Introduction

SPEAKER:

Ben Myall, CEO, Everybody Agency  

KEY TAKEAWAYS:  
- ChatGPT's 800 million weekly users signal urgent pharmaceutical AI visibility opportunity  
- Organic search traffic projected to decline 50% by 2028 disrupting current engagement models  
- Generative engine optimization requires content structured as modular question-and-answer units  
- Benchmarking AI overview and LLM visibility is now measurable and strategically essential  
- One life sciences client achieved 82% increase in AI overview citations after optimization  
 
The search landscape that pharmaceutical marketers have relied upon for two decades is undergoing a fundamental structural shift and patient engagement strategy must shift with it. Generative AI platforms are rapidly becoming the first destination for health information seekers, demanding that pharma brands develop new capabilities in visibility, content architecture, and performance measurement before the window for first-mover advantage closes.  

Where Patients Are Searching Now  
The scale of the behavioral change already underway is striking. ChatGPT reports 800 million global weekly users, and that number continues to climb steeply. More than 5% of all prompts entered into the platform are healthcare related, translating to approximately 40 million health queries every single day. "It's clearly an opportunity for us to be in the right place to provide personalized information when people are seeking it about healthcare," said Ben Myall, CEO of Everybody Agency, presenting at Pharma 2026.  

The reach of this shift extends beyond patients. Research indicates that 28% of healthcare professionals are using AI tools in their clinical practice, and roughly half of those are using consumer platforms such as ChatGPT and Perplexity rather than purpose-built clinical environments. For pharmaceutical brands accustomed to segmenting their digital strategies between patient-facing and HCP-facing channels, this convergence demands a reconsideration of how and where content is deployed.  

The urgency is compounded by what is happening to conventional search. AI overviews in Google already reduce click-through rates to first-position organic results by approximately 35%. By 2028, organic traffic to pharmaceutical websites is projected to decline by around 50%. Myall is direct about the implication: brands that continue to rely solely on traditional search engine optimization are investing in a channel that is structurally contracting.  

Building the GEO Strategic Framework  
The methodology Myall presents for generative engine optimization draws on the established logic of search strategy while introducing critical adaptations. Insight gathering, benchmarking, content planning, technical optimization, and measurement remain the core stages. What changes is the precision required at each step.  

Audience insight now demands a dual-channel approach. Keyword research, particularly long-tail queries that reveal the specific questions patients are entering into search engines, provides a reliable proxy for the prompts they are likely to use in generative AI platforms. Social listening adds a second layer, supplying the emotional context and conversational language that keyword data alone cannot capture. "If the search behavior tells us what people are looking for, then the social listening can tell us why they're looking for it," Myall explains. Together these inputs build portfolios of intent that drive both content planning and optimization priorities.  

Benchmarking now requires two distinct lenses. The first is AI overview visibility within Google: understanding which queries trigger AI overviews, whether the brand appears within them, and who else does. The second is LLM visibility across platforms such as ChatGPT and Perplexity, now measurable through tools including SEMrush, Profound, Omnia, and AccuRanker. "When I was speaking about this back in late 2024, there's basically no platform that told you anything about LLM visibility," Myall notes. That gap has closed, and the data it now produces should be treated as a core strategic input.  

Content Architecture for AI Retrieval  
The most consequential shift in execution concerns how content is structured. The web page as a unit of optimization is no longer sufficient. Generative AI does not retrieve pages, it retrieves modules. "The way that AI retrieves content is they retrieve it in modules," Myall explains, describing a model in which each discrete section of a page is built to answer a specific, identifiable question. This reframes the content brief: rather than designing a page to satisfy a broad need state, teams must map individual questions to individual content units, each formatted for direct retrieval. 

 
Myall identifies several structural elements that measurably improve AI citation rates. Key fact blocks allow generative platforms to extract and surface precise answers. Bullet-pointed responses to common questions provide retrievable structure. Comparison tables in HTML format are particularly effective, "for some reason AI loves a table," he observes, noting that well-structured comparative data is retrieved with high frequency. YouTube content is also a significant and underutilized factor: "Your YouTube content and its correct optimization will affect your AI visibility," Myall states, drawing on benchmarking conducted across multiple pharmaceutical and life sciences clients.  

Schema markup and structured data represent a technical priority that many pharmaceutical teams have not yet addressed. A site-wide schema audit can substantially improve the contextual signals available to AI platforms when determining whether and how to cite content. Internal linking architecture, mobile performance, core web vitals, and the handling of JavaScript, which AI crawlers cannot process as effectively as traditional search engines, are all technical factors that directly affect generative retrievability.  

From Optimization to Measurable Outcome  
The strategic case for acting now is supported by a case study from Everybody Agency's work with a leading life sciences company on a global portal. Following a structured GEO program encompassing technical optimization, FAQ-structured content, schema updates, and internal linking improvements, the client achieved an 82% increase in AI overview citations, a 30% increase in share of available AI overviews, and a 222% increase in LLM referral traffic, all while the overall opportunity in the market was simultaneously expanding.  

Myall is clear-eyed about what LLM referral traffic does and does not represent. Because generative AI delivers answers within the platform experience itself, referral volumes will always be modest relative to traditional click-through. The metric's value lies in trajectory: a doubling or tripling of GenAI referral traffic over twelve months indicates that brand visibility within those environments has grown proportionally.  

On the question of AI-assisted content creation, Myall acknowledges the regulatory friction that pharmaceutical companies face. Early experiments have demonstrated that efficiency gains in production are frequently offset by extended MLR review cycles when content is AI-generated. His recommendation is to invest in training proprietary LLMs to align more closely with MLR requirements, a process he describes as iterative but ultimately viable. "I do believe that experimentation will pay off really soon," he says.  

The strategic imperative resolves to three actions: act immediately rather than deferring to future planning cycles, embed GEO thinking into content planning as a standing operational discipline, and build measurement infrastructure now so that insights continuously inform optimization. The brands that treat generative AI visibility as a future consideration rather than a present competitive reality risk ceding patient engagement at precisely the moment the channel is growing fastest.  

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 Customer Engagement USA 2026 (October 27-28, Philadelphia) - where commercial, marketing, medical, data and AI pioneers converge. Explore the agenda here.

"QUOTE"

 Text

Strategy stability is not the same as rigidity 
Petruccelli's question on strategy-versus-pivot discipline exposed one of the most operationally useful tensions of the session. Every launch team talks about agility. Very few have thought through the mechanics of when to use it.

Veliz Salce offered the clearest mental model: "I see a strategy as the destination, but just like Waze sometimes tells me you need to take a different path because there is a roadblock here." Route changes are not destination changes. The discipline is in distinguishing between signals that warrant a course correction and noise that triggers premature abandonment of a sound strategy.

Doshi pushed back on the instinct to pivot too fast, particularly at the regional level. The strategy set before launch typically reflects sound national-level evidence. What it often misses is execution variance at the state or health system level - a Medicaid prior authorization policy that's ambiguous, a field team undertrained on medical necessity pathways for rare disease products that may not receive formulary access for a longer period of time. "Having that cross-functional team at the regional level and giving them a little bit of autonomy to make that little course correction" is how one can manage this. Local teams with real-time prescription data and the authority to act on it, without triggering a national level strategic overhaul.

 
Smith's contribution is structural: pre-defined pivot points. Build the inspection moments into the launch design before going to market. Decide in advance that at month three you will evaluate payer readiness against original assumptions. This prevents the two failure modes that plague launches - reactive pivoting based on early noise, and stubborn course-holding when the data is actually telling you something real. Smith has experienced both. "We took it out to the market and found out that they were not anywhere close to the curve," she said of one access strategy. The pivot was necessary. What would have helped was designing the trip wire earlier. 

AI is moving faster than the launch model it's replacing 
The panel's AI discussion surfaced a more uncomfortable truth than most AI-at-launch conversations produce. The teams using AI to improve launch execution are largely doing productivity work - faster insight synthesis, virtual HCP training, real-time field intelligence. That's real. But the more significant AI dynamic may be happening on the other side of the physician relationship. 

Smith named it directly: "We were spending so much time thinking about how we were going to use AI that we weren't thinking enough about how our customers were using AI." Payers are using AI to accelerate prior authorization reviews. Younger HCPs are using AI to answer clinical questions they would previously have asked an MSL. The traditional access model - rep, then MSL, then medical information - is being compressed at the front end by a generation of physicians who want a tailored answer in real time and won't wait twenty-four hours. Doshi's team is piloting portal-based AI responses to HCP inquiries, built on the recognition that physicians "want the response early and they want the response tailored to their question." 

Veliz Salce has deployed AI as a virtual physician for field force training, feeding the model with clinical evidence, treatment guidelines, and segment-specific market research so reps can practice against realistic objections before they're in front of a real prescriber. The infrastructure investment is modest. The intellectual investment - deciding what the AI needs to know to simulate the right physician - is where the work is. "It's more the thinking behind to ensure that I feed the AI with the right information," she said. 

The harder problem is narrative control. Doshi's argument here is the one launch leaders should carry back to their teams: "If you are not dynamic in real-time action, the narrative is going to be controlled by other people through social media, through AI, through the access to all of the information." Scientific communication plans were built for a world where the manufacturer controlled the pace of information flow. That world is gone. A competitor's off-label data, a caregiver forum's adverse event thread, an AI chatbot trained on unapproved sources - these now shape prescribing perception faster than a field force can respond. Veliz Salce pointed to the partnership implication: AI platforms currently serving HCPs lack access to approved, current product information, often by more than a year. Companies that move on strategic AI partnerships now will have a material advantage in the next launch cycle. 

The questions that determine whether day one actually works 
What the panel collectively described is the obsolescence of the sequential launch model. The old architecture ran in phases: evidence generation, then value story, then payer strategy, then field deployment. Each handoff lost fidelity. By the time insights reached the rep, they were months old and stripped of nuance. 

The model companies are building toward is continuous and parallel. HEOR embedded in clinical development to capture patient voice before endpoints are locked. Regional cross-functional teams empowered to interpret real-time data and correct local execution. AI surveillance monitoring narrative formation in channels the field can't reach. Social determinants layered onto clinical data to fill the gap between what patients experience and what physicians hear. 

Veliz Salce's challenge applies at the organizational level as much as the individual one: "We live in a world that is data rich but sometimes there are no actionable insights." The teams that close that gap - not by acquiring more data, but by asking sharper questions of the data they already have - are the ones whose day one actually becomes day one. Smith's provocation is the right place to start: "Ask this question: Who have we not listened to?" In pharma launches, the answer to that question almost always identifies the signal the launch plan is missing - and the patient population still waiting for access to a therapy that already exists. 

To get you highlights of Pharma USA 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 Customer Engagement USA 2026 (October 27-28, Philadelphia) - where commercial, marketing, medical, data and AI pioneers converge. Explore the agenda here.