SPEAKERS:
- Sam Rolfe, Sales Director, Exeevo (Moderator)
- Chetak Buaria, VP, Global Commercial Operations, Merck
- Philippe Guisset, Business Unit Director, Alfasigma
- Dr. Konstanze Wagner, Executive Director Strategy & Operations, Bristol Myers Squibb Germany
- Michael Zaiac, Head of Medical Affairs Oncology EUCAN, Daiichi Sankyo
KEY TAKEAWAYS:
- Field teams are architected for message delivery, not signal interpretation
- Depending on the specialty area, reps sometimes access only 20% of their physician universe, yet spend under five minutes preparing each call
- Digital twins of HCPs now enable signal-detection practice in consequence-free environments
- Cross-functional leadership misalignment is the primary brake on sustained upskilling
- HCPs querying ChatGPT about your drug may be your strongest buying signal, not your biggest threat
As Head of Medical Affairs Oncology for Daiichi Sankyo across Europe and Canada, Michael Zaiac maintains patient contact , uniquely positioning him on both sides of the healthcare professional engagement spectrum. From the clinical perspective, his stance is unequivocal: "When someone approaches me with a problem, I seek assistance in resolving that issue," he asserted at Reuters Pharma 2026. "I do not wish to become a conduit for delivering a message." This reversal—from being a problem-solver to merely a messenger—is precisely the dynamic that the pharmaceutical field model has institutionalized on a broad scale.The external pressure compounds it. "HCPs are overwhelmed and they're disengaging at scale," observed moderator Sam Rolfe of Exeevo. "Opt-out is becoming the new no-see." Organizations have responded by investing in more channels, more consent infrastructure, more behavioral analytics, generating more signals than at any prior point. The paradox is that this data richness is widening the capability gap, not closing it. The industry has built an increasingly sophisticated listening apparatus and left the field teams without the skills to listen anything.
The Architecture Is the Problem
The upskilling challenge facing commercial and medical field teams is not primarily a technology deficit. It is a design choice that has calcified over decades, and the panel made clear that adding new tools on top of it will not fix it.
"Traditionally we have trained field teams for message delivery," said Chetak Buaria, VP of Global Commercial Operations at Merck. "In fact, we spend hundreds of thousands to even measure message recall. We don't train them to pick up signals and convert them into insights." The measurement infrastructure tells the full story: when the KPI is recall, the training produces recall. Signal interpretation isn't a skill gap that can be patched with an afternoon workshop. It requires a fundamentally different conception of what field interaction is for.
Guisset illustrated how deep the behavioral default runs. Comparing call preparation to a job interview, he noted that reps who treat each HCP interaction as genuinely high-stakes invest meaningfully more effort in reading context and personalizing their approach — yet most do not operate that way. The data is damning: reps access only 20% of their physician universe, "and they don't spend more than five minutes preparing the call and they copy-paste what will be the content." When access is already scarce, spending it on templated message delivery is not just a capability failure. It is a strategic misallocation of the most limited resource in the commercial model.
Zaiac located the problem at the level of mindset before it becomes a question of tools or training design. "Often on the medical side we go into a discussion with a very set aim," he noted. The skills missing are sequential: first, the disposition to listen rather than transmit; second, the content depth to move the conversation from product to therapeutic area when the HCP's real question requires it. Neither skill is developed by training programs oriented around message recall scores.
Buaria's contextual signal example captures what genuine signal literacy looks like in practice. A physician showing a drop in digital engagement looks, in isolation, like a disengagement problem requiring intervention. Viewed across the full omnichannel picture, that same physician may be increasingly active in peer-to-peer events the company organizes, meaning the signal actually indicates advancement on the advocacy ladder, not withdrawal from it.
The appropriate response is the opposite of what the isolated metric would suggest. Training teams to read signals without the contextual framework to interpret them doesn't produce better decisions. It produces faster wrong ones.
Who Is Actually at the Center
The organizational question underneath the capability question is one of orientation. Commercial models have been built around the rep as the primary unit of engagement. The panel was consistent that HCP-centricity and rep-centricity are not the same thing, and the gap between them is where execution fails.
"Not talking about the rep, but talking about the signal, talking about the engagement with the HCPs where we want to create value and impact," said Dr. Konstanze Wagner of Bristol Myers Squibb Germany. The distinction matters operationally: when the rep is the organizing principle, signals get filtered through what reps can capture, which means owned channels, direct interactions, and first-party data dominate. Guisset pressed on what that leaves out: third-party signals, the conversations happening outside pharma's channels entirely, the HCP's behavior on platforms the company doesn't control. Most signal architectures have a systematic blind spot toward anything not generated by a pharma-owned touchpoint.
The KPI structure reinforces the misorientation. "We need to come up in industry with KPIs which really allow us to measure the output of teams," Zaiac argued, "because a scientific team cannot exist without a commercial team and vice versa. Yet we are having somewhat very different KPIs. We are not encouraging teamwork." Separate measurement systems for medical and commercial functions don't just create reporting friction. They actively prevent the cross-functional signal aggregation that produces genuine HCP insight. When an MSL's performance metrics have no relationship to a rep's, there is no structural incentive to share what either has learned.
Wagner added a layer that often goes unexamined in upskilling conversations: "If senior leadership is not on board on sustainably supporting the journey on upskilling to move in that direction, you're much slower as an organization than you actually want to be." Field teams frequently reach ahead of the organizational systems designed to support them. They adopt a signal-interpretation mindset, then discover that the workflow for converting insight into brand material hasn't changed, that global teams are still operating without local intelligence, or that the next-best-action recommendation hasn't been updated in six weeks. Buaria noted that insights frequently get trapped at the local level, leaving global organizations with blind spots that no amount of field team upskilling can compensate for. The bottleneck, in those cases, is not the rep. It is the infrastructure the rep feeds.
Discover more on this topic at Pharma Commercial Data & Tech Europe 2026 (4-5 November, London) Europe’s collaborative home for data and tech pioneers. Visit the website here.
Where Technology Actually Helps
The technology story from this panel was not triumphalist, and that credibility is worth preserving. The adoption curve for field-facing tools follows a recognizable pattern: initial engagement is high, automation attempts disappoint, and teams reach a frustration plateau that determines whether the investment survives.
Guisset described it precisely. His team combined human expertise with AI tools for training and content development, and the output was strong. Then came the attempt to scale by reducing the human layer, and quality dropped. "What didn't work is that once you try to scale it and to make it efficient without as many human resources on the back, it was not as good." The lesson is not that automation fails, but that the human-machine calibration point is earlier in the process than most technology rollouts assume.
Wagner was equally direct about the patience required. "The value might only translate in a year or two years," she acknowledged. "You somehow need to bridge that time also and continue to translate and explain the value story for the field teams." Her organization is working through AI-enhanced next-best-action tools and voice-recording capabilities that give reps anonymous call feedback, useful for both individual development and marketing insight. Neither is delivering immediate visible ROI. Both require sustained change management investment before adoption reaches the level where the value story becomes self-evident.
Buaria's digital twin initiative addresses the capability problem more directly. HCP personas modeled on behavioral segmentation allow reps to practice signal detection in a consequence-free environment. The assessment is not message recall. It is how effectively the rep identified and responded to the signals the avatar was generating, and how they advanced the conversation based on what they heard. His ROI proxy is behavioral: "Is the rep coming back on its own to conduct that training?" Professionals who find a development tool genuinely useful return to it without being required to. Voluntary adoption is a cleaner signal of value than any metric a program manager could design.
The Forcing Function No One Planned For
The panel's closing exchange on LLMs surfaced a reframe that deserves more strategic attention than it typically receives. HCPs are already using large language models, including Zaiac himself, on both the clinical and medical affairs sides of his work. Pharma cannot out-scale the general-purpose LLMs, and attempting to build a competing information layer is a category error. "We shouldn't compete with the health version of ChatGPT," Zaiac argued. The sustainable position is proprietary: an agent built on company-specific clinical trial data, layered on top of what public models already provide and supported by the companies’ respective Medical teams where appropriate.
Buaria's framing of pharma's role is the more consequential shift. "Ideally we are the unbiased orchestrator of scientific information and help the physician contextualize the information that he or she would have received maybe from an LLM." That repositioning, from information provider to contextualizer, requires exactly the skills the panel spent the session describing as absent. If the HCP already has the data, the rep's value is in contextualization, not download. The training model that produces message delivery is not just inadequate for that role. It actively undermines it.
Guisset offered the reframe that should change how commercial teams respond to the LLM question entirely. "If I have an HCP searching for my drug or my disease on ChatGPT, it's the best buying signal." An HCP actively seeking information about a specific product or disease area has signaled intent more clearly than almost any behavioral metric pharma currently captures. The organizations treating LLM engagement as a threat to manage are looking at the same signal and drawing the wrong conclusion.
Rolfe closed with the framing that ties the session's argument together: "Translating these signals is clearly a strategic choice about how you build the capability into your organization, how you measure success, and the return on those investments to embed learning for the long term." The word "choice" is precise. Signal infrastructure can be purchased. The organizational will to retrain a field force around listening rather than transmitting, to align KPIs across commercial and medical functions, to give senior leadership a stake in upskilling — those are decisions, not implementations. The companies that have made them will find that LLM adoption by HCPs accelerates their advantage. Those that haven't will discover that in oncology and specialty care, where HCPs are already querying trial data through AI tools before the rep arrives, more signals processed through an unchanged commercial model will produce only more sophisticated noise.
All panelists are employed by pharmaceutical companies or by entities providing services to the pharmaceutical industry. The views and opinions expressed in this publication are solely those of the panelists and do not necessarily reflect the views, positions, or policies of their respective employers.
To get you highlights of Pharma 2026 faster, we are using generative AI technology to summarise the transcripts of the sessions. If you have any feedback about the summary, please contact lucy.fisher@thomsonreuters.com.
Discover more on this topic at Pharma Commercial Data & Tech Europe 2026 (4-5 November, London) Europe’s collaborative home for data and tech pioneers. Visit the website here.
