S2 Ep8: AI Discoverability Is a Leadership Problem - Not a Content One

Transcript

Transcript

The following is a lightly edited transcript of the A Splice of Life Science Marketing podcast episode recorded 4 March 2026. Matt Wilkinson and Jasmine Gruia-Gray discuss AI discoverability in the life science sector - what the data shows, why most companies haven't acted, and why this is fundamentally a leadership challenge rather than a marketing execution problem.

Setting the Scene: What the Data Actually Shows

Speaker: Jasmine [00:41]

Hey, Matt.

Speaker: Matt Wilkinson [00:42]

Hi Jasmine, how you doing?

Speaker: Jasmine [00:44]

Good, good. Everything's with you.

Speaker: Matt Wilkinson [00:46]

Very good, thank you.

Speaker: Jasmine [00:48]

Okay, so today we have another special episode on what has been keeping up a lot of marketers these days. Life science companies are being sold - in quotes - answer engine optimisation, AEO, as the next frontier of marketing investment. You just wrote an article titled "Separating Signal from the Snake Oil in AI Search" - and I love this title - and it cuts through the hype by grounding the debate in Google's own statements and independent data.

In August 2025, Google's John Mueller publicly described the proliferation of AEO and GEO acronyms as spam and scamming. Danny Sullivan, on the other hand, Google's search liaison, was more measured but reached the same conclusion: these are subsets of SEO, not new disciplines. The rebranding is largely commercial, and buyers should beware.

The underlying shift is real. AI-generated overviews now appear in 18 to 25% of Google queries. And when they do, organic position one results lose 34.5% of their clicks. LLM referral traffic grew 155% in eight months versus 24% for traditional search. The trajectory matters more than the current 0.77% share of desktop activity. The structural disruption is visible in publisher analytics now, not in projections.

The commercial stakes for life science tools companies are sharpened by three recurring patterns from Strivenn's own exhibitor research across ELRIG Drug Discovery 2025, where we surveyed about 107 exhibitors, and SLAS 2026, where we surveyed 43 exhibitors. 62% of exhibitors had never asked an AI to recommend companies in their category. 44% had AI tools without any structured adoption programme. And AI recommendation lists consolidate to roughly five brands per B2B category - a dynamic the article names citation compression. The window to establish AI visibility is closing, and most of the sector has not yet looked at whether they are in or out. So Matt, what's going on?

The Sector-Wide Blind Spot

Speaker: Matt Wilkinson [03:36]

Yeah, so it's really interesting, isn't it? On the one hand, we have this really important set of new channels. And I won't just call them one channel. It's like AI search is really a bit like calling it social media - where you've got the likes of ChatGPT, you've got Perplexity, you've got Gemini, and you've got Google's own search within the Google window that uses Gemini as well. So you've got a lot of these new answer engines, if you will.

And what really is staggering to me is that 62% of exhibitors at the booths had never once asked an AI to recommend companies in their category. They had never bothered to look - do we appear? That's a real sector-wide blind spot to me. People aren't yet waking up to the fact that if you're not part of that search, if you don't appear in those five mentions, you're not necessarily going to be included in anybody's thinking. It's not like Google where there are ten, a hundred pages that you can go through. There are just five answers that get surfaced, and if you're not in those, you may not ever get that call.

The Commercial Cost - Is the Crisis Framing Justified?

Speaker: Jasmine [04:40]

I have the same experience. Some folks had a really blank look on their face when I brought this up. Others said, that's a really good idea, I'm going to try that as soon as you leave the booth. But the overwhelming majority, as the data show, have not looked at this.

But what's the actual commercial cost? Life science purchase decisions don't happen because a researcher typed a query into their favourite LLM and bought whatever appeared first. They happen through application specialists, distributor relationships, conference conversations, and sales cycles that span months. The channel driving this urgency is still less than 1% of desktop web activity.

Speaker: Matt Wilkinson [05:08]

It's hard to disagree with that for sure. And particularly for capex investments, absolutely, people are going to want to make sure that they're surveying everything.

But I think that that's missing the point. Those conversations happen because you already know something exists - when people already have brand affinity or they know about the brand, what Mark Schaefer would call the brand override. The framing you provided treats AI discoverability as a traffic channel, and it isn't.

As I mentioned, if you rank fifth on Google, you're still in consideration - people are still seeing your brand. A buyer can still find you even though they have to scroll down the page a little bit further. If AI doesn't include you in a category recommendation, the buyer may never know that you exist at all. And that's a categorically different problem. I think that's where this is really, really important to recognise - just how important this is. And just to see the growth of these engines is absolutely staggering.

Speaker: Jasmine [06:28]

So I completely agree on the binary nature of this. But here's a different perspective. Among the exhibitors who actually did run the test, 75% found themselves listed, or at least claimed that they were listed. So the unconsidered set is mostly made up of companies that haven't looked, not companies that are genuinely absent. The crisis framing may be doing more work than the data warrants.

Speaker: Matt Wilkinson [07:01]

Hard to disagree with. The 75% figure cuts both ways. It means that 25% who looked found they were not listed. And the ones who never looked - what's their assumption? They assume they're fine, or do they just not even think of this yet? The commercial cost of that assumption is deals you are not losing because you were never in the running. There are no losing signals. You just lose an RFP request to silence. So I think the fact that you don't know what you're missing out on - that's really the key challenge here.

Speaker: Jasmine [07:33]

Yeah, it's kind of like that Rumsfeld statement - the known knowns and the known unknowns and so on.

Speaker: Matt Wilkinson [07:41]

It's definitely an unknown unknown. If you're not looking and you're not featured, you're an unknown unknown, and that becomes a very wicked problem to solve.

The Budget Question - Is the Investment Justified?

Speaker: Jasmine [07:48]

I agree. That is a compelling way to frame it. But I keep coming back to the budget question. A 20 to 30% uplift to do this properly is not trivial for a lot of life science marketing teams already stretched across conference season, content production, CRM, and digital marketing. The marginal hours spent on Wikidata entity consistency is probably worth less than the marginal hours spent on well-targeted campaigns to an existing segment.

Speaker: Matt Wilkinson [08:25]

Hard to disagree, particularly if you're measuring direct input hour for hour. But that does assume that the Wikidata work doesn't compound. That's the citation compression argument. The brands establishing AI visibility now are setting the baseline models default to for future years. Training data is not refreshed continuously. Early absence does not create a gap you can easily close.

Life Science's Structural Advantage - and Why Most Can't Access It

Speaker: Matt Wilkinson [11:07]

I actually think that life science companies have an unfair advantage in AI citation - or at least many do. Those that have large amounts of data or peer-reviewed data that feature their products and services, mentioned in methodologies, that gives them a really, really good advantage. Peer-reviewed publications, named scientists with verifiable institutional affiliations, clinical validation data with specific results - that's all gold for AI engines.

An analysis of 36 million AI overviews found NIH content accounting for 39% of citations in health and science queries. Social platforms barely registered for those same queries, partly because the conversations that scientists are having are happening elsewhere - on the NIH, in peer-reviewed journals. That's where those conversations happen. It's letters to the editor. That's where the high-value conversations happen. Social platforms may be talking about tools and brand affinity, but they don't talk about the real science.

Speaker: Jasmine [12:10]

I can't dispute that advantage exists. What I often think about is whether organisations can actually access it. Some of these high-impact journals - Nature, Science, Cell - are all gated. The reason content is ghostwritten and unattributed is not because life science marketers are unaware of attribution's value. It's because named scientists are busy or reluctant to engage with marketing processes. They want to be seen as independent, not as shills for a particular company, and they operate under institutional guidelines about what they can claim publicly. That's not a content strategy problem. That's an organisational change management problem.

Speaker: Matt Wilkinson [13:03]

Fair, but bear in mind the open access movement is making things a lot more visible to a lot of people anyway. The gating issue is more solvable than the attribution issue. We don't have to gate content. We don't have to put our white papers and our deep scientific information in PDFs behind a gate that AI can't access. We can make that available on the web. We don't need to hide things behind a form. AI can't cite it if it's there, and it doesn't know it exists if it's behind a form.

So we need to create ungated structured summaries. We need named authorship, schema markup. Those are really, really important things to do. And there are tools that help with this - I'm a fan of a tool from a small developer called Super Schema. The AIs will generate and push data into the header of pages to help share content at as low a barrier as possible. That's easy to do. It's the sort of thing we need to start thinking about. Not just: we need to do SEO, we need to do AEO, we need to do GEO. We need to think about how do we make our materials genuinely available to all of these new channels - this entire new marketing discipline.

Speaker: Jasmine [14:30]

Those are things that we can control, right? The content on our website, we can control. The content on these high-impact journals, we can't. That's how their subscription model works - some of it is ungated, but the large majority is gated for subscribers.

That is one aspect that I think needs to be solved in this industry. Another aspect is: assuming legal signs off, assuming the scientist agrees to be named, assuming HR, communications, and sales can agree on a single company description - the approval cycle alone is soul-crushing. It can be upwards of a month, which is unrealistic for any organisation of meaningful scale. You're describing the output as if the input is trivial.

This Is a Leadership Decision, Not a Marketing Project

Speaker: Matt Wilkinson [15:35]

So your argument is that the structural inertia is just too deep to overcome?

Speaker: Jasmine [15:44]

The barrier is misdiagnosed. If you frame this as a content optimisation project, you'll get incremental progress and hit a ceiling. If you frame it as a cross-functional mandate - one senior commercial leader who owns AI discoverability and has the authority to move compliance, science, and marketing in the same direction - then you might get somewhere. Most organisations haven't made that decision yet.

Speaker: Matt Wilkinson [16:14]

That means the window is open for the ones that do. The competitive dynamic isn't life science versus other industries. It's the life science companies that solve the internal coordination problems first versus the ones that don't. That advantage is real. The question is who captures it first.

Speaker: Jasmine [16:33]

And that comes down to leadership - not necessarily a marketing question, but leadership and strategy, which is probably why it's not being answered quite yet.

Speaker: Matt Wilkinson [16:46]

That's hard to dispute. Raising AI literacy, AI maturity, and AI discoverability as a leadership conversation is absolutely crucial. The unconsidered set - it's not a knowledge gap. It's a prioritisation gap. And it's a leadership prioritisation gap. The companies that never ran the test may never have thought about it, or maybe they assumed they were fine. Some of them will be right. But a quarter aren't there. And the ones who were right today may not be right in 12 months' time. Because citation compression does not pause when you decide whether to take it seriously or not. It keeps going. And the brands building AI visibility now are not just winning referrals - they're setting baseline model defaults that compound. The baseline gets harder to displace the longer it runs.

Speaker: Jasmine [17:40]

But there's also a deeper problem. Life science companies have a structural AI citation advantage that no other industry can replicate - peer-reviewed content, named experts, verifiable institutional authority, the exact signals AI models weight most heavily in technical domains. But the same culture that produces that advantage - compliance review, distributed ownership, scientists who answer to research directors, not marketing managers - is precisely why most organisations can't access it. The barrier is not awareness, it's accountability. And nobody owns this right now.

Speaker: Matt Wilkinson [18:25]

And I guess that's marketing's job. It really has to be our job to elevate this from being a content strategy question to being a leadership question. The companies that will win AI discoverability in life science are not the ones with the best SEO or the most diligent Wikidata updates. They're the ones where a senior commercial leader decides that AI visibility is a cross-functional mandate, names someone accountable for it, and gives that person the authority to move compliance, science, and marketing in the same direction. Most organisations haven't made that decision yet. Most aren't even thinking about it yet.

The ones that do are not just solving a discoverability problem - they're building a structural position their competitors will spend years trying to close. The question is not whether to act. It's whether your organisation has both the authority and the incentive to make it happen.

Speaker: Jasmine [19:16]

So maybe a good place to end here is a call to action to all those leaders listening - or those listening who have leaders who aren't quite grappling with this issue yet - to get on it. This is your opportunity. Leveraging it now is going to be really critical to the future of your business.

Speaker: Matt Wilkinson [19:39]

Absolutely. Fascinating to dig into this, and thank you for the conversation, Jasmine.

Speaker: Jasmine [19:44]

Thank you, Matt. See everybody again soon at the next episode of A Splice of Life Science Marketing.

Q&A

How do I actually check whether AI recommends my company right now?

Open ChatGPT, Perplexity, and Gemini. Type: "Which companies provide [your category] for [your application]?" Run it three to five times across different phrasings. Screenshot every result. If your brand doesn't appear consistently, you have a baseline problem. This takes under an hour and costs nothing. Do it before any other AI discoverability work.

We ranked well in the test - does that mean we're fine?

Not necessarily. Citation compression means AI recommendation lists consolidate over time as models reinforce existing training data. Appearing today doesn't guarantee appearing in 12 months, especially as competitors invest in structured, ungated content. The question isn't where you are now - it's whether you're building the signals that make your position durable.

What's the single highest-leverage action a stretched marketing team can take this week?

Audit your gating. Pick your three most commercially important content assets - white papers, application notes, validation data. If they're behind a form or locked in a PDF, AI can't cite them. Move one to an ungated HTML page with named authorship and basic schema markup. That single change increases AI accessibility more than any optimisation tactic.

How do I make the case to leadership that this is their problem, not marketing's?

Run the AI visibility test first. Then show the result alongside your competitive set. Frame it as a pipeline risk question, not a marketing question: "If a buyer asks an AI to shortlist suppliers in our category, are we in the five? If not, what's the cost of that absence per quarter?" Leadership responds to revenue risk framing. Content strategy framing will get you a content budget, not a mandate.

Who should own AI discoverability in our organisation?

Someone with authority to move compliance, science, and marketing simultaneously - typically a Chief Commercial Officer, VP of Marketing with senior backing, or a dedicated cross-functional lead. The role doesn't need to be new, but it needs explicit ownership and accountability. Without that, AI discoverability stays a marketing project and hits the approval-cycle ceiling Jasmine describes. Name the person. Give them the authority. Start there.