A CMO I worked with last year opened a Miro board on our first call. It was genuinely beautiful. Nine swim lanes, color-coded emotions, little avatar faces going from confused to delighted, quotes pulled from twelve customer interviews. An agency had run a two-day workshop to produce it. It cost them somewhere north of $20,000.
I asked one question. "What changed after this?"
Long pause. Then: "We moved a section on the pricing page."
That is the honest outcome of most B2B journey mapping. A team spends real money and two days of everyone's calendar producing an artifact, and six months later the artifact is a tab nobody opens and the pipeline looks exactly the same.
The mapping was not the problem. Their map was actually good research. The problem is that a journey map drawn in Miro has no way to touch anything. It cannot route a lead, change a deal stage, trigger an alert, or show up in a forecast. It is a picture of a system, sitting next to the system, unable to talk to it.
I have rebuilt this exercise maybe fifteen times now for B2B companies between Series A and about 300 people. What follows is how I run it so the output is something your CRM can hold and your team has to act on.
Why most B2B journey maps do nothing
Forrester has been saying a version of this for years: the mapping workshop is the easy part, and what separates a map that stays alive from one that ends up behind a cupboard is what happens in the weeks after. Their own research on why mapping efforts fail keeps landing on the same causes. No clear decision the map is meant to inform. No owner after the consultants leave. And no mechanism to update it when reality moves.
I would add a fourth that is specific to B2B, and I think it is the biggest one.
Most journey maps are drawn from the inside out. Somebody takes the company's existing funnel stages, relabels them with buyer-sounding verbs, and calls it a journey. "Awareness, Consideration, Decision, Purchase." That is not the buyer's journey. That is your pipeline with nicer nouns on it. The buyer does not know they are in your consideration stage. They know they have a problem their CFO is asking about, and they have four tabs open.
A journey map is a data model, not a diagram.
If a stage on your map has no exit criterion, no tracked event, and no owner, it is decoration. Three columns are what makes it real.
The part of the journey you cannot see
Before you map anything, sit with the size of the blind spot.
Gartner's number that gets quoted most is that B2B buyers spend about 17% of their total buying time in direct contact with any potential vendor. Split that across the three or four suppliers on the shortlist and a single rep is getting roughly 5% to 6% of the buyer's attention across the entire evaluation. The other 83% is peer conversations, internal debate, analyst pages, Reddit threads, and a Slack DM to someone who used the tool at their last job.
Share of total B2B buying time spent with all potential suppliers combined, per Gartner. Any single vendor gets about a third of that. Your CRM timeline is a record of the smallest part of the decision.
It got more extreme in 2026. Gartner's March survey of 646 buyers found 67% now prefer a rep-free buying experience, up from 61% the year before. 45% used AI tools during a recent purchase. But 69% of buyers still go to a salesperson to check what the AI told them, and self-service purchases produce measurably more purchase regret.
That combination is the interesting bit, and it is what your map has to solve for. Buyers want to do the work alone. They then want a human to confirm they did not get it wrong. Those are two different jobs at two different moments, and most companies build for neither. They build for a linear handoff that assumes the buyer wants to talk from the start.
Map the buying group, not a persona
The other thing that kills B2B maps is mapping one person.
The average B2B buying group is six to ten people, and each of them turns up with four or five pieces of research they gathered independently. Gartner reported in 2025 that 74% of buying teams experience unhealthy conflict during the process, and that groups who do reach genuine consensus are 2.5 times more likely to describe the outcome as a high-quality decision.
Read that again, because it changes where the work is. Your biggest competitor in most deals is not the other vendor. It is the buying group's inability to agree. If 74% of them are fighting internally, the highest-value thing your journey map can describe is not "how do we make the champion like us." It is "what does the champion need in order to win the argument on Thursday when we are not in the room."
That reframe produces a completely different map. Instead of one lane called The Buyer, you get lanes for the roles that actually show up: the person with the pain, the person who owns the budget, the security or legal reviewer who can veto in one email, and the internal skeptic who used a competitor at their last company. If you have not done that role mapping, start with buying committee mapping before you touch the journey at all. And if your buyer personas are still demographic sketches with a stock photo, fix those first too.
What a journey map looks like when it is a data object
Here is the reframe I use with clients. Every stage on the map has to carry three things, or it does not go on the map.
The right column is not prettier. It is worse to look at and far more useful, because every row of it can be a property, a workflow condition, or a report filter. That is the whole trick. You are not producing a picture of the journey. You are producing the specification for how the journey gets recorded.
The exit criterion is the part that does most of the work. Most sales pipeline stages are defined by what the rep did, not by what the buyer did. "Demo completed" is a rep activity. "Buyer has confirmed budget exists and named the approver" is a buyer state. Stages defined by buyer state are the only ones that forecast reliably, because a rep can complete an activity into a dead deal all day long.
How I actually run the build
Five steps, usually across three to four weeks with a client. Not two days.
Step two is the one people skip and it is the one that produces the arguments. On a recent build the leadership team was certain their sales cycle was 45 days. The export said the median from first meaningful touch to closed-won was 111 days, and that the long pole was a security review nobody had put on the map because it happened after the deal was already marked Verbal Yes. That single finding was worth more than the rest of the project.
Instrumenting it in HubSpot without inventing a data warehouse
Assume HubSpot, since that is what most of our clients run. The build is smaller than people expect.
Deal stages carry the buyer-state definitions. Put the exit criterion in the stage description field so it shows in the UI when a rep drags a card. Sounds trivial. It cuts stage-definition arguments by about half because the rule is visible at the moment of the decision.
Lifecycle stage stays separate from journey stage. This trips up nearly everyone. Lifecycle is a contact and company property describing the relationship. Deal stage describes one opportunity. Mixing them is how you end up with a customer who is somehow also a marketing qualified lead. If that distinction is fuzzy in your instance, lifecycle stage versus lead status is worth reading before you build anything else.
Use custom events for the pre-deal signals. Pricing page view, docs page view, security or compliance page view, second and third contact created at the same company domain. Those four tell you more about where a buying group is than any form fill. HubSpot custom objects are useful once you need to model something the standard objects do not hold, like a buying group as its own record with roles attached.
Add a buying-group role property on the contact. A simple dropdown: champion, economic buyer, technical evaluator, blocker, end user, unknown. Reps will not fill it in unless it is required to move stages, so require it at the stage where it matters. Deals where three or more roles are identified close at a visibly higher rate than single-threaded ones, and now you can prove that in your own data rather than quoting someone else's study. That measurement is what makes multithreading a management conversation instead of a nag.
Set a time-in-stage alert. If a deal sits past the 75th percentile for its stage and segment, someone gets a Slack message. We build these on n8n so they run on the client's own infrastructure and can pull from more than one system, but a HubSpot workflow does the job at smaller volumes.
The dark part, and what you can honestly proxy
You cannot instrument 83% of the journey. Anybody selling you attribution software that claims otherwise is selling you a model, not a measurement. I have written at length about why multi-touch attribution mostly lies, and journey mapping runs into the same wall.
What you can do is add three honest proxies and stop pretending about the rest.
Ask a self-reported question on the demo form: "How did you first hear about us?" Free text. It is messy and it is still the single most accurate source data most B2B companies have, because the buyer knows and your pixel does not.
Watch account-level web activity rather than person-level. Three unknown visitors from the same company in a week is a signal even when you cannot name them. Visitor identification makes that usable without becoming creepy about it.
Check what the AI tools say about you. With 45% of buyers using AI during the purchase and most of them then asking a human to verify it, the answer an LLM gives about your category is now part of the consideration stage whether you mapped it or not. That is a real journey touchpoint with no analytics behind it, which is the argument for treating answer engine optimization as a journey problem rather than a marketing channel.
What to measure once it exists
Four numbers per stage. Conversion rate to the next stage, median days in stage, the count of deals currently past that median, and win rate for deals that passed through the stage versus those that skipped it.
That last one is quietly the most useful. If deals that skip your "security review" stage win at the same rate as deals that go through it, the stage is not a stage. It is a habit. Cut it. I have removed more stages than I have added in these projects, and the maps got better every time.
Run the review monthly for the first six months. After that, quarterly is enough unless something moves by more than about 20%.
Mistakes I keep running into
Mapping the journey you wish they had. Founders describe an evaluation where the buyer reads the case study, books a demo, and decides. Real buyers go quiet for five weeks, come back with a question about SOC 2, and disappear again. Map the disappearances. They are the most important part of the timeline and they are always missing from the workshop version.
Treating post-sale as somebody else's map. The journey does not end at closed-won. Renewal and expansion behaviour is set in the first 60 days, and if onboarding is not on the same map with the same instrumentation, you get the classic handoff cliff where nobody owns weeks two through six.
Making it too detailed to maintain. Forty-touchpoint maps look thorough and are dead within a quarter. Six to eight stages with real exit criteria beats a masterpiece nobody updates.
Building it without the reps. If the people who have to log the data were not in the room when the stages were named, they will work around it. Two AEs in the definition sessions is the cheapest adoption insurance available.
Assuming the map is the deliverable. It is not. The deliverable is a CRM that produces a stage-conversion report you trust on a Monday morning. The map is just the design document for that.
Journey map that never made it out of Miro?
We turn mapping research into CRM stages, exit criteria, and reports your team actually runs on. Book a free 30-minute audit and we will show you the three fixes we would make first.
Book an audit →FAQ
What is customer journey mapping in B2B?
It is the exercise of documenting the steps a buying group goes through from first recognizing a problem to renewing a contract, including the parts that happen without you. In B2B the useful version is different from the consumer version in two ways: it maps a group of six to ten people rather than one shopper, and it treats internal consensus-building as a stage in its own right rather than background noise.
How long should a B2B customer journey map take to build?
Three to four weeks of part-time work, not a two-day workshop. The interviews take a week, pulling and arguing about the CRM timeline data takes another, writing and agreeing the stage definitions takes a third, and building the properties and reports takes a few days after that. Anything faster is a workshop output rather than an operating system, and workshop outputs do not survive contact with a quarter-end.
How many stages should a B2B journey map have?
Six to eight. Fewer than six and the stages are too broad to tell you where deals actually stall. More than eight and reps stop updating them accurately, which makes every downstream report wrong. If you feel you need twelve, you probably have two different journeys running (self-serve and sales-led, or new business and expansion) and you should map them separately.
What is the difference between a journey map and a sales funnel?
The funnel is your view of your pipeline: your stages, your activities, your forecast. The journey map is the buyer's view of their problem, including everything they do when you are not there. A well-built map produces a better funnel, because it forces stage definitions based on what the buyer has done rather than what the rep has done. If your map and your funnel have identical stage names, you have mapped your funnel and called it a journey.
Do we need a journey mapping tool, or is a spreadsheet fine?
A spreadsheet and your CRM are fine for almost everyone under 300 people. Miro or a dedicated mapping tool is useful for the interview synthesis phase and useless afterwards, because the map has to end up as configuration in the system your team works in. Spend the budget on the data work instead. Buying a journey orchestration platform before your stage definitions are agreed is a fast way to automate a process nobody trusts.
Rebuilding how your revenue data is structured? That is most of what we do at Ziel Lab. Have a look at our CRM and RevOps work, our go-to-market builds, or get in touch.