SPEAKERS:
- Ravit Ansal, SVP, Chief Solutions Officer, Enterprise Commercial Solutions, Indegene
- Pippa Postins, Global Head of Commercial Marketing, Global Pharma Strategy, Roche
- Carla Benedito, VP, Head of Global Commercial Operations, Takeda
- Justin Schneller, Global Director, Content Excellence, CSL
- Dr. Michael A. Kurr, Founder & Content Management Transformation Expert, Dr. Kurr Advisory
KEY TAKEAWAYS:
• Sixty to seventy percent of pharma content never reaches a customer, representing billions in structural waste
• The six-month lag between global creation and local deployment is a competitive vulnerability, not an operational nuisance
• AI scales existing dysfunction, organizations without feedback loops will amplify fragmentation, not eliminate it
• Modular, claim-level content architecture is the prerequisite for both local adaptability and AI optimization
• Decision ownership is the unsolved variable that defeats every other investment
When Ravit Ansal asked the room at Pharma 2026 how many organizations had feedback loops that actively informed what content got created next, almost no hands went up. That moment deserved more attention than it received. Pharma companies collectively spend more than $2 billion annually on content that never reaches a physician, a payer, or a patient. The absence of hands wasn't a confession of technical immaturity. It was an admission that the industry has been running a one-way content pipeline for decades and calling it a system.
The panel that followed didn't offer a roadmap. What it produced was more useful: a precise diagnosis of why the dysfunction persists, and why the arrival of generative AI has made the cost of inaction measurable in ways it wasn't before.
Six Months Is a Strategy Problem
The standard framing of global-to-local content failure centers on volume, too much produced, too little used. The more damaging variable is time. "In 2026, the greatest waste actually isn't content that Global creates that isn't used by affiliates," Postins said. "It's the six-month time lag where Global and affiliates aren't talking to each other." When headquarters and country teams operate on disconnected cycles, affiliates don't wait. They commission their own agencies, create their own assets, and by the time global delivers its package, the local version is already in market. The result isn't localization. It's fragmentation dressed as process.
Schneller named the downstream consequence that tends to get underweighted: "If we have all these local variations of content, the AI models are going to be struggling to keep up with what is the true source of truth coming from us." Every unauthorized local variant is a data point that dilutes the signal global spent months building. The brand problem and the AI training problem are the same problem.
Benedito positioned the failure as organizational rather than operational. "Strategy doesn't end at claims definition," she argued. "Operations does not end at campaign approval and MLR approval. This ends at impact, at changing behavior, at changing patterns of utilization of our medicines." The metric misalignment she described is structural: global teams are measured on content produced and approved; local teams are measured on speed of execution. Neither incentive rewards content reuse. "Global today does not have the same metrics as far as content is concerned as local does," she noted, which means the feedback loop has no shared language even when the data exists to build one.
Postins condensed the ambition into a phrase that captures the actual goal: "We want one voice with 80 local accents." Achieving that requires something the industry hasn't built — a connected system where local signal informs global creation in near real time, and where global strategy arrives early enough that affiliates don't need to improvise.
AI Scales What You Already Are
The AI layer should be where this gets fixed. Instead, it's where the dysfunction gets amplified. Kurr, drawing on more than 15 years of transformation work across major pharma organizations, was direct about the trajectory: "We are not fixing anything here. We are actually scaling the problem." He cited Gartner data indicating that nearly a third of marketers report no effective system for data-enabled decision-making, alongside MIT findings suggesting 95% of AI initiatives are failing to meet expected business outcomes. The organizations deploying AI fastest onto broken content operations are accelerating toward a more expensive version of the same failure.
The mechanism is straightforward. "AI is always only scaling who you are and what you have in place already," Kurr said. "Which means wherever there are still cracks in the system, AI will expose this pretty brutally." Schneller made the same point from the implementation side, describing what foundations have to precede any AI deployment: taxonomy and metadata systems defined at enterprise level, modular design systems, governance embedded across the workflow. "Without that, it just becomes another shiny thing that you do. And in most cases it becomes a failure."
The MLR bottleneck sits at the center of this problem. Schneller framed its logic precisely: "It makes no point to report back on whether a piece of content worked or not if it takes three months to make that change visible to our customers." Benedito's proposed solution isn't faster approval, it's less content requiring approval. She argued for a structural shift toward modular, claim-level content: atomic pieces with embedded references, qualifiers, population data, and epidemiology that can be assembled locally without triggering full re-review. "We are producing too much content," she said. "We don't need so much creativity these days. What we need is good blocks of content and great processes."
Kurr added the governance dimension that makes the MLR debate more complex than it appears. The approval process exists not primarily to ensure content quality but to establish accountability. "This discomfort potentially is not going away of the people who need to sign up," he cautioned. Speeding up review with AI doesn't resolve the underlying question of who owns the decision. It surfaces the question faster.
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.
The Dashboard Nobody Reads
Most large pharma organizations have content performance dashboards. The problem isn't data availability. Schneller was candid about where CSL currently sits: "A lot of that, in my personal opinion, is a little bit vanity metrics, how many times a video has been watched or so forth." The gap between reporting and decision-making is where the feedback loop breaks. "The gap is really around moving from just reporting to generating an insight and an action from that," he said.
Kurr proposed the KPI that's almost universally absent: "Did my last decision lead to a better outcome than the iteration before?" That question is deceptively simple. It requires knowing who made the decision, what change it produced, and whether the outcome improved, a chain of accountability most organizations haven't formalized. "As long as it is unclear who is really owning the ultimate decision, who is reading the dashboards, who is deriving the insights, who is owning the decisions that are being made on the back of these insights, and then who is on the hook for the outcome on the back of this decision, the system will continuously break," he argued.
Postins added a structural precondition that often gets skipped: "The challenge for a lot of organizations is that they may not have the systems and data structured in such a way that the feedback loops actually exist." The dashboard problem is often actually a data architecture problem. Feedback loops that appear to be functioning are, in practice, operating on word of mouth and informal market intelligence rather than structured signal.
Physicians Are Asking AI, Not Visiting Portals
The global-to-local problem is now intersecting with a third variable that changes the urgency calculus entirely. Ansal introduced Generative Engine Optimization, the emerging discipline of ensuring pharma content surfaces accurately inside AI-generated responses, not just in search results or branded platforms. "38% of GPs rate information from AI as critical," Postins noted. "That's a fundamental shift in our world."
Physicians who previously navigated to a company's medical portal are now getting synthesized answers from ChatGPT, Claude, or Perplexity, answers assembled from whatever data those models have indexed. Postins reframed the strategic goal accordingly: "How do we become that hidden trusted partner for the doctor so that we are the primary quoted source that the AI is using? We build our success not only on the narrative that we're telling, but on building the layers of truth within the AI models."
Ansal introduced the risk dimension that makes this more than a positioning question. When physicians search AI engines, they receive responses that no one at the company reviewed or approved. Dosing information, label details, and comparative claims can be misrepresented without any visible signal that something went wrong. Ansal called this silent risk, the company doesn't know the error occurred, and neither does the physician who acted on it.
Benedito connected this to resourcing with unusual directness: "If we want to do the work the same way... our teams are going to collapse." Her prescription was explicit. Stop producing standalone creative assets, build modular content structured for LLM consumption, and make explicit trade-offs about what the organization stops doing. "We need to decide what are we not doing," she said. That's not a rhetorical question. It's a resource allocation decision most content teams haven't been forced to make. Schneller reinforced the structural shift: at CSL, the team is actively reconsidering its agency model to move away from the pattern of launching with traditional assets and hoping markets use them.
The Commitment the System Requires
What emerged across the session is that pharma's content operations problem is neither a technology deficit nor a strategy gap. It's a decision-ownership problem at scale. Organizations have invested in platforms, AI tools, dashboards, and governance frameworks, and the feedback loop still doesn't close because no one has been made specifically accountable for acting on the signal those systems generate.
Kurr's commitment for the following Monday was the most operationally honest statement of the session: "I will be much more explicit in my engagements with asking the people who is owning which kinds of decisions, and I will start writing it down to have the right level of formality in place." Schneller's was measurement-first, focus on content reuse as the single metric that exposes duplication, fragmentation, and inconsistency simultaneously. Fixing reuse, he argued, produces a three-for-one return on cost, speed, and brand coherence.
Kurr's point about execution consistency cuts against how most content teams still operate. "The consistency of execution trumps perfection every single time," he said. Pharma's orientation toward perfection, every asset optimized, every claim legally ironclad, has produced a system where the pursuit of perfect content crowds out the consistent delivery of “good-enough” content. The modular, atomic content model Benedito and Schneller both described is, at its core, a bet that consistently good is more commercially valuable than intermittent perfect.
The organizations that close this loop first won't do it by deploying better AI. They'll do it by answering the questions the technology can't answer for them: who decides, who acts, and who is accountable when the outcome doesn't improve. Until those questions have names attached to them, the feedback loop remains a dashboard feature. And AI, in the meantime, will keep scaling whatever answer already exists, correct or not, into every channel the company thought it controlled.
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.
