The Third Thing
Data tells us what happened. AI is getting frighteningly good at telling us what will happen next.
Neither of them can change what happens.
That's the part I keep coming back to, because our industry is currently investing enormous amounts in the first two and treating the third as a nice-to-have. Only creativity changes outcomes. Not because it's magic, but because feeling something is what precedes doing something, and no dashboard has ever made anyone feel anything.
What we’re actually building
VCCP Media's work with Dr. Karen Nelson-Field found that 1.5 seconds of genuine attention is enough to encode a memory, provided distinctive brand assets are doing their job. Short exposure isn't wasted exposure. [1]
But where those assets aren’t working, brands levy what the research calls a distinctive asset tariff on themselves: up to 66 cents wasted in every dollar of digital spend. Invisible on every dashboard you look at. And it’s a creative problem, not a media one, which is why buying smarter won’t touch it.
That reframes what we’re doing when we make an ad. We aren’t delivering a message. We’re building a memory structure that has to survive months of nothing, and then fire at the exact moment someone is sitting across from their physician trying to remember why they wanted to ask about this.
That’s a craft problem. It’s also the only part of this job that AI can’t do for us.
Why our audience is harder than most
Here’s what makes healthcare different, and it isn’t the regulation.
In a consumer category, the person at second zero is indifferent. Not for you, not against you. Scrolling.
In ours, they arrive already unsure who to believe. Trust in your own doctor fell from 93 percent to 85 percent in two years, and trust in the FDA fell from 65 percent to 53 percent over the same period. [2] These are not fringe numbers. They describe the ordinary person in the waiting room, and brands are further from the centre of that circle than either the physician or the regulator.
But the finding that rearranged my thinking is smaller and barely gets quoted. A third of adults now take health questions to an AI tool, which puts it level with social media as a health information source. [3] Speed is the main reason people give. More than a third, though, say it is because they feel more comfortable asking privately, and that is the number worth sitting with.
I want to be careful here, because this is not a story about doctors failing. Clinicians are doing extraordinarily complex work inside fifteen-minute appointments, and the privacy people say they are looking for is a function of a system under strain, not of anyone’s character. It’s also worth saying plainly that a chatbot which is warm and confidently wrong is not an improvement on anything.
But it does tell us something about the mindset we’re designing for. Being right is not the same as feeling received. And that gap is not a data problem. Data can tell you it exists. It cannot close it.
CHESS, and what it’s really for
VCCP’s shorthand for what breaks through in those opening seconds is CHESS: character, humor, emotion, surprise, sound.
In consumer categories these get read as attention levers. In Pharma, they’re doing something more specific: each one is a signal about who is speaking and how. Which is precisely what a guarded audience is screening for before they’ve processed a single claim.
Character is the most underused asset we have, because we’ve confused protagonist with spokesperson. A character is someone who comes back. Ours is usually an anonymous patient-actor recast next cycle. Bring the same person back three times and you have an asset. Do it once and you have casting.
Humor is almost entirely absent, and the reason given is always that the condition is serious. That confuses the subject with the tone. Anyone who has spent an afternoon in a patient community knows people living with a diagnosis are frequently, startlingly funny about it. That humor is unmistakably not the voice of an institution, which is what makes it work.
Emotion is the one we think we’re good at and mostly aren’t, because what we reach for is uplift. Resolution, sunlight, the moment after. Uplift is the category’s factory setting and recognizable across a room in half a second. Specificity is what lands: the small true detail that makes one person go, quietly, that’s me.
Surprise is structurally the most valuable lever we have, because it’s by definition the thing a guarded viewer has no file for. Someone who can’t immediately classify what they’re looking at hasn’t yet decided to dismiss it.
Sound is the biggest waste and the cheapest fix. Sonic identity is a real distinctive asset, which is why the research strips it out when testing. Ours is a shared library of the same reassuring piano, which means our sonic asset belongs to the category rather than to any of us. Ask Massive Music, sonic branding builds brands and deeply enters sensorial association.
Run the test yourself. Strip the logo, the ISI, the corporate blue, the piano from your last campaign. Could anyone tell it was yours?
None of this works without the insight
I want to be clear that CHESS is not a licence to be interesting for its own sake. Distinctiveness without insight is just noise with better production values, and our category has plenty of that already.
The levers only work when they’re aimed at something true about how a specific person is actually thinking. What they’re frightened of at 2am. What they’ve already been told by four other people. What they’re not saying in the consultation because they don’t want to seem difficult. That understanding is the raw material. Everything else is delivery.
Which is where data earns its place, and it’s a serious one. It maps the mindset. It tells us where the resistance sits and who is carrying it. AI can increasingly tell us how that will move next quarter.
Then somebody has to make something that changes it. That’s the part that has never been automated and, I’d argue, can’t be, because the thing we’re trying to shift isn’t an input in a model. It’s how someone feels about a decision they’re frightened to make.
What it’s for
In January 2020, 32% of Americans had a positive view of the pharmaceutical industry. By February 2021 it was 62%. By the end of 2025 it was 43%. [4]
I find that trajectory hopeful rather than depressing. It happened once, inside living memory, and most of us were there for it. Trust moved thirty points in thirteen months because for one extraordinary stretch people could see what the science actually does.
We have that science. It’s real, and in most of our categories it’s the most useful thing in the room. What we haven’t consistently had is work that makes anyone feel it.
That’s the whole job, and it’s worth doing properly. Not because a better ad builds a better brand, though it does. Because someone books the appointment they’ve been putting off for eight months, and asks the question they were too embarrassed to ask, and the science finally gets to do what it was developed to do.
Data can tell us that person exists. AI can tell us how likely they are to move.
Only the work can move them.
u0026nbsp;
u0026nbsp;
References
Hacking the Attention Economy, VCCP Media with Dr. Karen Nelson-Field and Amplified. 20,000+ views of 72 ads tested in original and "bad twin" versions.
KFF Tracking Poll on Health Information and Trust, January 2025. Nationally representative sample of 1,310 US adults, fielded January 7 to 14, 2025.
KFF Tracking Poll on Health Information and Trust: Use of AI for Health Information and Advice, March 2026. Nationally representative sample of 1,343 US adults, fielded February 24 to March 2, 2026.
The Harris Poll, US pharmaceutical industry positive perception tracking. January 2020 (32%), February 2021 (62%), end of 2025 (43%).