How a B2B Buying Committee Actually Buys

Six jobs, worked through out of order

A B2B buying group has six jobs to complete before it can sign: problem identification, solution exploration, requirements building, supplier selection, validation and consensus creation. The framework is Gartner's, and the most important thing Gartner says is that the jobs are not sequential: B2B buying does not play out in any predictable, linear order, and customers instead engage in what Gartner calls looping, revisiting jobs across a purchase.

The questions being asked are worth learning in Gartner's phrasing. Problem identification is we need to do something. Solution exploration is what is out there to solve our problem. Requirements building is what exactly do we need the purchase to do. Supplier selection is does this do what we want it to do. Validation is we think we know the right answer, but we need to be sure. Consensus creation is we need to get everyone on board.

Read those six as things a website must answer at any moment, in any order, for a reader arriving on a different job from last week's. That is not the brief a funnel implies.

Who is actually doing the buying

Gartner's press release of 7 May 2025, reporting a survey of 632 B2B buyers fielded in August and September 2024, describes modern buying teams as ranging from five to 16 people across as many as four functions. Three findings from the same research change how you write for them:

• Around 74% of buying groups exhibit unhealthy conflict — members holding conflicting objectives, disagreeing on the best course of action, or being overruled by external decision-makers.
• Groups that reach consensus are 2.5 times more likely to report the deal was high quality.
• Group-level relevance improved consensus by 20%, while individual-level relevance reduced it by 59%, and buyers experiencing buying-group relevance were three times more likely to report a high-quality deal.

That last finding cuts against a decade of personalisation marketing. Material addressed to the group's shared problem helps the group agree; material tailored to each individual's concerns appears to make agreement harder, presumably because it hands each function a different reason to care and no common ground.

What each job needs from the site

Two of those rows are missing from most corporate websites. Requirements building is the more damaging gap: if nothing on your site helps a buyer specify the purchase, the specification gets written by whichever competitor did publish that material, and every requirement then favours them.

What a champion needs in order to sell you internally

Consensus creation is the job you can least help with directly: it happens in meetings you do not attend. Somebody inside the organisation argues your case, and what you gave them decides whether they win. Assume that person is competent, busy, and less fluent in your product than you are.

• A short business case they can paste into their own document — the problem, the options, the recommendation, the cost shape, the risk of inaction.
• A one-page summary for an executive who will never speak to you and will read nothing longer.
• Answers to objections from functions not in the room: security, legal, finance, and whoever chose the incumbent.
• Procurement paperwork before it is requested — certifications, insurance, data-processing terms, subprocessor lists.
• A reference in their sector at roughly their size.

Make it forwardable: a stable URL beats an attachment, and a document behind a form cannot be forwarded to a chief financial officer at all, so your most shareable asset should never be your most gated one. And write for the group, not the individual.

Which pages do the work, in practice

Mapping the six jobs onto a real website produces a page inventory different from the usual one, in rough order of how often each is missing:

1. Problem-level pages describing a symptom rather than a product category — what somebody finds when they know something is wrong but not what to buy. Usually absent because teams name things by product.
2. Requirements and specification resources: what to measure, what to ask, which standards apply, what a complete scope of work contains.
3. Comparison and alternatives pages, including approaches you do not sell and cases where you are the wrong choice. Buyers who find a limit later treat it as concealment.
4. Proof pages: case studies with figures the customer has agreed you may publish, and a security section that answers the questionnaire before it arrives.
5. Commercial-model pages explaining how you charge — the unit, what drives it up or down, what is included. Silence there is read as evasion.

Internal site search has to work, because people arrive mid-loop hunting one thing they half-remember, and URLs have to be stable, because they get pasted into emails and reopened months later.

Why content mapped to funnel stages produces the wrong assets

The standard exercise is a grid: stages across the top, personas down the side, one asset per cell. It produces coverage, and coverage was never the objective. Three failures follow. It assumes each asset is consumed once, at a fixed point, by one person, so nothing is designed to be re-found or read out of sequence — yet in a looping process the valuable asset is the one a buyer returns to on a third visit. It fragments what should be consolidated, producing thin pieces where the group needs complete resources that survive being forwarded across four functions. And it gates by stage rather than by purpose, which is how the most forwardable document in the library ends up behind a form.

The journey map on the wall has the same problem: it is a device for agreeing vocabulary, and it misleads the moment it is used to schedule content or attribute revenue to a stage.

One more piece of conventional furniture should be retired. A family of widely repeated percentages claiming to quantify how far buyers get before engaging a supplier circulates constantly in this field. The versions conflict, the methodology is not publicly retrievable, and the idea is incompatible with a process that has no fixed order. Do not use it, least of all to justify a content budget.

Measuring a process that has no order

Reporting built on a linear model describes something that is not happening. What can be measured honestly consists mostly of signals of re-finding and forwarding rather than of progression.

Worth reporting: repeat visits to the same pages from the same organisation over weeks; direct traffic to deep URLs, the signature of a link pasted into an internal email; internal site-search queries, which show what buyers expected to find and did not; and the documents that won deals mention when you ask.

Worth discounting: stage-conversion percentages from a funnel that does not exist; single-touch attribution across a decision spanning several quarters; and account-engagement counts from visitor-identification tools taken at face value. Reverse-IP identification tells you which organisation is assigned the address a request came from — often a residential broadband provider, a mobile carrier, a VPN or a serviced-office landlord rather than the visitor's employer — and no vendor publishes an audited accuracy rate. Measure over the right window, too: document and webinar engagement accumulates for weeks.

Where to start, given all of that

Take your last five won deals and last five losses and ask each three questions: what changed that made you look, who else was involved and when, and what did you read or send to a colleague. Ten conversations will tell you more than any published journey model, because they reveal the order your buyers used and the documents that did the internal work.

Then audit the site against the six jobs and find the job with no page behind it. In most corporate websites the two gaps are the same: nothing helps a buyer with requirements building, so the specification gets written to a competitor's strengths, and nothing exists that a champion can present without you in the room. Building those two — a specification resource and one business-case document designed to be presented by somebody else — beats another quarter of blog posts.

Watch for the temptation to rebuild the funnel; it returns because it makes budgets legible. Keep the six jobs on the wall instead, and when somebody proposes an asset, ask which job it serves and who will forward it to whom.

Frequently Asked Questions