Webinar wrap-up: How pharma can leverage AI-driven synthetic personas
In a recent webinar sponsored by Lumanity – Using AI-driven synthetic personas to take your insights further – Damian Eade, Technology Transformation Officer, Insight, at Lumanity and Stéphane Lebrat, Global Insights and Analytics Director at Takeda, discussed how, as AI continues to reshape the healthcare landscape, pharma teams are beginning to leverage synthetic personas.
Synthetic personas transform static data into dynamic, interactive conversations. They can be deployed at multiple points in the brand and insight cycle – from enriching qualitative research, to helping local teams interpret and implement segmentation, to ongoing pressure testing of campaigns.
Interactive, evidence-backed simulations
Synthetic personas turn static, traditional personas into interactive, evidence-backed simulations that let teams ask nuanced questions, probe motivations and explore objections – without replacing real customer research.
“Essentially, [segmentation] provides a classification system for us,” explained Eade. “It helps us to prioritise, target, and allocate resources and [is] completely valid, completely evidence based. But – and this is the big part – it can also feel quite abstract. That's the risk with your segmentation […] People don't always necessarily connect with it emotionally in the way that we would hope. So, this is where we have the persona, which is really, by contrast, a way to add that human translation.”
Indeed, a traditional persona takes a strategic segment and turns it into a more ‘human story’, with a name, a face, motivations, anxieties, behaviours. “It's how you make a segment memorable and usable for teams who really need to then be able to design experiences or create messages or be able to plan their campaigns,” said Eade.
“The need for the persona is absolutely critical [and ] with the approach of the synthetic persona, you have the best of both worlds […] It's a game-changer.”
Stéphane Lebrat
A synthetic persona, however, provides an interactive simulation.
“It's not just about reading a persona,” continued Eade. “It's the ability to actually have that two-way dialogue to be able to ask questions, probe motivations, really dig in and explore hesitations, pressure points, et cetera.”
For Lebrat, the importance lies in the global translation potential, also.
“I'm working in a global team – so, you do one persona right, globally,” stated Lebrat. “The need for the persona is absolutely critical [and ] with the approach of the synthetic persona, you have the best of both worlds […] It's a game-changer.”
Resonance over numerics
Synthetic personas complement digital twins, but are not the same. Digital twins model “what happens if” with numeric outputs, while personas explore “what resonates and why” via contextual, conversational responses for messaging, positioning, and channel choices.
“I think one of the things that I often see coming more and more now, is a kind of further confusion or conflation between this idea of synthetic personas […] and this idea of digital twins, which are often used and often talked about in the very same breath,” noted Eade. “They are actually really different tools that answer very different questions, fundamentally.”
Digital twins are probabilistic statistical models; they simulate decision making in a very repeatable, structured way. “They're best when the question is, ‘Well, if I change X, what happens?’,” explained Eade.
“Synthetic personas are narrative synthesis models […] They simulate archetypal behaviours, attitudes,” he went on. “The output here is answer generation: it's conversational, it's contextual, it's exploratory responses. And so, as a result, they're much better for when the question is, ‘What's the hesitation with my customer?’ [or] ‘How might they react to X message?’.”
For Lebrat, it goes back to simplicity of use: the synthetic persona leverages the data a team already has. Synthetic personas bridge the “one-slide compression” problem by grounding every conversational answer in cited segmentation data, publications, and research, reducing shortcuts and speculation across global and local teams.
“Teams are going to refer to the persona and refer to what they have here and they will have some questions about who's the target, what's the need, how to communicate with them – we have tonnes of data,” Lebrat explained. “I've just done this exercise […] on segmentation in five countries: we had 130 variables. [Now,] you just go through the persona and then you speculate. It's not even a guess, it speculates.”
Conversational strategy accelerator with guardrails-driven value
Success depends on solid segmentation inputs, of course, and disciplined usage (clear verification of citations, no leading prompts, no copy paste), with clear global guardrails for local adaptation. Further, value metrics should be tied to use cases (e.g., time saved, higher engagement, better brand-plan rigor).
“[It’s about] how to make sure the asset will address the unmet medical need of your target population of your persona, of your key segment,” explained Lebrat. “The data we've collected through primary market research, we complemented this with some cultural elements because the US is not France, it's not Switzerland, it's not Brazil, it's not India, it's not China, right? So, you put some element of context in the same database.”
“But in one year from now, we'll have a new brand plan, right?” continued Lebrat. “So, they will revisit the persona again, revisit what they've done, what worked, what didn't […] They will continue to work with that persona, [transformed] into a true conversation.”
Eade agreed that it is this idea of an interactive layer that sets synthetic personas apart.
“It's about keeping insights alive,” added Eade. “It's about the ability to activate segmentations beyond the debrief: pressure test earlier, iterate faster. Fundamentally, this comes back to this central idea [that] often we hear [about] research sitting on the shelf, a problem of collecting dust. It's a way to make sure that we're shaking off the dust and we're getting engagement back into the research in a different way, something that's much more dynamic.”
“[It can help] local markets, sales teams, cross functional partners be able to engage with segmentation in a way that's more usable for them,” Eade continued. “We can think about it almost like the ability to role play objection handling with customers.”
“[Synthetic personas] allow adaptation, to go straight to a local team in whatever the country. The tool is designed to consider the local specifics and talk in their own language.”
Stéphane Lebrat
And the local to global adaptation stage? Deploying them as an interactive layer across discover–design–deploy phases speeds hypothesis formation, sharpens research design, pressure-tests strategy and creative, and trains field teams through role-play and objection handling, often in local languages.
“There are risks of failure that can become somewhat predictable: a one-size-fits-all global truth that ultimately doesn't really land locally as it would be hoped; a fragmented local reinvention where every market is quietly doing their own thing and creating their own version of the customer,” explained Eade. “So, it really becomes a question of how you give local teams the ability to be able to adapt, while keeping that kind of core strategic spine intact.”
“I work with medical, commercial, marketing, so many people, communications as well,” noted Lebrat. “So, there are a lot of different partners. This is just for the global layer; if you put on top regional, and then local, you have an army of business partners to manage in that context. [Synthetic personas] allow adaptation, to go straight to a local team in whatever the country. The tool is designed to consider the local specifics and talk in their own language.”
Getting to the deep human truth from quality data, and marketing it
The webinar included for a Takeda-specific case study, as well as a demonstration of Lumanity’s EMULaiTOR technology.
“This tool provides a lot of added value […] going to the insights, helping us to leverage the full data set that we collect, going as deep as possible,” enthused Lebrat.
Questions from the audience included queries about how healthcare consultancies interested in adding synthetic personas as an offering should go about transforming segmentation outputs into such interactive synthetic personas.
“You need a bit of a tech infrastructure of course, but I think probably one of the most important things is being really clear in terms of how you manage and treat the data,” replied Eade. “Pre-processing of the data is such an important factor and making sure that [you’re] curating quality data in the first instance, that [you’re] also managing that data so that it is appropriately indexable, so quality responses can then be provided.”
“It is a fantastic tool, but it shouldn't be used to replace going out and speaking to customers – that is not going away. It's going to help you inform and speak to them more effectively.”
Damian Eade
And in terms of trust and preventing an overreliance on AI outputs?
“It’s worth nothing if we don't spend due care, attention, and diligence around the deployment and the human aspect of this,” insisted Eade. “It’s really important to make sure that we are educating people, that we're giving them the rules of engagement, the etiquette that I spoke to earlier. Simple factors that help really ensure that people are able to use [this] responsibly, that they are using it in a way that they're going to get effective and efficient responses. This is still AI, so we need to make sure that people aren't copy and pasting.
“It is a fantastic tool, but it shouldn't be used to replace going out and speaking to customers – that is not going away. It's going to help you inform and speak to them more effectively,” Eade concluded.
Meet the panellists
Damian Eade, Technology Transformation Officer, Insight, Lumanity
With over 20 years in healthcare market research and more than a decade specialising in digital and technology-driven methods, Damian Eade leads AI and tech implementation at Lumanity Insight, ensuring advanced tools deliver meaningful, impactful insights. Previously global head of digital and managing director of the social media insights and analytics expert team, his career spans leadership roles in technology-enabled insight agencies and client-side experience at GSK.
Stéphane Lebrat, Global Insights and Analytics Director, Takeda
Stéphane Lebrat works at the intersection of strategy, insights, and storytelling. An expert in insights, analytics, and competitive intelligence within the pharmaceutical industry, he has partnered with a broad range of leading pharma to shape strategy, drive innovation, and turn complex data into clear, actionable business insights. With a strong focus on the human perspective, local realities, and real world impact, Lebrat has played a key role in global vaccine launches.
About Lumanity
Lumanity is a leading global strategic partner for biopharmaceutical companies. We engineer breakthrough value to tackle our clients’ toughest challenges and transform lives. By combining deep scientific, clinical, medical, regulatory, and commercial expertise with advanced technology and AI-driven tools, Lumanity guides complex decision-making and execution across the entire medicine value creation journey.
With 1,200 highly specialized experts working in more than 50 countries, and offices in North America, the UK, the EU, and Asia, we collaborate with nearly all of the top pharmaceutical companies and over 100 biotechs worldwide.
Our integrated approach brings together strategy, evidence, engagement, and technology - helping clients navigate market complexities with confidence, accelerate commercial success, and ultimately improve patient health outcomes.
