Can AI build an accurate B2B healthcare persona?

Understanding who to target used to take lots of money and lots of months for surveys, focus groups, interviews, and market research.

Now, of course, we have artificial intelligence tools that can do some of this for us. In a few minutes’ time you can prompt your AI tool of choice to provide a buyer persona that presumably tells you something about the people you’re trying to reach.

But is AI a good substitute for traditional audience research methods? Can it deliver an accurate persona for a specific company and industry, like a healthtech vendor that sells to health systems or physician practices?

And, more to the point for content and communications teams, to what extent can an AI-generated persona serve as a guide to developing content that target audiences notice and respond to?

Here’s what people working at the intersection of marketing and AI have to say about personas and their cousins, ideal customer profiles:

“They make it incredibly easy to ‘talk to’ your ideal customer and understand exactly what makes them tick — whether you’re brainstorming blog ideas, auditing your website for conversions, optimizing paid ads, etc.” – Andy Crestodina, content marketing speaker and writer, Orbit Media Studios (LinkedIn post, November 2024)

“As with so many other things, people have a tendency to treat ideal customer profiles as just another document on the pile, when in reality they’re the basis for some of the most powerful generative AI work you can accomplish. You know who your customer is, or who you want your customer to be.” – Christopher S. Penn, Trust Insights (Almost Timely newsletter, Oct. 6, 2024)

Four-step process

Let’s take Andy and Chris at their word and assume that AI can help build a pretty good persona. What’s the best way to go about it?

Andy recommends a simple four-step process for building a buyer persona. This works equally well for an ideal customer profile (ICP) – but note that the two aren’t the same. This process also removes some of the demographic assumptions that don’t seem as relevant in the B2B world (“Susan is in her early 50s, has two teenage daughters and lives in the suburbs”).

Copy his sample prompt (modifying for your company, industry and audience) into an AI tool
Refine it by adding, removing, or revising elements
Save it as a PDF for future use with your AI tools
Chat with it

Following the advice Andy and Chris have published, and combining it with what other content marketing experts have to say, here’s a prompt I used in ChatGPT 4o:

Build a persona of a chief operating officer at a multi-specialty medical group that has 25 clinics in three states. The chief operating officer is a non-clinical leadership role responsible for daily operating activities, IT, real estate, purchasing and contracting, overseeing practice managers, and related duties. This person is looking for help with adopting artificial intelligence tools that will help clinicians improve efficiency and deliver more personalized patient care. List their hopes, concerns, emotional triggers, and decision criteria for selecting a potential partner or vendor that provides AI tools for medical practices.

I got back a 700-word draft that could serve as a reasonably accurate starting point for a company that provides AI scribes and other AI tools for practices.

ChatGPT 4o told me that the medical group COO is concerned about administrative burdens on clinicians, patient outcomes, HIPAA compliance, interoperability, IT implementation challenges, and other issues common to practice management.

This list – which is part of the full persona I got back – is a decent starting point for developing content. A few ideas spring to mind:

Thought leadership from your CEO on the industry’s need for seamless integration as healthcare providers adopt more AI tools
A case study about a successful adoption of your technology
A sales piece on financial advantages of using AI, complete with a calculator or similar tool on your product’s landing page

Chris and his Trust Insights colleague, Katie Robbert, made an interesting point about using an AI-generated persona in this podcast: Watch out for assumptions. If you’re using the persona to develop a content strategy, chances are it reflects a potential buyer who already understands the problem your company solves. That might not accurately reflect the thinking of a prospect who’s unaware of both their problem and your solution.

Go deeper

There’s probably not much in an AI-generated persona that you wouldn’t come up with on your own or with your team. But that’s going to take a lot longer than the 15 seconds or so that AI can churn something out.

I think most people who regularly use generative AI have learned to go deeper than simply accepting the first output. For example, Andy’s step 4 is “chat with it.”

In the medical practice example, some questions you might think of to ask AI:

Where do you get your information about possible IT vendors for the practice? What websites, blogs, trade publications, newsletters, podcasts, and influencers do you follow?
Where do you go to learn more about tools like AI scribes?
What do you not like about talking to vendors?
What’s one thing you think is misunderstood about operating a medical group?
What do you wish the clinicians in your practice understood and appreciated more about AI?

The responses won’t entirely hit the mark. Maybe only a small percentage would square with what you and your team already know about your buyers and prospects.

Still, there are probably ideas and insights you wouldn’t have thought of. Taking those and running them by team members could spur ways to improve marketing tactics for the target audience.

A few limitations

No AI-generated persona is sufficient without going deeper. AI can’t build a useful one without accurate inputs and a thorough analysis by a human expert. Here are some warnings about relying on AI personas:

This exercise is based on assumptions, not data. AI tools are helpful, but not as much as what your actual customers and prospects are saying and how they’re behaving.
Generative AI tools work best when you follow up with more questions, refining as you go to get better results. Ask AI to check its work.
One persona probably won’t cut it, given the lengthy and complex buying processes in healthcare organizations. Not only do you need personas for those involved in purchasing decisions, you probably need to define those who might influence the decisions, including third parties like industry analysts and media.

Once you’re reasonably satisfied with your persona, save the document. Give the AI tool a draft content piece, upload the persona and ask it how well the piece matches the persona’s informational needs.

Despite the limitations, using AI to generate a profile of your marketing’s target audience is a good starting point, especially for smaller companies unable to devote big dollars to market research. It’s certainly better than guessing at who’s likely to be interested in your content.