%20png%20copy.avif)
Most B2B startups still build their go-to-market around a buyer they can see. The problem is that the buyer now does most of the work before you know they exist.
By the time a prospect lands on your site, they have often already run the category query through ChatGPT or Perplexity, pulled a shortlist of three or four vendors, drafted an internal comparison, and decided where you sit on it. The shortlist gets assembled in a place your analytics will never show you. You either made the list or you didn't, and you usually find out after the decision that mattered was already made.
This is the part founders underestimate. AI did not just add a channel. It moved the first, shortlist-deciding moment of the buying journey out of your funnel entirely — and then changed who the buyer trusts once they re-enter it.
Here is what actually changed, and where the leverage is for a startup that can move faster than its category incumbents.
The traditional B2B motion assumed a sequence: a buyer becomes aware, visits your site, converts on a form, enters nurture, talks to sales. Every stage was observable and addressable. You could buy the keyword, retarget the visitor, score the lead.
That sequence still happens. It just no longer starts where you can reach it.
According to Forrester's State of Business Buying, 2026, 94% of business buyers now use generative AI in their purchase process, up from 89% the year before — and twice as many buyers named generative AI or conversational search their most meaningful source of information than named any other source, outranking vendor websites, product experts, and sales reps. Gartner's survey of 645 B2B buyers, run in late 2025, puts active GenAI use at a more conservative 45%, mostly for gathering information on vendors and products.
The two numbers disagree, and the gap is worth sitting with rather than averaging away. Forrester surveyed a broad global buyer base and counted any AI use; Gartner counted GenAI specifically among seven information sources buyers reported using. The honest read is not "94% or 45%." It is that a large and fast-growing share of buyers now begin vendor research inside an answer engine, and that share is rising every survey cycle. The startups that win the next two years are the ones that treat that as already true.
When discovery moves into an AI answer, three things break for a vendor optimized around the old funnel:
Strip away the hype and three concrete shifts matter for a startup. Each one changes a different part of your GTM.
The mechanism is straightforward. A buyer types a category question into ChatGPT, Perplexity, Gemini, or a private enterprise model. The answer comes back as a short, confident list of vendors. That list is the new top of funnel, and it is built from what the model has been trained on and can retrieve — not from who bid most on the keyword.
What earns a citation is different from what earns a ranking. A Muck Rack analysis of more than a million AI prompts found that over 85% of non-paid AI citations originate from earned media — third-party publications, not vendor-owned pages. Peer-reviewed GEO research from Princeton and Georgia Tech (Aggarwal et al., KDD 2024) found that adding specific statistics to content raised its AI citation rate by 30–40%, and that citing credible sources raised it further. The model rewards content that looks like a source, not content that looks like a sales page.
A necessary caveat, because the GEO market is loud with inflated numbers: as a traffic channel, AI referral is still small. Benchmark data from 2026 puts generative-search traffic at roughly 1% of an average B2B site's sessions, against around a third from Google organic. You will also see vendors claim AI visitors convert anywhere from 4x to 23x better than organic. Those multiples come from small samples, are wildly inconsistent, and should be treated as directional at best. The defensible claim is narrower and still important: buyers arriving from an AI answer tend to arrive pre-informed and pre-shortlisted, which is a higher-intent state than a cold organic click — not a magic conversion multiplier.
So the discovery shift is not "AI traffic is huge now." It is "AI now decides your shortlist inclusion, even though it sends little traffic." That asymmetry is exactly why it is easy to ignore and expensive to ignore.
The more interesting behavioural change is what buyers do after the AI gives them an answer. They don't trust it.
In Forrester's survey, 36% of buyers said GenAI made them more confident in their decision, while 20% said it made them less confident because they hit unreliable or inaccurate information. Gartner found the skepticism is now even-handed: 51% of buyers said they were more likely to encounter misleading information from GenAI, and 49% said the same about a sales rep. Buyers now distrust the machine and the salesperson at roughly the same rate.
The result is a two-step trust process. The AI produces a fast draft of reality — a shortlist, a comparison, a business case. Then the buyer turns to a human to validate it. 69% of buyers told Gartner they turn to sales reps specifically to validate AI-generated insights. Gartner expects this to persist: it projects that by 2030, 75% of B2B buyers will still prefer sales experiences that prioritize human interaction over AI for the moments that carry risk.
For a startup, this is the optimistic half of the story. AI compresses early research, but it raises the value of the human who can confirm or correct what the machine said. The seller's job shifts from delivering information the buyer could get faster elsewhere to providing judgment, context, and reassurance the model can't. A founder-led sales motion is unusually well-suited to this — provided the founder shows up as a validator, not a brochure-reader.
A reasonable assumption would be that AI shrinks the buying group: if one person can research faster, fewer people are needed. The opposite happened.
Forrester reports that the typical B2B buying decision now involves 13 internal stakeholders and nine external influencers, with more on complex purchases — and that procurement is now a decision-maker in 53% of cycles, engaging from the start. Faster research lowered the cost of involving more people and consulting more outside voices, so buyers did exactly that to de-risk the decision. Notably, 94% of buyers in groups of six or more reported clear benefits from the larger group: broader perspective, shared validation, easier budget approval.
AI is also entering the committee as a participant, not just a tool. Forrester found that 61% of purchase influencers say their organization has used or will use a private GenAI engine to support purchasing, and predicts that in 2026 at least one in five B2B sellers will have to respond to AI-powered buyer agents — including agent-led quote negotiation. Gartner goes further, projecting that AI agents will be involved in $15 trillion of B2B purchases by 2028. That horizon is uncertain and the figure is a forecast, not a measurement — but the direction is consistent across both firms.
The practical takeaway is unglamorous: you are no longer persuading one researcher. You are equipping a committee, some of whose members are software, to build a case for you internally when you are not in the room.
Take those three shifts together and most of the standard GTM advice needs re-pointing.
Content stops being for humans only. Your content now has two audiences: the buyer and the model that summarizes you to the buyer. That means structured, specific, statistic-dense pages that an LLM can retrieve and cite cleanly — not thin keyword pages built for a 2018 search algorithm. The page that gets cited is the one that reads like a credible source, with concrete numbers, clear claims, and named evidence.
Earned media becomes a distribution channel, not a vanity metric. If 85% of AI citations come from third-party sources, then getting referenced on analyst pages, review platforms, respected industry publications, and credible roundups is now a core GTM motion, not PR garnish. Forrester expects 75% of enterprise B2B companies to increase influencer-relations budgets for exactly this reason. For a startup, the move is to be present and consistently described across the handful of sources models actually pull from in your category.
Category presence beats keyword presence. Models recommend entities they understand and see consistently described. A startup with one strong, coherent positioning statement repeated across its site, its founder's writing, third-party listings, and review profiles is more legible to an LLM than one with scattered messaging and a big keyword footprint. Consistency is a retrieval advantage now, not just a branding nicety.
Founder-led sales is a structural edge, not a phase to grow out of. Because buyers want a human to validate the AI's draft, the founder who can speak with authority at the validation moment closes the confidence gap the model leaves open. Don't rush to replace that with a generic SDR script the moment you can afford one. The thing the buyer can't get from ChatGPT is the thing you should staff most carefully.
Your sales material has to survive being read by a machine. When a buyer feeds your one-pager or pricing page into their own AI to compare you against competitors, ambiguous claims get flattened and over-claims get exposed. Specific, verifiable, well-structured collateral now does double duty: it persuades the human and it survives the model.
The mistake nearly every team makes first is assuming the AI shift is a tooling update — rename the SEO checklist, add "ChatGPT" to a slide, keep doing the same content at higher volume.
It isn't a renamed checklist, for three reasons.
SEO optimizes for a click; GEO optimizes for a citation inside an answer the buyer may never click through. The zero-click reality means your influence often happens entirely inside the model's response, with no session to measure. SEO rewards your own pages; AI citation disproportionately rewards third-party sources you don't control. And SEO is a volume game; AI recommendation is an authority-and-consistency game, where being clearly and repeatedly described as a credible entity matters more than publishing ten more posts.
A team that responds to the AI shift by shipping more blog posts is solving the wrong problem faster. The work is to become the kind of source models cite — which is closer to building genuine authority than to gaming a ranking.
Run this as a sequence. It works for any B2B category.
Before changing anything, ask the answer engines your category's buying questions — "best [category] tools for [ICP]," "alternatives to [competitor]," "is [your product] good for [use case]." Note whether you appear, how you're described, and what sources the model cites. That output is your current top-of-funnel. Most founders have never looked at it.
Look at which third-party publications, review sites, analyst pages, and roundups the AI cites when answering your category questions. Those are the sources that decide your shortlist inclusion. Getting accurately represented on that specific short list of properties is higher-leverage than almost any owned-content work.
Rebuild your most important pages to be retrievable and quotable: clear claims, specific numbers, named evidence, clean structure, and an unambiguous statement of who you're for and what you do. The goal is that a model can lift a correct, flattering sentence about you without having to interpret. Vague positioning is now a retrieval liability.
Assume a committee of 13-plus, some of them using private AI, will build the internal case without you. Give them the artifacts that survive that process: a comparison that holds up when fed to an LLM, ROI evidence that isn't hand-wavy, and a crisp answer to the procurement and risk questions that now arrive early. Make it easy for an internal champion — or their AI assistant — to argue for you accurately.
Put your most credible human — often the founder, early on — at the point where buyers seek validation, and resist automating it away prematurely. Then revisit the whole picture every quarter, because this environment moves fast: models change what they cite, competitors move into the same sources, and buyers adopt new tools. A GEO and trust strategy that worked six months ago can quietly go stale.
AI did not abolish the B2B fundamentals. Buyers still want to reduce risk, build internal consensus, and trust a human before they commit. What changed is where the journey starts and who shapes the first impression.
Discovery moved into answer engines, so shortlist inclusion is now decided by what third parties say about you, before you ever see the buyer. Trust split in two, so the AI drafts reality and a human validates it — which raises the value of a credible founder, not lowers it. And the buying group grew and absorbed AI agents, so you are equipping a partly-automated committee to argue your case in your absence.
"We rank #1 for our keyword" is a 2019 sentence. "When a buyer asks ChatGPT for the best tools in our category, we're on the list, accurately described, for the right reasons" is the 2026 version. The startups that internalize the second sentence — and act on it before their category incumbents do — get to be the default answer while everyone else is still optimizing for clicks.
If you want help turning this into a concrete GEO and buyer-enablement plan for your category, that's the kind of work Big Moves Marketing does with growth-stage B2B teams.
%20-%20white.svg)