?”
The new model sells the deliverable
The new model sells a finished, review ready deliverable at a fixed price. The client buys the asset it needs and knows the price before work starts.
Each brand begins with an onboarding step that captures its guidelines, approved claims and regulatory requirements in a knowledge bank. Every later project is produced using that knowledge. The tenth asset should be cheaper and safer to produce than the first.
A proof of concept is delivered early, when a misunderstanding costs very little to correct. Compliance documentation is built into the deliverable, including the claims matrix, annotated references, safety information placement and change log.
The new model does not replace the client's medical writers, reviewers or regulatory team. It should make their work shorter by giving them a medically compliant draft with the supporting material attached.
On one observed flyer project, the conventional agency used an estimated 240 staff hours and billed US$60,000. Medware produced the project with 15 human hours at an introductory price of US$10,000. That is one data point, and it needs to be tested across more projects, but it shows how different the cost structure can be.
What changes for the client
The invoice changes first. Fixed prices remove the estimate, change order and quarterly reconciliation. A brand manager can plan content against a catalogue rather than a rate card.
The better measures also change. Instead of focusing only on cost per asset and agency responsiveness, clients can measure:
- cost per used asset
- time to a review ready draft
- first cycle approval rate
- reuse across channels and markets
These measures expose waste that the old model can hide.
What does not change
The regulator does not change. Every piece still needs medical, legal and regulatory review, and medical accountability stays with the people the regulator expects to hold it.
Brand strategy does not change. Someone still needs to decide what the brand should say. An AI production model is a poor place to make that decision.
Relationships do not change either. Pharmaceutical companies buy from people they trust, and a fixed price does not create trust on its own.
The transition has already started
The likely transition will happen one brand at a time. The new model will win the production layer first, while established agencies retain strategy, media and parts of the client relationship. For a period, clients will use both.
The agencies, offshore hubs, platforms and consultancies now all promise savings from AI. The question for clients is no longer who promises the largest saving. It is whose saving appears on the invoice.
The agency after headcount is not an agency without people. It is an agency where people do only the work that needs them, and where the client pays for the outcome rather than the pyramid.

