Expert voices and AI discovery: What think tank comms teams should know

Recent analysis* of roughly 700,000 social-platform citations in ChatGPT responses found that on LinkedIn, personal profiles (25.4%) and individual posts (25.1%) together account for just over half of citations. Company pages, by contrast, account for around 18%. (*Source: Profound analysis of ChatGPT social citations, 2025)

That’s a big difference, and it suggests that when models look for authority in a field, they are far more likely to surface named individuals than institutional pages. 

Authority has always been human

Think tanks and research organisations have always traded on expertise carried by named individuals — fellows, directors, senior researchers. Their judgement and clarity make the institution legible.

What’s changed is that AI systems now operationalise those signals.

When someone asks a model about industrial policy, AI governance or climate finance, it synthesises from across the web. It looks for signals of authority, clarity and consistency. And increasingly, those signals are attached to people.

If your executive team appears only for occasional announcements, or maintains sparse, generic profiles, you are under-represented in one of the fastest-growing discovery environments.

The goal here isn’t to turn leaders into influencers; it’s about recognising that executive visibility is now part of your communications infrastructure.

Citation isn’t the same as trust — but it shapes perception

It’s worth being clear-eyed here. As we all know, LLM citations are not guarantees of accuracy. Research into generative search has shown misattributions and fabricated links are still common. Nor does inclusion necessarily drive traffic — often the value lies in being present in the answer itself. But presence does matter.

If a policymaker, journalist or adviser asks an AI system a question in your field, the model is likely to present a cluster of voices. That cluster becomes a proxy for “who counts” in the conversation.

Consistent presence in that space doesn’t happen by accident; it requires deliberate design.

At its core, this is about trust

In our work exploring how public-interest communications may evolve over the next decade, one theme keeps resurfacing: trust becomes harder to earn, and human signals of credibility become more important.

In an LLM-shaped information layer, trust is assembled from visible patterns:

• Consistent expertise over time
• Public reasoning, not just outputs
• A clear, accountable voice

For think tanks, research and policy orgs those signals sit naturally with senior leaders and subject-matter experts.

This is where comms teams play a vital role: helping leaders show up in ways that are coherent, safe and sustainable.

What comms teams can do now

The challenge is rarely persuading leaders that LinkedIn matters (although we know that can be a challenge!). It’s making participation structured and manageable.

A workable approach looks like this:

• Define narrative focus: Each executive or expert should have 3–5 clear themes they consistently return to. This repetition builds clarity — for humans and machines.

• Treat profiles as landing pages: Headlines, summaries and featured work should signal expertise immediately. (And a reminder that this is about positioning, not biography!).

• Build light editorial scaffolding: Provide prompts drawn from existing research:

— One chart, one implication
— Behind the paper
— What we mean when we say…

Ultimately, consistency matters more than volume.

• Put governance in place: Clear guardrails on sensitive topics, embargoes and corrections reduce risk and increase confidence.

• Measure patterns, not spikes: Track profile views, engagement and — periodically — whether your organisation’s voices appear in AI answers to core sector questions. Look for repetition over time, not one-off mentions.

A realistic pilot

For most organisations, this doesn’t require a full strategy overhaul. We’d suggest starting with a 90-day pilot:

• 3–5 priority leaders
• One post per week (or fortnight)
• A shared content bank built from existing outputs
• A simple review loop
• A monthly check-in on visibility and confidence

Key takeaway

In many ways, visible expertise has always mattered for think tanks. Ideas travel through people as much as publications. What’s changing is the extent to which digital systems — including LLMs — interpret organisations through the voices and expertise they can clearly identify.

That creates an opportunity. When leadership, research and editorial content are clearly expressed and easy to navigate, organisations are more likely to be understood — and surfaced — in these environments.

For think tanks with substantial bodies of work, the question becomes less about publishing more and more about how expertise is structured, expressed and connected across your digital ecosystem.

If you’re thinking about how your organisation shows up in AI-mediated discovery — from expert visibility to editorial strategy and content structure — we’d be glad to explore it with you.