Here's a test you can run on any existing journey map in about five minutes. For each stage, emotion and pain point on it, ask: which customer said this? If you can't name them for most items, the map documents what your organisation believes — rendered convincingly enough that nobody questions it any more.
The room determines the output
Most journey maps are produced the same way. Twelve colleagues, a workshop, sticky notes, a facilitator. Everyone contributes what they think customers experience. The output gets designed properly and hung on a wall.
What that process reliably produces is consensus. What it can't produce is evidence, because nobody in the room is a customer.
The failure isn't that the assumptions are wildly wrong — they're usually partly right, which is worse. A map that's 70% accurate and treated as 100% accurate sends you confidently in a slightly wrong direction, and the confidence comes from the artefact looking authoritative rather than from anything underneath it.
The whole discipline in one rule Every item on the map should be traceable to a specific person who said it. Anything you can't trace is a hypothesis, and should be labelled as one.
The fix isn't a better workshop. It's replacing most of the workshop with conversations.
Eight interviews beat a room of twelve
Qualitative research hits diminishing returns fast. For a single segment, eight to twelve conversations is usually enough — by around the eighth you're hearing variations of what you've already heard, which is the signal to stop rather than a reason to worry.
That's a week of work and it produces something the workshop never can.
Talk to recent buyers, not long-standing customers. Someone who bought three years ago has reconstructed the decision into a tidy story. Someone who bought last month still remembers the confusion.
Talk to people who didn't buy. Harder to arrange and disproportionately valuable, because they experienced the failure points your customers pushed through. Closed-lost contacts will often speak candidly precisely because there's nothing at stake.
Talk to the people who talk to customers. Sales and support are not a substitute for customer conversations, but they're a good source of hypotheses to test in them — and they'll tell you where the map is wrong faster than anyone.
The backwards interview
One technique worth more than the rest, and it's simple.
Ask people to narrate backwards from the decision. "You signed on the fifteenth. What happened immediately before that? And before that?"
Forward narration produces a clean retrospective story with the messy parts smoothed out — people reconstruct their decisions as more rational than they were. Working backwards surfaces the steps they'd otherwise skip: the colleague they asked, the comparison they abandoned halfway, the search that went nowhere, the three weeks where nothing happened because someone was on holiday.
And ask what nearly stopped them at each point rather than what they liked. Obstacles are actionable; enthusiasm mostly isn't. Those obstacles are also directly reusable as a content plan, which is the connection to building content around sales objections — the same raw material serves both.
Systeme.io
Affordable all-in-one marketing platform with funnels email and course builder included
Best for: All-in-One Platform
What analytics can't see
The second structural problem, and it's grown considerably.
A map built from analytics maps the visible portion of the journey — which is now the minority of it. What sits outside your data entirely:
- Research inside AI assistants. People form views about categories and vendors in conversations you'll never see, arriving already decided about things you never got to address.
- Peer conversations. Slack groups, communities, a message to a former colleague. Frequently the most influential input and completely invisible.
- Internal discussion. In B2B especially, most of the decision happens in meetings you're not in — the dynamic covered in how buying committees actually decide.
- The gaps. Long stretches where nothing happens because of budget cycles, competing priorities, or someone being away. These look like disengagement in your data and are usually just life.
This is also why journeys don't resemble the funnels drawn around them — the argument worked through in how buyer journeys changed in 2026. For mapping purposes the practical consequence is narrower: interviews reach the invisible parts and analytics don't, which is the main reason the conversations aren't optional.
Journey map versus funnel audit
These get conflated and they answer different questions.
| Funnel audit | Journey map | |
|---|---|---|
| Describes | Your process | Their experience |
| Built from | Your measurement | Their account of it |
| Answers | Where do people drop out? | Why, and what happened elsewhere? |
| Covers | What you can see | Including what you can't |
| Best used | To locate the problem | To understand it |
They work best in sequence: an end-to-end funnel audit tells you where the loss is concentrated, then interviews with people who dropped at that point tell you why. Running the map first, without knowing where the problem sits, tends to produce a comprehensive document with no focus.
What to skip
Journey mapping templates accumulate rows. Several aren't worth filling.
The invented emotion row. Nearly every template has one, and nearly every map fills it with adjectives chosen in a workshop. "Frustrated," "excited," "anxious" — drawn by people who never asked. If you interviewed customers you can record what they actually described; if you didn't, leave it blank rather than fabricating it.
Exhaustive touchpoint inventories. Listing every possible interaction produces a wall of items with no weighting. Three touchpoints that determine the decision matter more than forty that exist.
Persona proliferation. Six personas each with their own journey is usually five more than anyone will use. Map your primary segment properly first. If you're generating personas with AI assistance, the caveats in using AI for customer research apply doubly here — synthetic personas are hypotheses, not evidence.
The beautiful artefact. Design effort on the map correlates poorly with use. A rough version that gets revised is worth more than a polished one that gets laminated.
A workable process
- Pick one segment and one journey. Not all customers, not all paths. Scope is what makes this finishable.
- Look at the data first. Where does measured drop-off concentrate? That tells you where to point the interviews.
- Write down your assumptions before interviewing. Explicitly, as a list. This is the workshop's real job — generating hypotheses to test, not conclusions to document.
- Interview eight to twelve people. Recent buyers, some non-buyers. Backwards narration, five questions, thirty minutes.
- Build the map from what they said, attributing each item. Mark anything unsupported as a hypothesis rather than deleting it.
- Compare against your assumption list. The gaps between what you assumed and what you heard are the findings. This step is where the value is, and it's the one most often skipped.
- Pick three things to change, with owners and dates.
- Refresh annually, or sooner if the market moves.
Step three deserves emphasis. Writing assumptions down before the research is what turns the exercise into a test rather than a documentation project — and it protects against the common failure where interview findings get quietly absorbed into what everyone already thought.
Turning a map into changes
The map's output should be a short list of specific fixes, and they tend to cluster in predictable places.
Where people had to work something out for themselves. Every point in an interview where someone says "I couldn't tell whether…" is a gap you can close with information. These are cheap fixes with immediate effect.
Where they needed something you didn't provide. Frequently a document for internal approval, a comparison you avoided making, or a reference in their sector.
Where the handover was rough. Marketing to sales, sales to onboarding. Transitions between teams are where journeys break most reliably, because each side assumes the other handled it.
Where they went somewhere else to find out. If people consistently left to check reviews, ask a peer, or search a comparison, that's a signal about what you're not saying — and increasingly about where you need to be present, given how much research now happens through AI-mediated answers rather than on your site.
Where they couldn't find something that existed. Often a navigation and labelling problem rather than a content one, which puts it in the territory of information architecture.
Why maps stop being used
Three predictable causes, all avoidable.
Nobody owns it. A map produced by a project team and handed over belongs to no one. Assign a person and a review date at the point of creation.
It's too big to consult. A map covering six segments across twelve stages with forty touchpoints is a reference document nobody opens. One segment, clearly, gets used.
It went stale silently. Behaviour keeps moving while the artefact stays fixed. And an out-of-date map is more dangerous than none, because it carries the authority of research without the currency — people trust it precisely because someone once did the work.
The same decay affects most structural marketing artefacts, and the remedy is always the same: an owner, a date, and a small recurring refresh rather than a large occasional rebuild.
If the honest constraint is that nobody has a week to run interviews alongside everything else, that's a capacity problem rather than a method one — and it's where an outside marketing partner conducting the conversations often produces more candid answers than an internal caller would anyway.
The short version
Run the traceability test on any map you already have: if you can't name which customer said each item, it documents your assumptions. Replace most of the workshop with eight to twelve conversations, weighted toward recent buyers and people who didn't buy. Ask them to narrate backwards from the decision, because forward narration smooths out the messy parts that matter. Write your assumptions down before you interview, so the exercise becomes a test rather than a documentation project — the gaps between assumed and heard are the actual findings. Skip the invented emotion row and the exhaustive touchpoint list. And give it an owner and a review date, because a stale map carries the authority of research without the currency.
Confident about your customer journey without having asked them?
We run the customer conversations and turn what they reveal into specific, prioritised changes.
Explore Brand Strategy →Frequently asked questions
What makes a customer journey map useless?
Building it entirely from internal assumptions. A workshop of colleagues produces a diagram of what the organisation believes about its customers, rendered convincingly enough that it stops being questioned. The test is simple: for each stage, emotion and pain point on the map, can you name which specific customer said it? If most items fail that test, the map documents your beliefs rather than their behaviour, and decisions made from it will inherit whatever those beliefs got wrong.
How many customer interviews do you need for a journey map?
Fewer than most people expect — usually eight to twelve for a single segment, because qualitative research reaches diminishing returns quickly. By around the eighth conversation you tend to be hearing variations of things you have already heard, which is the signal to stop. That is a far better use of a week than a workshop with twelve colleagues, because it produces evidence rather than consensus. Interview recent buyers rather than long-standing customers, since recent buyers can still remember the decision.
What is the best way to interview customers about their journey?
Ask them to narrate backwards from the decision. Start at the moment they chose and work in reverse — what happened immediately before that, and before that. Forward narration produces a tidy retrospective story that skips the messy parts, whereas working backwards surfaces the steps people forget they took: the colleague they asked, the comparison they abandoned, the search that went nowhere. Ask what nearly stopped them at each point rather than what they liked, because the obstacles are the actionable material.
How is a journey map different from a funnel?
A funnel describes your process — the stages you have defined and can measure, viewed from inside the business. A journey map describes the customer's experience, including the parts that happen nowhere near your properties and produce no data at all: conversations with peers, internal discussions, research inside AI assistants, and the long periods where nothing visibly happens. The funnel tells you where measured drop-off occurs; the journey tells you why, and frequently locates the cause somewhere the funnel does not look.
How often should you update a customer journey map?
Refresh the evidence roughly annually, and treat any significant change in your market, product or buyer behaviour as a trigger to revisit sooner. Maps decay quietly because the behaviour they describe keeps moving while the artefact stays fixed on the wall, and an out-of-date map is more dangerous than no map because it carries the authority of research without the currency. A short refresh of three or four interviews is usually enough to confirm whether the existing picture still holds.
