By Sean Pollard, VP Insights & Analytics
Over two days at the AMEC Global Summit in Dublin earlier this year, communications, measurement, analytics, and technology leaders explored some of the biggest forces reshaping the industry. While artificial intelligence was a central theme running through many keynotes and panel discussions, the conversation extended far beyond AI itself.
What stayed with me after the AMEC Global Summit in Dublin was not simply the pace of change around AI, but the opportunity it gives communicators to make better decisions with the information they already have.
That same direction was reinforced last month in AMEC’s GEO Fundamentals Webinar, which brought one of the Summit’s clearest themes into sharper focus: AI-led discovery should not be treated as a visibility exercise, but as a measurement challenge requiring clear objectives, transparent testing, reliable sources, and a clear connection to trust, behavior, and outcomes.
From my perspective in healthcare communications, three themes stood out.
Measurement must help teams decide what to do next
Communications measurement has long been asked to show what happened after the fact: coverage, impressions, engagement, sentiment, and share of voice. Those indicators still have a role, but AMEC Dublin made clear they are no longer enough on their own.
Leaders need more operational answers: what risks are emerging, where resources should be focused, and what should happen next? This point surfaced across several sessions on measurement, communications intelligence, reputation, and organizational decision-making: measurement creates value when it helps teams answer better questions and support better decisions.
One framing captured the shift neatly: organizations need to move from measurement as validation to measurement as navigation. Validation explains whether work happened and whether it performed. Navigation helps teams understand context, anticipate change, and make better choices while there is still time to act.
The same point came through in discussions about AI ROI, where speakers challenged the assumption that faster work automatically creates value. Faster workflows and more automation are useful only if they improve decision-making, stakeholder outcomes, innovation, risk management, or business performance. In healthcare communications, this is where measurement earns its place: helping teams understand patient and healthcare professional needs, identify reputation or access risks earlier, and advise on the next best action.
One practical application we are exploring is the use of synthetic audiences to support healthcare communications planning before real stakeholder engagement begins. Drawing on qualitative research, media signals, and stakeholder data, AI-built personas representing policy audiences — HTA bodies, payers, advocacy groups — allow teams to pressure-test messages and anticipate objections quickly, before the cost and time investment of live consultation.
The approach only works if the discipline around inputs and validation is maintained. Personas built on thin or unrepresentative data produce plausible-sounding outputs that do not reflect how real audiences behave. Market-specific data, native-language inputs, and expert review at each stage are the guardrails that make the tool reliable.
That is the design principle: AI as a decision-support capability, with human expertise setting the parameters and interpreting the results.
AI is reshaping how audiences discover and interpret information
The second major theme was the rise of AI as a discovery and interpretation layer. Large language models and generative search tools are increasingly sitting between organizations and their audiences, producing synthesized answers that blend, prioritize, and interpret information on their behalf.
One AI visibility session described this as the "era of answers." As AI visibility and generative search become more important, communications teams need to understand how organizations appear in AI-generated responses, which sources AI systems rely on, whether AI-generated narratives are accurate, and how AI visibility contributes to reputation and influence.
A brand-side session brought this challenge into practical focus, emphasizing machine readability, answer-first content, and whether published information can be found, interpreted, and recommended by AI systems. In simple terms, organizations now have to consider how they "market to the machine" as well as how they communicate with people.
One data point shared at AMEC brought this into sharp focus: approximately 84% of citations in AI-generated responses came from third-party sources rather than brands’ owned domains.
That suggests earned media and credible external sources are increasingly shaping the information AI systems draw on to answer questions and form responses. But AI visibility should not become another vanity metric: being present in an AI-generated answer does not automatically mean a brand is trusted, accurately represented, or influencing behavior.
This diagnostic work is now central to our practice. Our GEO audits assess how a brand, medicine, or disease area is represented across major AI platforms, using long, conversational prompts that reflect how patients and clinicians actually ask questions, rather than keyword lists inherited from SEO.
In one recent healthcare engagement, a brand in a specialist therapy area had strong AI visibility ahead of an important evidence milestone, but an audit of AI-generated responses across leading platforms found multiple accuracy issues, including misread clinical evidence and overstated claims. High visibility was amplifying an inaccurate picture.
The strategy shifted from building presence to ensuring accuracy: structuring evidence so AI systems could read, interpret, and cite it correctly, rather than drawing from partial information.
That distinction matters sharply in healthcare. Visibility without accuracy is not an asset. When AI confidently presents an inaccurate clinical picture to a clinician or patient before a correction can reach them, it compounds the risk rather than reducing it.
Trusted signals and human judgment determine value
The third theme was trust. Across discussions on reputation, source quality, misinformation, AI-generated content, stakeholder confidence, and organizational resilience, the conference repeatedly returned to the same idea: more information does not automatically create better intelligence.
One case study illustrated how AI-supported analysis helped identify financial challenges and stakeholder concerns but did not determine the right course of action. The solutions came through consultation, listening, negotiation, and human decision-making. The broader warning was clear: analysis should not be confused with judgment.
Another session made a similar point through the JEEPS framework: judgment, experience, expertise, people, and strategy. As information becomes more abundant and AI makes analysis faster, competitive advantage shifts toward interpretation.
The differentiator is not simply who has the largest dataset or the quickest tool; it is who can understand context, identify what is meaningful, build trusted relationships, and decide responsibly.
That human layer depends on signal quality. The same issue came through in discussions on source intelligence, audience intelligence, narrative intelligence, and data governance: the challenge is not simply volume, but relevance. Speakers also demonstrated how conclusions can change depending on which sources are included in analysis, reinforcing that trust and reputation have measurable consequences for enterprise value, stakeholder behavior, and organizational resilience.
This is how we approach AI-enabled work: the guardrails matter as much as the models. When we build synthetic audiences, personas are constructed from curated, market-specific data in the relevant local language, validated by native-market experts, cross-checked against real-world research, and used to inform stakeholder engagement, not replace it.
When we audit AI visibility, responses are assessed against an evidence base of regulatory, scientific, and peer-reviewed sources, with appropriate medical or scientific review informing accuracy judgments.
In both cases, technology accelerates the analysis; people who understand the science, the market, and the consequences decide what it means and what should happen next.
What comes next
AMEC Dublin did not point to a future where measurement is simply more automated. It pointed to a higher standard for the function itself. Communications and marketing teams will need measurement systems that connect fragmented signals, evaluate AI visibility and source quality, protect trust, and support better decisions.
For healthcare communicators, that means building intelligence that is both technologically enabled and deeply human: fast enough to spot emerging signals, rigorous enough to protect credibility, and thoughtful enough to support responsible decisions in complex stakeholder environments.
The next phase of measurement will not be defined by measuring more things. It will be defined by turning what is measured into better choices. That, for me, is the real shift from reporting to decision intelligence.
AI-enabled approaches have become a fundamental part of how we help teams plan, test, measure, and optimize healthcare communications, from AI visibility audits and synthetic audience testing to source quality, audience intelligence, and decision support.
Interested in hearing more? Connect with us here.