B2B Buyer Journey: What the Published Stage Models Disagree On

First result declines to draw a journey

Search the term and Gartner comes back first. Open it looking for stages and you get a different kind of answer. Its own FAQ says buying teams work on multiple tasks concurrently or individually, without a consistent order or journey.

John Dawes at the Ehrenberg-Bass Institute puts a size on the audience this applies to. About 20 percent of business buyers are in the market over a year, and something like 5 percent in a quarter.

He calls the figure a heuristic, not a measurement.

The 20 and 5 percent are Dawes’ figures, and Gartner publishes something else. Asked what the stages are, Gartner answers with categories instead of steps, and it is not a caveat buried in a footnote.

Today's B2B buying journey follows a nonlinear purchasing path that more closely resembles a set of distinct buying jobs or sets of tasks that buyers must complete to finalize an actual purchase. Most buyers revisit at least one buying job during their purchase journey.

Gartner, The B2B Buying Journey, FAQ, read 5 September 2026

The four categories it names are problem identification, solution exploration, requirements building and supplier selection. On best practice the same page is blunter still: catalogue the jobs and the tasks underneath them, instead of drawing a path.

Stage models get drawn anyway, and they do not line up. Here is what each one gives.

Google is not settled on a number either, and its own suggested questions are the place to see it.

How many buying jobs a journey contains

A number does get attributed to Gartner, and it is not Gartner's. One published framework is titled Gartner's six stage framework and lists problem identification, solution exploration, requirements building, supplier selection, validation and consensus creation. Gartner's own page names four, and the other two are absent from it.

The page states the attribution plainly: Gartner's framework breaks down the B2B buying process into six clear stages. It was updated on 22 June 2026.

The attribution, in full, on the page that makes it

1. 1The heading, the sentence and the table all name Gartner. This is a clear claim, not an implication.
2. 2Rows five and six, validation and consensus creation, are the two that do not appear on the Gartner page.
3. 3The table itself is well made. The problem is not its quality, it is that a reader cannot confirm it at the source it names.

What this does not prove: that the six job model was never Gartner's. It may well have been published by Gartner elsewhere. The material that would settle it sits behind an email form, so the question stays open.

What it does establish is checkable and enough. A reader who takes six stages from a third party, then goes to Gartner to confirm them, will not find them. They will find four categories and a recommendation to stop mapping.

This is worth being precise about because it is a common failure mode in this category. A framework gets summarised, the summary gets cited, and the citation outlives the page it came from.

The same pattern shows up in the statistics these publishers carry. One page cites Gartner for a figure that buyers spend 17 percent of their purchase journey meeting with potential suppliers, and a second cites the same number. Neither links to a Gartner page carrying it, and Gartner's public page on the buying journey does not state it.

None of that makes the number wrong. It makes the claim unverifiable from where a reader stands, and that is a different complaint. Treat any figure repeated across four vendor blogs without a primary link as a claim, not as a measurement.

One statistic in this area carries its own primary source, and it is the most useful of them. One page cites McKinsey's B2B Pulse for 2024 on buyers using an average of 10 interaction channels across a purchase. Ten channels is the practical reason a single ordered path is hard to observe. How many of those ten touch a system you own is not something McKinsey counted, and it is the question your own record answers.

Tracking jobs instead of stages

If your buyers work on tasks concurrently, a single position field cannot describe them. Your record holds one stage at a time, so you have to choose for a buyer doing three things at once. A workable answer starts by giving up that single field.

1. List the jobs, not the steps

2. Attach one observable signal to each job

3. Record jobs as flags, not as a position

4. Keep one stage field for the handoff only

5. Review which jobs stall

The stage field that survives step four is the marketing to sales boundary, and the stage model behind it is set out in the B2B marketing funnel. Two models, two jobs: one tracks ownership, this one tracks buyer progress.

What buyer intent data costs

The commercial answer to buyer journey work is intent data: software that tells you which accounts are researching your category. Pricing sits behind a sales call across the category, so no published range exists to quote.

What can be priced is the record keeping. HubSpot publishes Marketing Hub Professional from $800 a month with 2,000 marketing contacts and a one time $3,000 onboarding fee, and Enterprise from $3,600 a month. Salesforce publishes Sales Cloud from $25 a user a month for Starter Suite up to $550 for Max, billed annually. Both vendors revise these figures.

Those licences buy you the fields, and custom properties come with both products at those plan prices. The flags are not a separate purchase.

The expensive part of this work is not software. Deciding which signal stands for which job is judgement, and judgement is not a line item.

One gap stays open: no vendor in the intent data category publishes a rate card, and no independent source publishes a median. A reader who wants a number will have to get quotes.

Say you would rather price the search half of your own journey: that is our B2B SEO service.

How does a buyer map go stale?

Journey maps are cheap to draw and expensive to keep. Each of these failures is visible in public, and that openness is useful in a category where everyone shows their working.

All three failures share one tell: your map gets treated as a finding when it is a drawing.

For example, if your map says buyers compare vendors in week three, ask two reps when the last three deals really did it.

Drawing a map with no measurement behind it. HubSpot's journey mapping guide gives eight ordered stages and, on the version read on 5 September 2026, carries no cited statistic at all. Another gives three stages, says in the same piece that the journey is often non-linear, and also cites nothing. A map without evidence is a hypothesis with confident typography.

Eight stages, numbered, on the same result page

1. 1The numbering is the claim. A numbered list asserts an order, and the introduction says most journeys follow it.
2. 2Onboarding, usage, expansion and advocacy all sit after the purchase, so this list spans the buyer journey and the customer relationship in one sequence.
3. 3The list carries no citation for any stage.

Building the map around what your team can see. 6sense surveyed 2,509 B2B buyers in 2024, and both of its headline figures describe the stretch before a seller is involved. Your CRM reaches that stretch only where your own systems were touched.

Treating a preference for self service as a preference for no help. Gartner publishes that 75 percent of B2B buyers prefer a rep-free sales experience. On the same page it states that self-service digital purchases are far more likely to result in purchase regret. Both halves belong to the same finding, and only the first half gets quoted.

Stalls and acceptance outlive stage labels

A conversion rate per buyer journey stage is not published anywhere. No standard set of stages exists, so a rate for stage three would not mean the same thing twice. Two numbers survive the disagreement about labels.

The first is where deals stop. Forrester reported in its 2024 State of Business Buying research that 86 percent of B2B purchases stall during the buying process, a figure quoted on a vendor page in June 2026. Forrester's report is not public, so the figure reaches this page second hand.

The structural point about any funnel drawing is the one Gartner already made: buying teams work on tasks concurrently, without a consistent order. People enter at any stage, and a journey map inherits that and then draws an order on top of it.

The second is what sales accepts. Your sales acceptance rate is the published barometer of a scoring program's health, and the same test works one level up. It measures whether your reading of the buyer matches theirs, and it does not depend on how many stages anyone drew.

A third number is easier to gather than you might expect: how many people from the account touched you before a deal opened. 6sense reports an average buying group of 11 people in its 2024 work, quoted second hand here.

Forrester publishes its own count of the same thing, first hand. Thirteen people inside the organisation are involved in the buying decision, and 89 percent of purchases involve two or more departments.

That release does not publish its sample, so read those as the shape of your buyer’s room and not its size. For example, your own contact count per closed deal is a rough version of the same thing, and it is already in your CRM.

One thing stays unpublished and is not invented here: a way to observe from outside when a buying job is finished. Gartner recommends cataloguing jobs and does not say how to detect completion. That is the open problem in the method, and anyone adopting it inherits it.