You've likely read several speculative articles about the potential of AI in healthcare marketing. This isn't one of those. While we will highlight future opportunities, this Insight takes a practical approach and outlines the steps you can take to get started with pharma marketing AI today.
Are there amazing things to come? Absolutely. But your ability to leverage this potential depends on what you do now. Consequently, this Insight gives you a framework that will help you implement AI in healthcare marketing — detailing how you can start in a way that is safe and provides long-term value.
Note: this Insight refers to topics explained in detail elsewhere on Anthill’s site. In particular, you may want to review our online guide to modular content. That provides a good overview of the topic — explaining how it applies to the healthcare industry. You can also download a PDF version here.
Pharma marketing AI presents a tremendous opportunity to enable omnichannel at scale.
Generative AI enables high-volume content creation, increasing your ability to deliver more personalised HCP engagements.
However, there are risks that AI generates biased or false information — which is unacceptable in healthcare.
The need for accuracy at all times, and the requirement to follow good medical legal review (MLR) practices, means that healthcare companies must handle AI differently than other industries.
Pharma needs to match AI to its established procedures — integrating it thoughtfully into existing workflows and systems.
In this way, you can gain the benefits of pharma marketing AI, extend your ability to deliver omnichannel engagements, and work with confidence — knowing that safe and ethical practices are being followed.
Issues with generative AI and healthcare marketing
The regulated nature of the healthcare sector means that we must apply AI differently to other industries. We can’t, and shouldn’t, cut and paste what works for regulation-light sectors to the life sciences.
AI produces astonishing results by determining the most probable answer. It can be helpful to think of AI less as 'artificial intelligence' but rather as 'applied statistics'. And it’s very good at it. Generative AI can produce content in seconds that is likely exactly what you need. And it can suggest astonishingly insightful answers — solutions you may never have considered.
But AI is not always right. Because it deals with probabilities, AI will also generate content that looks plausible but is incorrect. And that’s unacceptable in healthcare where ‘always being right’ is a necessity — both because we have a duty to ensure that the information we release is correct and because we need to be fully compliant with all regulations. In other words, healthcare communication has to be 100 percent correct every time — not mostly accurate.
This is a problem. AI must never invent plausible-sounding information about healthcare products and services. But we do want to access the potential of machine learning to generate content faster, be more creative, and enable customer-centric omnichannel communication.
What to do?
Integrate with established pharma practices
The solution is to integrate AI with established practices — applying it thoughtfully to existing workflows and systems. To understand how that’s possible, let’s start by defining three overall stages of the healthcare communication supply chain:
commissioning and creation
review and approval
localisation and delivery
This categorisation matters because it enables us to be selective and apply AI differently, depending on where we are in the supply chain. This way, we can get the best of what pharma marketing AI offers while ensuring that all materials are accurate and compliant.
Key concept: AI before and after MLR
At Anthill, we are working with our clients to implement AI around a simple division: before and after MLR. This separation ensures that potentially inaccurate content can never be provided to HCPs.
In other words, you work with generative AI technologies when developing content and then apply different AI technologies later in the content supply chain. This separation ensures that AI is never generating content at the moment that HCPs engage with it.
In this way, the ‘before and after MLR’ structure provides two related uses for pharma marketing AI that we will detail below: content generation and then content selection and assembly.
Before MLR: AI content generation
The first stage — commissioning and creating content — can use generative AI because it occurs before MLR.
A rapidly growing suite of AI tools is available to you and your creative partners. You have probably already tried OpenAI’s ChatGPT and know that it can generate content ideas and copy with just a simple prompt. It can also change the tone of something or simplify complex information to make it understandable to different audiences. You may also have experimented with graphics generators such as DALL·E. And more solutions are being released every day.
For example, there is now an AI tool that looks at a video and suggests six interesting clips that will get the most engagement. And another that reads documents and then automatically generates 10 social media posts.
Generative AI is already very useful in pharma marketing. You will want to get good at it. That’s why we encourage our clients to invest time in learning what it can do. Your mind will start racing and likely find things right away. A good initial step is to task your team with understanding what’s available and to come back with two or three ideas for how they can improve your content supply chain. Once you start looking, the opportunities to increase content production speed are astounding.
Use generative AI correctly
That said, you don’t simply want lots of content. You want more of the right content. That requires a refined approach that adapts AI technologies to the demands of healthcare communication and the specific needs of your brands.
AI tools are trained on huge data sets. That makes them powerful, but it also means that the answers that they produce won’t be specific to your needs. The hours you save by using generative AI shouldn't be spent sorting through endless variations of irrelevant content and then remaking them to match your brand’s identity and tone of voice.
You also don’t want to block your MLR process. This is essential. Colleagues in MLR are already under pressure to review ever-increasing amounts of digital content. Adding generative AI for exponential growth just won’t work. If approvals become slower, the potential of AI cannot be realised in pharma. When you apply AI in healthcare marketing, it needs to work for everyone within your organisation.
How to solve it?
Always on-brand AI content
One solution is to use AI more sophisticatedly — ensuring that all generated content matches your requirements right from the start. That’s the thinking behind Anthill’s AI image services. We use our understanding of your products’ visual identities and train an AI to output relevant imagery every time. This way, large on-brand image sets can be created quickly and without licensing issues or expensive photoshoots.
The result is that you can provide affiliates with HCP and patient images that precisely meet local requirements — accounting for ethnicity, gender, height, weight, clothing, expression or any other factor. Achieving this at scale was impossible before generative AI. And it is a good example of how to use pharma marketing AI technologies to meet people’s real needs in healthcare.
If you want to start with AI in healthcare marketing, on-brand AI imagery shows what’s possible today.
Unlock AI’s potential with modular content
Another way to ensure AI fits the specific needs of healthcare is to work with modular content. While an investment to set up initially, it enables you to get far more value from pharma marketing AI. In fact, modular content can unlock AI's potential in all three stages of the supply chain: generation, approval, and delivery.
You can think about modular content as occupying a 'middle position' between individual content components, such as texts and graphics, and completed assets, such as a Veeva Approved Email or an eDetailing presentation. Each content module includes everything needed to tell a complete mini-story, e.g. product claims, references, copy, graphics, logos, and the ‘business rules’ that determine correct use. That means that modules can work on their own and in combination with each other without causing citation problems or other issues.
This ability to function like building blocks is why modular content has such potential for AI in healthcare marketing.
MLR, modular content, and AI
For colleagues in MLR, modular content simplifies approvals and avoids the duplication of work. Before modular content was available, MLR had to review each asset individually — even if it contained the same content as already approved materials. Now MLR can approve each content module once and be confident that it delivers precisely the same message, in the same way, no matter where it is used. As a result, you can combine and recombine modules as required. You can use the same modules in emails, presentations, websites and any other channel.
With regard to AI, modular content has two major advantages. Firstly, it enables MLR to work more efficiently. Because more content can now get approved, you can use generative AI to increase production volume. In other words, modular content makes life easier for your MLR colleagues and ensures they are not a ‘bottleneck’ in the supply chain.
The second advantage is even more significant. Modularising content enables you to use AI ‘after MLR’ to assemble assets with previously impossible efficiency and speed.
After MLR: AI content assembly
Following MLR, you can use AI in healthcare marketing differently — not to generate content but to find and sort information, assemble assets, and recommend materials that are more likely to meet people's needs.
This is a different way of working for many healthcare companies. But AI opens new opportunities that will change workflows. Many currently necessary marketing tasks will be increasingly seen as ‘pre-AI’ legacy activities.
For example, most companies currently have to define all required assets many months in advance, brief their creative partners to produce them, and then spend a great deal of time overseeing the whole production process. But with AI, you can re-engineer the supply chain — removing many steps to work faster and in more agile ways.
This transformation is possible because recent developments in AI research and cloud computing have moved the home of AI out of supercomputing clusters in labs and into the mainstream. The ability to use natural language, via written or spoken prompts, has democratised access to AI — and that accessibility turns potential into real change.
If you use natural language to tell an AI what you want to create — whether it’s an email, slide or website — very different workflows become possible. Combining an accessible UI, modular content, and AI's ability to find, sort and assemble that content, produces fast, simple and intuitive workflows that can build large volumes of final tactics in a safe and compliant way that meets all regulations.
In the example below, it’s important to note that the AI is inventing nothing. It is simply drawing upon a library of pre-approved content. It then uses this content to populate a relevant template. The AI can then determine which modules apply to that specific task and use customer data to suggest the most compelling subject lines.
These systems are already possible, and we believe they will become standard practice across the healthcare industry.
Closer to the customer
These new workflows, powered by an AI trained on your pre-approved content, mean that you can be more responsive to customer needs. For example, local brand teams can assemble materials to match a specific customer profile — such as a set of Veeva Approved Emails for HCPs who have a preference for video content. And because only relatively light approval of assets is needed for already-reviewed modular content, this can happen at speed.
This efficiency moves your focus from the content supply chain and redirects it to where it should be pointed — the customer. Because you spend less time managing processes, you have more time to understand HCP needs, develop strategies, and build customer experiences.
It also frees your partners to work more creatively. Agencies can now reassign the time currently spent on routine layout tasks to explain your product in the most impactful way possible. In other words, they can move 'upstream' by creating the modules and templates that will be AI-assembled whenever required.
At Anthill, we think it's important to keep these two benefits of pharma marketing AI — efficiency and creativity — balanced. It’s reflected in the discussions we’re having, the work we’re doing, and how we’re structured. On the consultancy and technology side, people are excited about leveraging AI to produce content at a previously inconceivable speed and scale. But we are also an agency. As such, Anthill’s creative team are just as enthusiastic about AI, though their focus is on how it expands their capacity for originality and creativity and enables a new level of omnichannel engagement.
Healthcare already leading in AI
The healthcare industry is actually at the forefront of AI. The 2023 Artificial Intelligence Index Report from Stanford University highlighted that the AI focus area with the most investment was medical and healthcare (6 billion dollars), followed by data management, processing and cloud (5.9 billion), and FinTech (5.5 billion).
Healthcare companies are making extraordinary advances in accurate diagnoses, such as improved radiomics that can detect clinically relevant features in imaging data beyond that perceptible by the human eye. And there are exciting developments in using AI to enable more personalised treatments.
That said, pharma marketing AI is just getting started. That same Stanford report noted that healthcare marketing and sales use of AI was less than half the average across all industries. And it was considerably behind sectors such as business, legal and professional services and financial services. This will change.
As an AI leader, the healthcare sector will be looking to apply AI technologies to customer engagement soon. Currently, companies are figuring out how to do it and make their first moves. So, let’s circle back to the title of this Insight: How to start with pharma marketing AI today?
To answer, we provided a simple framework — ‘before and after MLR’ — that Anthill is using to structure our work integrating AI in healthcare marketing with our clients. This simple division enables you to start safely by applying AI differently in different stages of the content supply chain. We also looked briefly at how you can make generative AI more relevant to your needs, using the example of Anthill’s new AI image services.
There are tremendous opportunities to apply AI in healthcare marketing, and this Insight has just touched on the many possibilities. Combining AI's analytical and predictive strengths with its content generation and assembly capabilities will change everything. Ultimately, it will realise the long-standing ambition in healthcare to increase customer-centricity and provide individually relevant information at scale.
Such profile-driven communication is currently challenging, but AI — used safely and ethically — can change that. What's important now is to set up your organisation to unlock the potential. It will be a journey. If you would like to explore the potential with Anthill, please get in touch with us.
In this ‘fireside chat’, Paul Simms interviews Anthill’s Chief Technology Officer, Tor Kristensen, about the realities of AI for pharma marketing. Tor explains how we should be thinking about it and what we should be doing to get all the potential benefits in a way that is safe and relevant to healthcare.