There are 22 people involved in the average B2B purchase decision. Thirteen internal stakeholders, nine external. Most companies have two contacts in their CRM and a pipeline report that says the deal is "progressing."
Forrester published the 22-stakeholder number in 2025 and most teams skipped past it because the implication is uncomfortable. Eighty-one percent of buyers already have a preferred vendor before the first outreach. Eighty-five percent have written their requirements before any seller is involved. The "sales process," for the most part, is theatre staged for a decision that already happened.
That is the disorienting part. The hopeful part is the other side of the same fact. The companies that win these deals are not the ones with the best closers. They are the ones that were useful six months earlier, when one person on a 22-person committee was trying to figure out who actually understands their problem. The decision was made in a room you were not invited to. The job is to be remembered when that room meets.
This piece covers three things in order: what the room actually looks like, what the invisible buyer is reading, and how to be the company that gets remembered.
What the 22-stakeholder reality actually looks like
A B2B purchase committee in 2026 is not a single room with a single decision-maker. It is a distributed group with overlapping mandates, asynchronous research, and at least three competing framings of the problem. Each person sees the purchase through the lens of the function they own.
The internal thirteen typically include an executive sponsor, a champion who runs point, a budget owner in finance, a technical evaluator in IT or engineering, a security or compliance reviewer, an operations lead who will use the thing daily, a procurement function that runs the negotiation, a legal reviewer for contract terms, and a handful of cross-functional consultees pulled in to validate fit. Some are deeply engaged. Some receive a Slack message asking for a thumbs-up and never read the proposal. Both count as stakeholders.
The external nine are the harder group to model because most teams forget they exist. They include consultants and analysts the buyer trusts, peers in adjacent companies who get an unofficial reference call, advisors and board members who weigh in informally, and — increasingly — the AI tools the champion uses to generate a vendor shortlist before the company even formally opens an evaluation. These are the people you are competing against without knowing it.
Each vendor in the consideration set gets roughly five percent of the buyer's total evaluation time. That is your window. Five percent. You are not running a sales process inside that window. You are running a memory-and-clarity check. The committee is asking, in effect, "of the options on this list, which one understands what we are actually trying to do." Whoever answers that question most clearly with the smallest amount of effort from the committee tends to win.
What the invisible buyer is reading
The invisible buyer does not read your homepage first. They start in a search bar or an AI chat, with a phrase that sounds like a problem rather than a product. "How do other manufacturers handle [specific operational pain]." "Best approach to [niche regulatory constraint]." "What questions should we be asking vendors about [a thing they only half understand]."
In 2026, that initial query gets answered in two places before they ever click through to a website. The first is an AI Overview or a generative chat response that synthesises an answer from a small set of cited pages. The second is a vendor-comparison summary the champion will literally paste into a Slack thread for committee review. Both surfaces are built from retrieved content. Both reward the same thing: pages that contain a passage strong enough to be lifted directly into an answer with attribution.
The invisible buyer is not reading a long article cover to cover. They are reading the paragraph the model decided was the best evidence for their question. If your content does not contain that paragraph in a form a model can cite, you are not in the conversation. The pages they do read in full tend to be the ones cited in those generated summaries — which means the structure of how generative search retrieves and cites content has direct line-of-sight to whether your sales team gets a meeting.
We covered the mechanics of this in The Citation Economy and Your Traffic Is Down and Your Rankings Haven't Moved. The point here is narrower. The invisible buyer is not a content-marketing audience. They are a research function inside a 22-person decision, and they are using AI tools to triage the choice on your behalf. The format that wins is the format the model will quote.
There is a second pattern under that one. The invisible buyer is also reading what their peers wrote about you. Earned mentions in industry publications, podcast appearances, conference talks, slide decks shared on LinkedIn, code repositories, customer case studies — everything outside your owned property that an AI engine might use as a source. The discoverable surface is wider than your site. Most teams optimise only their site and ignore the other ninety percent of the surface their buyer is actually reading.
How to be remembered when the room decides without you
The first move is to stop optimising for the part of the journey you can see and start optimising for the part you cannot. Most marketing dashboards measure visits to the website and conversions on the form. Those are the visible part of the decision — typically the last five percent. The other ninety-five percent of the decision happened before that visit, in places you do not control and in formats your analytics stack does not capture.
That is uncomfortable. It is also clarifying. If most of the decision is invisible, the right strategy is not to chase higher attribution on the visible part. The right strategy is to be useful and specific in the formats the invisible buyer is actually consuming.
Three concrete moves we run with clients.
Be the canonical source on the entities your buyer cares about. For every problem your ICP genuinely struggles with, identify the entity at the centre of it and build the page that is unambiguously the resource a serious researcher would bookmark. Not a top-of-funnel "complete guide to X" with no point of view. The reference. We covered the workflow for this — entity inventory, consolidation, lift-test rewrites — in The Citation Economy. The output is pages that get cited, not just ranked.
Frame the problem before the buyer frames it themselves. The seller-buyer perception gap on what the actual problem is averages around fifty-four percent — more than half of the time, both sides are describing different situations and neither knows it. The most useful content you can publish is the content that gives the champion a better way to frame the problem internally. If your framing becomes how the committee talks about the problem, you have already won the meeting you are not in. We talk about this dynamic in The Confidence Trap.
Make a champion's job easier inside the company. The visible buyer is the champion. The invisible buyers are the other 21 people they have to align. Anything that reduces the cognitive load of selling internally — a clean one-pager that maps your offering to the questions IT will ask, a comparison framework procurement can adapt, a proof point that lands with finance — pulls weight none of your sales reps will ever see. The companies winning in this environment are the ones who treat the champion as the user, not the lead.
The going-to-market thesis underneath all three is the same. More information has not made B2B buying rational. It has multiplied the friction. The opportunity is not in feeding the research machine. It is in being the company that helps the buyer think clearly when everything else is noise — and in being clear enough, in the formats the room is actually reading, that they remember you when the decision is made without you.
Further reading
- The Citation Economy — the operating manual for being the source AI engines build answers from.
- The Confidence Trap — why AI-armed buyers arrive more confident, more dissatisfied, and harder to align.
- The Handoff Tax — why fragmented agency stacks fail in environments where the invisible buyer expects coherence.
