TL:DR

Gen Z discovery has two new front doors. TikTok is now a mainstream search surface for them, a clear majority use it to search, and it's a $15B+ US shopping channel that collapses discovery into purchase,though, read honestly, it complements Google rather than replacing it. And AI answer engines have scaled to ~900M weekly users, with nearly half of Gen Z discovering brands inside ChatGPT and over a quarter starting their search there. The catch: an AI answer has one slot, not ten. Visibility didn't get less important. It got concentrated. Show up at both doors, or be invisible at the moment of choice.

Two New Front Doors

In an interesting conversation recently, someone asked me where I'd put my attention if I were building a brand for the next generation of buyers. My answer was simple, and I'll stand behind it here, with the data: for Gen Z, two new front doors have opened on discovery, TikTok and the AI answer engines, and a brand that isn't present at both is increasingly invisible at the exact moment a young buyer is forming an opinion.

I want to make that case properly, including the parts the hype gets wrong. Because the strongest version of this argument isn't “Google is dead”, it's more interesting, and more useful, than that.

Start with what hasn't changed

Every brand has the same two needs it had fifty years ago: to be seen, and to be remembered. Visibility and attention. Byron Sharp's name for the second one is mental availability, being easy to bring to mind in a buying situation. That north star doesn't move. What moves, every decade, is the machinery in the middle: the path from an impression to a transaction, and the surfaces where the impression happens at all.

For Gen Z, that surface has split. The single front door of “rank on Google” has become at least three: traditional search, social-video search, and the AI answer. Two of those three barely mattered three years ago.

Front door one: TikTok is a search engine now (but read the data honestly)

The headline most marketers half-remember is from 2022, when a senior Google executive then overseeing Search said publicly that almost 40% of young people, looking for somewhere to have lunch, go to TikTok or Instagram rather than Google. It was true, and it was instantly over-read into “Gen Z abandoned Google.”

The more recent, more honest picture is better for strategists because it's more precise. Independent data from Adobe, a January 2026 survey of around 800 US consumers , found that 65% of Gen Z have used TikTok as a search engine, and that roughly half of all consumers (about 49%) now have, up from 41% two years earlier. (A separate study commissioned by TikTok put weekly Gen Z search usage as high as 86%, close behind traditional search , treat that one as directional given who paid for it, but the direction is the same across both.)

Here's the nuance that actually matters, and that the “TikTok killed Google” crowd misses: the share of Gen Z who prefer TikTok over Google is small and falling , from about 8% in 2024 to around 4% by 2026, per Adobe's tracking. So this is not a replacement story. It's an and story. Gen Z runs different searches on different surfaces: TikTok and Instagram for visual, social, “show me what's actually good” discovery; Google for the factual and transactional. The job isn't to bet on one. It's to be present where each kind of search happens.

And TikTok has quietly done something Google never did: it collapsed discovery and purchase into one surface. TikTok Shop's US gross merchandise value reached about $15.1 billion in 2025, up roughly 68% year over year from around $9 billion in 2024 (Momentum Works; some trackers put it higher still). When the place you discover a product is also the place you buy it, the distance from impression to transaction shrinks to a single tap , and the brand that owns that moment of discovery owns a disproportionate share of the sale.

What earns visibility there is not a banner. It's content people actually search for: about a third of TikTok users say they're prompted to search by something they saw on the platform, and the formats they want are video tutorials, product reviews, and creator recommendations , not ads dressed as ads.

Why creator recommendations carry such weight: a behavioural-science detour

It's worth pausing on why a recommendation from a person outperforms a message from a brand , because the answer is older than marketing, older even than humans.

In a 2005 study at Duke University Medical Center, neuroscientists Robert Deaner, Amit Khera and Michael Platt gave male rhesus macaques a simple choice: take a juice reward, or give up some of that juice for the chance to look at a photograph of another monkey. The monkeys paid, they sacrificed a reward they actually wanted, to look at the faces of high-status, dominant members of their group (and at images of female hindquarters). To get them to look at low-status monkeys, the researchers had to pay them extra. Looking at the powerful, in other words, was so valuable to the primate brain that it was worth giving up juice for. (Deaner, Khera & Platt, “Monkeys Pay Per View,” Current Biology, 2005.)

We run much the same wiring. Humans have an evolved bias to find high-status individuals magnetic, to watch them, attend to them, and copy them. Anthropologists Joseph Henrich and Francisco Gil-White called this prestige bias: because we can't know in advance who is worth learning from, we use a shortcut, we learn from the people other people are already paying attention to (Henrich & Gil-White, “The Evolution of Prestige,” Evolution and Human Behavior, 2001). Attention itself becomes the signal of who to trust. If that sounds like a follower count, that is precisely the point: the visible metrics on a creator's profile are an industrial-strength version of the cue our ancestors used to choose who to learn from.

So an influencer is not a new kind of advertising. They are a very old kind of person, the high-status, widely-attended-to model the brain is built to learn from, wearing a modern interface. And a second mechanism stacks on top. As far back as 1956, sociologists Donald Horton and Richard Wohl observed that mass media let audiences form one-sided bonds with performers that feel like real friendships — “intimacy at a distance,” they called it, coining the term parasocial interaction (Psychiatry, 1956). A Gen Z viewer doesn't just watch a creator; across hundreds of hours of casual, face-to-camera video, they come to feel they know them.

Put the two together and you have why the creator beats the brand at the moment of choice. The creator is at once a prestigious model the brain is wired to copy and a trusted friend it is wired to believe. Two of our oldest decision shortcuts, learn from the high-status, trust the people close to you,fire at the same time, on the same screen, in the seconds before a purchase. A brand logo triggers neither.

This is also why authenticity isn't a nice-to-have; it's the load-bearing wall. Prestige, in Henrich and Gil-White's words, is freely conferred, deference an audience grants, not attention a brand can buy. The instant a partnership reads as bought rather than believed, the brain reclassifies the creator from “trusted model” to “salesperson,” and the effect doesn't just fade, it can invert. Which loops straight back to the discipline: find the creators who are genuinely part of a community, not the ones who will rent you their feed, and keep all of it anchored to the sale.

Front door two: the AI answer engine

The second new front door didn't exist for consumers three years ago, and it has scaled at a pace with few precedents in consumer technology. ChatGPT reached roughly 800 million weekly active users by late 2025 and about 900 million by early 2026 (OpenAI). Whatever the precise figure this quarter, the direction is unambiguous: a meaningful slice of human “how do I / what should I buy” questions now goes to an AI assistant first.

Gen Z is leading that shift into commerce. Per Adobe's May 2025 survey, 47% of Gen Z say they've discovered a new product or brand inside ChatGPT, and 28% now start a product search in ChatGPT before they ever open a search engine. Separately, about one in three (33%) say they prefer AI platforms over any other channel for product research, ahead of Amazon, retailer sites, and social (Commerce × Future Commerce, 2025). And the behaviour is sticky: a 2025 PayPal holiday survey (Talker Research, ~1,000 US adults) found 61% of Gen Z have used — or considered using — an AI tool to help with a purchase in the past year, the highest share of any generation.

Now the part that should change how you think about visibility. A Google results page gave you ten chances to be seen; if you weren't first, you might be third, or recoverable on page two. An AI answer gives you one. The model synthesises a single response and names a handful of brands. If yours isn't in that answer, there is no page two, you simply don't exist for that query, with no second chance to be considered. Scarcity of slots is the whole story. Visibility hasn't become less important; it has concentrated.

This is why I treat LLM visibility, generative engine optimisation (GEO) ,as a real discipline, not a buzzword. In Sharp's language, mental availability now has to extend to the machine's “mind”: being the brand the model recalls and the feed surfaces, not only the human. Practically, that means understanding how these engines build answers — from their trained, “parametric” memory and from live, retrieved sources — and earning your way into both: the third-party evidence, the structured and authoritative content, the presence in the places these models read.

Why this supports the bigger strategy, not replaces it

None of this throws out the playbook. It relocates it. The chain I build everything around is unchanged — visibility → attention → familiarity → preference → transaction — and it's still backed by decades of evidence (Binet and Field on share of voice running ahead of share; Ehrenberg-Bass on the steady share erosion when brands go dark). What's changed is where each link now happens for a sixteen-to-twenty-five-year-old: the impression lands inside a creator's video or an AI's answer; the attention is won or lost in the first second of a scroll; the purchase closes in an in-app checkout that sits millimetres from the discovery.

What this means for a brand , and why FMCG is the hard case

So what does a brand actually do with this? Here's where it gets more demanding than “post on TikTok and optimise for ChatGPT.”

Being unmissable on TikTok and being the brand the model names are both, in the end, mental availability plays. They make you easy to think of; they build the memory. That's half the equation, and it's the half social and AI are brilliant at. But Byron Sharp's other market-based asset is physical availability: being easy to buy. The two only convert when they meet. A brand that's top-of-mind but not to-hand loses the sale at the last centimetre; a brand that's everywhere on shelf but absent from the mind never gets reached for. (One honest nuance: TikTok Shop blurs the line, adding a buyable layer to the social surface. But for a chilled impulse drink, the decisive availability is still real-world , at the exact moment of thirst.)

This is the heart of the FMCG problem, and nobody knows it better than Coca-Cola, whose own founding distribution doctrine, Robert Woodruff's instruction that Coke be “always within an arm's reach of desire”, is the most elegant definition of physical availability ever written. Mental availability creates the desire; physical availability has to put the product within arm's reach at the instant the desire fires, or it evaporates.

Here's the trap for Gen Z. “Within arm's reach” used to be solved by classic distribution: be in every supermarket, every Späti, every Rewe and dm. That still matters, and it still wins the planned shop. But a large share of this generation's consumption, and almost all of its discovery , happens not in the aisle but at the occasion: the festival, the gig, the gaming session, the campus, the run club, the moment a community gathers. If you're mentally available (they want you) but physically absent from where they actually gather and consume, you've built the desire and handed the sale to whoever is in the cooler at that moment.

So physical availability, for Gen Z, has to be mapped to community and culture, not just to retail. Coca-Cola itself points the way in public, and it's worth being precise about the two moves, because they do different jobs. Its global Coke Studio platform (relaunched 2022) makes music the brand's Gen Z passion point, that's largely a mental-availability play, building cultural presence and memory. The on-the-ground festival activations, Lollapalooza, Bonnaroo and the like, are where physical availability meets the occasion: the brand buyable and chilled inside the cultural moment, not on a banner beside it. The platform builds the wanting; the activation puts the cold can within arm's reach where the wanting fires. Both halves, meeting where Gen Z actually gathers.

So how do you map those communities, the part that's genuinely hard? A repeatable method, all of it from tools a brand already has:

• Listen before you map. Use social listening and the creator graph to find where your category's conversation actually lives — which subcultures, communities and creators anchor it. Don't guess the passion points; let the data show them (music, gaming, football, streetwear, wellness — each skewing to different cohorts).
• Translate online communities into offline gathering points. Every digital community has physical anchors — festivals, esports arenas, campuses, local scenes, run clubs. List them. These are your new “shelves.”
• Overlay purchase geography. Layer first-party and retail/geo data to find where Gen Z purchase density and those communities overlap. That intersection is where sampling, chilled availability and activation earn the most return, it turns “community marketing” from a vibe into a distribution decision.
• Show up as a participant, not a sponsor. Here's where the behavioural science from earlier bites: presence only builds equity if it's freely conferred, co-create with the community, don't rent a logo slot. The instant it reads as bought, the prestige effect inverts.
• Prove it. Treat each community activation as a geo experiment , hold out matched regions, measure the lift in availability-driven sales with marketing-mix modelling. That's what turns “be where Gen Z gathers” from a nice idea into a defensible line in the plan.

The takeaway for any FMCG brand , and most of all for the one that wrote the original availability playbook: mental availability gets you into the consideration set; physical availability at the level of community and occasion is what closes the sale. For Gen Z, the arm's reach of desire no longer runs only through the supermarket. It runs through culture, and the brands that win will be the ones that map it, show up in it authentically, and can prove the cooler was full when the moment came.

Sources

• “Almost 40% of young people, when they're looking for a place for lunch… go to TikTok or Instagram” , Google SVP Prabhakar Raghavan (Knowledge & Information org, incl. Search), Fortune Brainstorm Tech, July 2022 (cited US 18–24, dining-specific; underlying study never published). Multi-platform nuance: a SOCi survey of 1,002 US 18–24s found near-parity for local business search — 67% Instagram / 62% TikTok / 61% Google.
• 65% of Gen Z have used TikTok as a search engine; ~49% of all consumers (up from 41% in 2024) — Adobe (Jan 2026 survey, 807 US consumers, via Adobe Express / Search Engine Journal). Preference for TikTok over Google fell from ~8% (2024) to ~4% (2026) — same Adobe data.
• A separate TikTok-commissioned study (via Tubefilter / WARC, 2025) reported weekly Gen Z search usage as high as 86% — treat as directional; the source has an interest.
• TikTok Shop US GMV ~$15.1B in 2025, +~68% YoY from ~$9B in 2024 — Momentum Works (some trackers report higher, e.g. ~$15.8B / +108%; figures vary by methodology).
• ChatGPT ~800M weekly active users (late 2025) → ~900M (early 2026) — OpenAI / TechCrunch. (Some trackers estimate ~1B by mid-2026 — treat as estimate.)
• 47% of Gen Z discovered a new brand via ChatGPT; 28% start product searches in ChatGPT — Adobe (May 2025 consumer survey, ~1,000 respondents).
• ~1 in 3 (33%) Gen Z prefer AI platforms over other channels for shopping / product advice — Commerce × Future Commerce (Sept 2025).
• 61% of Gen Z have used or considered using an AI tool to help with a purchase in the past year (highest of any generation; 40% overall) — PayPal 2025 Holiday Shopping Survey (Talker Research, 1,000 US adults, Sept 2025).

Behavioural science

• Deaner, R., Khera, A. & Platt, M. (2005), “Monkeys Pay Per View: Adaptive Valuation of Social Images by Rhesus Macaques,” Current Biology 15, 543–548 (Duke University Medical Center). Male macaques forwent juice to view high-status faces and female hindquarters; required overpayment to view low-status monkeys.
• Henrich, J. & Gil-White, F. (2001), “The Evolution of Prestige…,” Evolution and Human Behavior 22(3), 165–196 — prestige bias / learning from whom others attend to.
• Horton, D. & Wohl, R. (1956), “Mass Communication and Para-Social Interaction: Observations on Intimacy at a Distance,” Psychiatry 19, 215–229 — origin of “parasocial interaction.”

Foundational frameworks (own thesis): Byron Sharp, How Brands Grow (mental + physical availability); Binet & Field (share-of-voice → growth); Ehrenberg-Bass “When Brands Go Dark”

Unsubscribe