Seen or assumed? The self-test that shows how much fiction is in your journey map

The self-test: take your latest process overview

Fetch the most recent overview your company has of the customer's path. A journey map from the last workshop, a process map from quality management, the service team's onboarding overview. You don't need a new tool. A printout and a pen are enough.

Go through the overview from left to right and mark each step with one of two words: "seen" or "assumed". Then do the same with every transition between two steps, meaning every arrowhead.

Transitions matter more than steps. That is where ownership changes, often the system too. And that is where something is most likely to slip through.

Only what you can back up counts as "seen":

• System data, for example timestamps in the CRM between two status changes
• Support tickets or call notes that concern exactly this step
• Recordings or logged customer contacts
• Shadowing real customer cases yourself
• Direct conversations with customers about exactly this stage

Everything else counts as "assumed":

• Workshop results
• Descriptions of the target process
• The experience of individual employees ("this is how it works here")
• Evidence so old that the process, tool or customer channel has changed since

Be strict. If nobody in the room can name a specific source, the mark is "assumed". Even if everyone is sure it is right.

Don't do the test alone. Sit down with sales and service, because each department knows different transitions. Appoint one person with a single job: to ask, for every "seen", how we know.

Then count. How many steps and transitions have you marked, and how many of them are "assumed"? Divide the second number by the first and write the result at the top of the overview.

This figure is not yet a verdict. Assumptions are a legitimate starting point for any map.

Why most journey maps show the inside view

A typical journey map is created in a meeting room in an afternoon. Sales, service and marketing each bring their part of the customer's path, and at the end a neat sequence of steps hangs on the wall. The people in the room are the ones who designed the process or run it every day. No customer is in the room.

What comes out is the target process: "Quote sent, customer signs, handover to onboarding, kick-off." The map does not show whether customers actually move through it this way. It shows how the company imagines the process.

Most of the maps we get to see are created exactly like this. It gets tricky when nobody checks afterwards: the workshop ends, the map hangs on the wall, and from then on the hypotheses on it serve as the basis for important decisions.

One reason lies in perspective. Most of the people involved see only their own stage. Sales knows the path up to the signature, service knows the tickets, marketing knows the campaigns. Nobody knows the whole path from the customer's point of view. A second reason: what people say about their behaviour often does not match what they actually do. A meta-analysis in Nature Human Behaviour (2021) compared self-reports of digital media use with logged usage data and found that self-reports only rarely reflect actual use accurately. What a team assumes about its customers' behaviour is even further removed.

Why this happens particularly often in mid-sized companies

Many mid-sized companies also have a structural gap. Customer data is scattered. The CRM knows the order, the ticketing system the complaint, the inbox the query, the analytics tool the website visit. These systems are rarely connected, and there is usually no dedicated BI (business intelligence) team to bring the data together.

So observed customer behaviour exists, but it is out of reach. If you ask in the workshop how long the customer waits for the first contact after signing, you get an estimate instead of a figure. The estimate goes onto the map and looks exactly like a step that is backed by evidence.

The question of ownership fits this picture. In the MiCX-X survey of 1,000 mid-sized companies in the DACH region (in German; December 2024 to March 2025), customer experience sits with marketing in 26% of companies and with management in 22%, with other departments sharing the rest. The authors frequently see a lack of clear responsibilities and no systematic use of data, even though many companies already invest in customer surveys and metrics.

The teams fill a data gap with the knowledge they have. The problem is the structure: as long as no one measures the customer's path as a whole, every map is built from the inside view. And as long as the map does not show which step is backed by evidence and which is assumed, you cannot tell by looking at it.

Where assumed transitions cost you customers

Look at your marks. A pattern usually stands out: the steps themselves are often well documented. The quote is created, the order recorded, the ticket opened. For these there are records, timestamps and people responsible. The gaps lie in between.

Typical transitions in mid-sized companies:

• Quote to order. The CRM says "sent". No system shows whether the customer spends the following days comparing, has a question or cannot pass the documents on internally.
• Sales to onboarding. Sales reports the deal as closed, and the project team takes over. Often nobody knows exactly when the handover reaches the customer.
• Service ticket to solution. The ticket is closed. Whether the problem is solved for the customer is another question.
• Tool to tool. When data moves from email to the CRM, from the CRM to the ticketing system or to accounting, manual steps appear. Each of them can get stuck.

Why here of all places? Each tool sees only its own stage. No system, and often no department, sees the transition as a whole. That is where the data is thinnest, and that is where assumptions fill the gap. Who is responsible for which customer data at such transitions is settled by a data ownership mapping.

We see two patterns here again and again.

The customer is waiting for a response that counts as done internally. Sales has handed over, and the status says "completed". For the company, the step is finished. For the customer, nothing has happened for days. The map shows a neat arrow. The waiting time appears nowhere, because nobody measures it.

The customer uses a channel the map does not know. The map assumes a contact form or the responsible contact person. In reality, the customer calls the person they know from an earlier project, or writes to a general address. These contacts end up in none of the systems the map depicts. If something goes wrong there, nobody sees it.

Internally, everything is in order in both cases. The process description is correct, and responsibilities are clear. The customer just does not behave as in the target process.

If a map like this becomes the basis for decisions, budget and attention flow into the steps that are well visible anyway. The transitions where customers actually get stuck are considered uncritical.

Your team only finds out how many customers you lose at these points once the transition is backed by evidence.

The mandatory column: data source per transition

The self-test shows once how much assumption is in your overview. To stop this from disappearing again after the first workshop, the map itself needs a fixed place for it.

Every friction map, meaning the overview of the points where customers get stuck, gets a mandatory "data source" column for each transition. It contains exactly one of three values:

• observed behaviour: system data, tickets, recordings, shadowing
• customer statement: what customers say about this stage in conversation
• team assumption: everything that comes from workshops, process descriptions or experience

In the self-test, the first two count as "seen" and the third as "assumed". Observed behaviour carries more weight than a statement, which is why you keep the two separate.

For context: the column is our proposal. It is not an industry standard. The only widespread advice is to check assumption-based maps against customer research afterwards. The column shows whether that has happened.

What the table can look like

The example is filled with typical entries rather than real customer data.

Each column has a job:

• Transition: where the customer moves from one department, tool or channel to the next.
• Suspected friction: what you think could go wrong there.
• Data source: one of the three values above. There are no others.
• Evidence: what exactly the assessment is based on. An employee's memory is an indication. It is not evidence. That is why the second entry stays at "team assumption".
• Status: "unchecked", "friction confirmed" or "no friction found".
• Next check: how an assumption becomes evidence, using data that already exists.

The rule: no evidence, no all-clear

In many workshops, a transition counts as unproblematic as long as nobody complains. That is a fallacy. Customers who get stuck at a transition often don't get in touch at all. They simply leave.

So the rule is: a transition without evidence is never classed as "uncritical". It stays at "unchecked" until there is evidence.

Evidence is not automatically reliable either. Four questions help with the assessment: how much knowledge is there? How cleanly was it gathered? How current is it? How broad and deep is it? Thin, outdated knowledge gathered in passing means high risk. Even if the column says "observed behaviour".

What you get from it as a decision-maker

A typical map shows where problems are suspected. With the mandatory column, it also shows where your company knows nothing.

For budget decisions, that is the more important information. A transition with known friction can be fixed. A transition with no data may be costing you customers right now without anyone noticing. How you can make the return on a journey measure provable before it starts has an article of its own.

From "assumed" to "seen": sources without a new tool

For most assumed transitions, you need neither an analytics project nor a BI team. A lot of the evidence is already in your systems, and you can get the rest with a phone call. The data is usually there. What is missing is someone who reads it specifically for one transition.

The order below follows the weighting from the column: first sources that show behaviour, then conversations.

1. Analyse existing system data

• CRM timestamps between status changes. Almost every CRM stores when a record changes from "quote" to "order" or from "won" to "in progress". Export these times for the last few months and look at the intervals. If a transition takes a long time or the waiting times vary widely, that is a first piece of evidence. If the status fields are not maintained at all, that is a finding too: nobody is watching this transition.
• Sort support tickets by transition. Take a sample and assign each ticket to the stage the customer was in at the time: shortly after signing the contract, during onboarding, after the first invoice. Where tickets pile up, a transition is generating queries. A spreadsheet with one extra column is enough for this.
• Email response times. How long does a customer wait for the first reply after an enquiry? How long after the handover from sales to the next team? Many email and ticketing systems show this without an add-on module. This is where the typical pattern shows: done internally, still open for the customer.
• Drop-off points in forms. If enquiries, registrations or orders go through web forms, your existing visitor statistics usually already show on which page or at which field visitors stop.

2. Shadow real customer contacts

Sit with service for an hour, listen in on an onboarding meeting, accompany a field sales visit. This is no inspection of staff. You are testing a single question: does what is on our map actually happen at this transition?

Write down deviations word for word, for example channels customers use to get in touch that are missing from the map. A few hours can be enough to confirm or refute several "assumed" marks.

3. Read tickets and call notes in the original wording

Categories smooth things over. Behind "delivery query" you cannot see that the customer is asking for the second time. So for every critical transition, read a handful of tickets or notes word for word. Look out for phrases such as "as discussed", "I had already" or "nobody got back to me". They reveal transitions that your map lists as done, while the customer experiences something else.

4. Talk briefly to newly won and lost customers

Call a few customers who have joined in recent weeks, and just as many who have dropped out or turned down a quote. A short conversation with three questions is enough:

• Where did you have to wait?
• Where were you unsure what would happen next?
• What almost made you give up?

New customers still remember the friction at the start. Lost customers can tell you at which transition they dropped out. Enter the result as "customer statement" and, where possible, compare it with system data.

How to keep the effort small

• One source per transition is enough to start with. You don't need complete proof, just one piece of evidence that moves the status from "unchecked" to "friction confirmed" or "no friction found".
• Samples and exports instead of interfaces. A spreadsheet is enough. You don't need new systems or integrations for this.
• Record every finding with its source and date. How current a piece of evidence is partly determines how reliable it is.
• Note missing sources openly. A transition that neither a system nor an employee can see is a finding in itself. It belongs at the top of the list.

Prioritising: which assumed transitions first

After the self-test, you have a list with many "assumed" marks. Checking them all at once would be a major project. You need an order.

The rule: first check the transitions that are assumed and that revenue depends on. Both must apply. An assumed transition with no link to revenue can wait. A transition linked to revenue that you have already backed with evidence is fine for now.

What revenue depends on

In most mid-sized companies, three kinds of transition decide where money is made or lost:

• First order: from quote to signature and from signature to first delivery of the service. This is where a prospect becomes a customer or not, and where it is decided whether they want to stay after the first few weeks.
• Renewal: the moment before the contract ends, a repeat order or a follow-up order. If you don't look here, you only notice churn when the order fails to arrive.
• Escalation: from the customer's problem to the solution. A ticket that is done internally but still open for the customer costs trust at exactly the moment they are thinking about switching.

The matrix: evidence times business relevance

Field 1 is your work list. Field 2 deserves a second look, because evidence ages too. The four questions on reliability from the section on the mandatory column help with this assessment.

An order for the coming weeks

1. Sort field 1. If there are several transitions there, start with the one carrying the most revenue. When in doubt, that is the renewal, because existing revenue is at stake there.
2. Choose one source per transition. CRM timestamps, tickets, response times or a conversation with a recently lost customer. One is enough for the first step.
3. Finish one transition after the other. Enter the data source, evidence and status. Only then move on to the next.
4. Then field 2. Once the most urgent assumptions have been clarified, check the evidenced, revenue-relevant transitions for how current they are.

You work through a short list. After each step, your leadership team knows what customers experience at one more critical point.

What changes when the map is honest

As soon as a journey map hangs on the wall, it becomes a basis for decisions. It is used to decide which process gets reworked, which tool gets added and where the team invests its time. If the transitions rest on assumptions, the budget ends up where the team expects problems. Customers often get stuck somewhere else.

An honest map changes three things:

• Investment follows the evidence. Improvements start where observed behaviour shows friction, rather than where the loudest voice in the workshop suspects it.
• Blind spots become visible. Management learns where nobody knows exactly what is happening. That is often where the expensive surprises lie.
• Discussions get shorter. At every transition, the first question is: how do we know? The answer is in the column.

That is why the data source column belongs permanently in every future map, beyond this one revision.

Your next step for this week: take an hour with your leadership team. Do the self-test on the current process overview and decide which two or three transitions you will check first.

Don't be alarmed by a high share. With it, you know for the first time where you stand.

If you would like support with this: UNIT ONE supports stocktakes of this kind as part of journey architecture, from the first mark to the map with a data source column. A first entry point is the Reactivation Audit.

And until then, the question for your next workshop: which arrow on your map is actually backed by evidence?

Frequently asked questions

What counts as "seen" and what counts as an assumption? "Seen" is whatever can be backed up with system data, tickets, recordings, shadowing or customer conversations. The experience of a long-serving head of sales is valuable, but it does not replace evidence.

Is a journey map with many assumptions worthless? No. It only becomes worthless when assumptions are not labelled and decisions rest on them as if they were backed by evidence.

How often should the data source column be updated? Whenever a process, a tool or a customer channel changes, and every time the map is revised.

Sources

• Nature Human Behaviour: meta-analysis, doi.org/10.1038/s41562-021-01117-5
• marktforschung.de: Customer experience has arrived in mid-sized companies, but is often fragmented and without a strategy (in German)