The B2B buying journey is splitting in two

Most B2B marketing still runs on the funnel. A buyer becomes aware of a problem, researches solutions, compares vendors, talks to a salesperson, and one of them wins. Awareness at the wide end, a signature at the narrow one. Campaigns feed the top, lead scores track the descent, a demo request counts as progress.

It is a tidy picture, and very little of it survives contact with how a company actually buys software now.

Here is what breaks it. A business purchase is not one person moving through stages. It is a group that forms around a single decision and dissolves once the decision is made. Nobody is formally appointed to it. It is simply everyone whose agreement, budget, signature or silence the purchase will eventually need: the person who wants the thing, the people who have to run it, the people who have to pay for it, and the people whose job is to ask what happens if it goes wrong.

Two things have happened to that group recently. Finding answers has become almost instant. And the group has got bigger. Those two changes pull in opposite directions, and most B2B marketing is still built for neither.

The half of the journey you never see

By the time a vendor learns that an opportunity exists, a great deal has already happened without them.

The starting point has moved. Forrester’s 2026 State Of Business Buying report describes genAI searches as where B2B buyers now begin. McKinsey’s Pulse survey shows the same shift from another angle: generative AI has climbed into the top five channels buyers use to discover and evaluate suppliers, alongside supplier websites, in-person conversations, web search and video calls. Asking a model to explain a category, name the main players and summarise the trade-offs is now an ordinary first move.

But an answer is not the same as a decision. Forrester is blunt about why: answer engines “often deliver incomplete or unreliable information, creating mistrust”, so buyers take what they have been given and check it against people and sources they already rely on. The need to validate has not gone away. It now begins the moment the AI answers.

That mistrust is not theoretical. Forrester’s 2026 predictions guide found that 19% of buyers using AI tools during a purchase came away less confident in their decision rather than more, because of inaccurate or unreliable output. Roughly one buyer in five is being actively misled by the tool they started with.

And yet the tool keeps winning their attention, even where you would least expect it. At the final commit stage, the last stretch before contracts are signed and money moves, buyers rated an exchange with a generative AI tool as a more meaningful interaction than a conversation with a product expert, by 30% to 17%. That is the point of maximum risk in a purchase, the moment when a person would traditionally want a human being on the other end of the line, and the machine is out-scoring the expert by nearly two to one.

Forrester expects that to reverse rather than continue. The reasoning is simple enough: as AI produces more information, and more misinformation, the genuinely scarce thing stops being an answer and becomes someone who can tell you which parts of the answer to believe.

Meanwhile the buyer is doing most of the work alone. 6sense’s 2025 Buyer Experience Report, based on nearly 4,000 recent buyers, found that buying groups get through around 60% of their journey before a vendor is involved at all. In that stretch they work out what the problem is, research quietly, draw up a shortlist, and often settle on a favourite.

That private window is narrowing, though, and the reason is worth sitting with. First contact moved from 69% of the way through the journey in 2024 to 61% in 2025, roughly six or seven weeks earlier. Buyers are not calling sooner because they need less help. 58% said what pulled them in early was needing to evaluate how vendors were implementing AI inside their products. AI in the product raised a question buyers could not settle on their own, so they went looking for a human sooner than they used to.

By the time they do make contact, the competitive picture is often close to settled. Forrester’s numbers on this are the most uncomfortable in the whole subject.

Read that chain again, because it reframes the job. A formal evaluation is often not a contest. It is a confirmation of a decision that was substantially made while nobody from the vendor was in the room.

So the first question for marketing is no longer how to generate a lead. It is how to be one of the answers a buyer runs into before there is a lead to generate. That answer might come from an AI tool, an analyst, a former colleague saying “we used them at my last company”, or something someone read six months ago and half remembers. By the time a demo request lands in a CRM, much of the real competition has already happened.

Buying got more human, not less

You would expect faster research and better self-service to gradually push people out of B2B purchasing. The evidence runs the other way. Forrester’s summary of the pattern is that leaders increasingly lean on two networks, colleagues inside the company and influencers outside it, to justify and de-risk the decision.

That is why the group has grown.

Thirteen colleagues and nine outsiders, on a typical purchase, and more on anything complex or strategic. Putting AI in the product multiplies it further: Forrester found that purchases including generative AI features carry buying groups twice the size of purchases without them. The more novel the technology, the more people a company wants looking at it.

A group that size is slow and awkward, and buyers know it. They want it anyway. Among buyers whose groups ran to six people or more, 94% reported clear benefits: broader perspectives, shared effort in validating solutions, a better ability to secure budget and a greater likelihood of approval.

That last pair is the tell. Securing budget and getting approval are not product questions. They are political ones. The extra people are not friction the buyer is trying to remove; they are how the risk gets spread thin enough that somebody is willing to sign.

Which sets up the real problem. AI has made information cheap and abundant. It has not made any of it easier to trust, and trust is what a group of twenty-two people actually needs in order to agree.

Foundry’s 2026 Customer Engagement Study, fielded among 676 IT decision-makers across North America, EMEA and APAC, measures exactly that gap. 81% say they are challenged to find high-quality content from vendors when evaluating major purchases, down from 87% in 2023 and 90% in 2021, but stubbornly high given how much more content exists now.

The reasons matter more than the number. The top complaints are marketing hype (31%), a lack of unbiased information (25%) and content that is too generic (24%). Volume complaints come below those: content overload at 22%, too many search results at 21%. Buyers are not drowning. They are looking at a lot of material and finding very little of it credible enough to put in front of a colleague.

There isn’t one journey, there are several running at once

This is where a traditional journey map stops helping.

Picture a company evaluating an enterprise AI platform. On a funnel diagram this is one dot, at one stage. In the building, it is six conversations happening simultaneously.

• Operational champion”Which platform has the best capabilities?”
• Finance”Why do we need another platform at all?”
• IT”Where does the data go?”
• Security”What happens if the model leaks something sensitive?”
• Procurement”Can we consolidate this with a supplier we already have?”
• Executive sponsor”Does the value clear the bar for the disruption it will cause?”

Six people, one purchase, and not one of them is at the same stage as the others. The champion is comparing feature lists while finance is still questioning whether the project should exist. No single piece of content moves all six forward, because they are not asking the same question.

They also do not proceed in order. Foundry’s study tracks seven stages a purchase passes through: determining business need, determining technical requirements, evaluating products and services, recommending and selecting vendors, selling internally, approval and authorisation, and post-sales engagement. Its finding is that buyers do not walk through them. “Enterprise IT purchase decisions have never been linear,” the report notes. “Buyers move back and forth between stages as requirements evolve, stakeholders change, and new information emerges.”

The material that carries a purchase changes as it loops. What wins an evaluation is a demo and a product test. What carries the same purchase through selling internally is credible external evidence, news coverage and analyst reports, because the champion has stopped choosing and started convincing.

And any of it can be undone. A security objection reopens supplier selection. A new CFO reopens the business case. A competitor ships a feature that rewrites the requirements. A weak proof of concept sends everyone back to arguing about whether the problem is worth solving at all. The CRM says Stage 4: Evaluation. Inside the buyer’s office, four different arguments are live.

The journey nobody in sales ever sees

Somewhere in that group is the person who wants this to happen, usually whoever found the vendor in the first place. Sales teams call them the champion. Their hardest work is not evaluating the product. It is carrying the decision through their own organisation, and almost none of that is visible to the vendor.

LinkedIn’s B2B Institute, working with Bain & Company, has put structure on this under the name “Buyability”. It splits the group in two. Target buyers evaluate the product: they read the material, take the demo, care about capabilities. Hidden buyers sit in finance, legal and procurement. They rarely download anything, rarely attend anything, and leave almost no trace in a marketing system.

They are also, collectively, about half the decision. The research puts roughly 50% of total decision-making influence with those hidden buyers. Forrester’s numbers point the same way: procurement now serves as a decision-maker in 53% of business buying cycles.

The two halves want different things from the same purchase. A target buyer is asking what the team could accomplish. A hidden buyer is managing downside risk, and the question in their head is whether this decision can be defended if it goes badly.

Get that wrong and the deal does not usually get lost to a competitor. It just stops. LinkedIn and Bain found 40% of deals stall because the buying group cannot agree, not because someone else won. Nobody says no. The thing simply never reaches a decision.

Familiarity is what moves it. A vendor is roughly 20 times more likely to be chosen when the whole buying group already knows and trusts the brand at the outset, compared with when only the technical champion does. In the same research, 81% of purchases went to vendors almost everyone in the group already recognised. Only 4% went to a vendor known solely by the person recommending it.

Foundry finds the same blind spot from another direction: 64% of IT decision-makers say vendors could do a better job of educating the non-technical functions inside their own organisation, a gap the report ties to how much influence business users now hold over technology decisions.

All of which gives brand marketing a job that has nothing to do with fame. Being known by the champion is worth surprisingly little. Being known by the finance director who has never heard of you, and now has to approve you, is worth almost everything. Brand is not there to make you famous. It is there to make agreement easier.

A good product is not enough

Once a few vendors clear the technical bar, a different contest starts, and it is not the one most vendors think they are in.

The champion is no longer choosing a product. They are preparing to defend a choice, in a room, in front of people who were not part of the evaluation, some of whom are actively looking for a reason to say no. What they need is not more features. It is ammunition.

So a B2B vendor is really selling two things: the solution, and the case for choosing the solution. The second is built from customer evidence, ROI, implementation proof, references, analyst opinion and third-party validation that did not come from the vendor. It is the part the champion can quote when they are in the room alone.

The evidence for how much this matters is stark. LinkedIn and Bain found buyers were three times more likely to choose a vendor heavily recommended by peers or existing customers than one simply promising a better product or a lower price. A better product loses to a safer one. Forrester’s 2026 predictions guide expects that balance to tilt further still, predicting that proof of success will overtake brand reputation as the leading nontechnical driver in purchase decisions.

Buyers increasingly want to generate that proof themselves rather than take it on trust. Forrester found more than 60% of buyers now run some form of trial before buying, rising to 78% on purchases of $10 million or more. On the largest deals, where you would expect the most deference to a vendor’s claims, buyers are the most insistent on checking.

The contradiction that isn’t one

Gartner’s most recent sales survey, run across 646 B2B buyers in August and September 2025, found that 67% prefer a rep-free buying experience: getting all the way to a decision without a sales conversation at all. Read that alone and the conclusion writes itself. Sales is being disintermediated.

Then read it next to Foundry, which found 71% of IT decision-makers say the vendor that responds to their questions quickly and thoughtfully usually gets the business. The same buyers who would rather not talk to anyone still hand the deal to whoever answers well when they finally do. And they are less patient about it than they were: the expected wait for a first follow-up has fallen to about 13 hours, from 17.5 hours in 2023.

Both things are true, because they are about different moments. Buyers do not want a salesperson standing between them and the information. They can pull up features, watch a demo, ask an AI to explain the category and compare vendors before lunch, and they resent anyone slowing that down.

What they cannot do alone is the hard part: work out how any of it applies to their own company, decide what belongs in the business case, learn how a comparable company actually rolled it out, weigh trade-offs nobody wrote down anywhere, and anticipate what finance is going to challenge.

That is the shift. The rep stops being a gatekeeper to information, which buyers now resent, and becomes the person who helps a group reach a decision it can defend later, which buyers will still pay for.

Marketing has the same blind spot

Most B2B marketing is still built to detect individuals. Someone clicked. Someone downloaded a report. Someone attended a webinar. Someone crossed a lead-score threshold.

But companies do not buy because one person accumulated enough points in a scoring model. They buy when enough people, arriving separately and for their own reasons, become comfortable with the same decision. The job is not moving one person from awareness to consideration. It is moving an organisation from uncertainty to confidence, which is a different task with different materials.

It is also harder to do consistently, because the organisation is meeting you in a dozen places at once. McKinsey’s 2026 Global B2B Pulse Survey, covering nearly 4,000 decision-makers across 13 countries, found buyers now use an average of ten channels in a single purchase and expect to move between them without friction.

What happens when that breaks is instructive. Asked what would make them stop working with a supplier, buyers put inconsistent information from different teams at the top, cited by 52%, followed by the plain difficulty of reaching anyone with the knowledge or authority to help. Not price. Not features. Two teams giving different answers, and nobody available who can settle it.

Trying to control a journey like that is hopeless. Making it easy to get a straight answer at any point in it is not.

Map the decisions, not the funnel

A useful journey map should not stop at Awareness, Consideration, Decision. Take a real purchase and ask a better question: what has to become true before this company buys?

Someone has to believe the problem is worth solving. Someone else has to believe your approach is the right one. Someone has to find the company credible before the economics get discussed at all, and someone has to prove those economics hold. The technology has to satisfy whoever is accountable for running it. The risk has to feel survivable to whoever answers for it if it fails. And eventually enough of those beliefs have to exist in the same building at the same time for money to move.

Every one of those is a different person, a different worry and a different kind of evidence. None of them are stages in a funnel.

Seen this way, AI does not remove marketing or sales from the purchase. It moves where their value sits. AI helps a buyer discover what is possible. A website lets them investigate without asking permission. A trial proves the product works. Customers make the claims believable. A brand that the whole group recognises lowers the risk of choosing you. A salesperson helps that group work through what is genuinely ambiguous. And the champion carries the decision the last few metres, through their own building, to a yes.

The useful question is no longer how to move a buyer through a funnel. It is what this buyer needs in order to move the decision forward, and who else in that building needs it too.

Questions worth answering

How many people are involved in a typical B2B purchase?

Forrester's 2026 State Of Business Buying report found a typical business purchase now involves 13 internal stakeholders and 9 external influencers, rising further for complex or strategic purchases. Purchases that include generative AI features carry buying groups of 14 members, double the seven on purchases without them.

How much does AI search change the B2B buying journey?

It changes where the journey starts, not how it ends. Forrester describes genAI search as the starting point for B2B buyers, and 6sense found buying groups complete around 60% of their journey before contacting a vendor. Buyers still validate what AI tells them against people and sources they trust before deciding, because answer engines often return incomplete or unreliable information.

What are 'hidden buyers' in B2B purchasing?

LinkedIn's B2B Institute and Bain & Company use the term for buying-group members in finance, legal and procurement who evaluate risk rather than product fit and leave little trace in marketing data. Their research puts roughly 50% of total decision-making influence with these hidden buyers, and Forrester separately found procurement acts as a decision-maker in 53% of business buying cycles.

Why do B2B buyers prefer a rep-free sales experience but still need sales reps?

Gartner found 67% of B2B buyers would rather buy without a sales rep, yet Foundry's 2026 Customer Engagement Study found 71% of ITDMs say the vendor that responds to their questions quickly and thoughtfully usually gets the business. Buyers do not want a rep gatekeeping information; they want fast, relevant help applying it to their own company and defending the decision internally.