B2B Strategy

A B2B Service Experience Map: Keep Customer Evidence Separate From Assumptions

A request disappears from the customer's view before it disappears from the team's queue.

The customer sends a message. Someone forwards it. A specialist investigates. The customer sees an acknowledgment, then silence.

A service experience map follows what the customer can encounter across that episode. It gives each recorded step an evidence reference. It leaves the person's thoughts unknown when nobody asks about them.

GOV.UK's experience mapping guidance starts with captured user experiences and organizes events into stages. It covers actions, channels, needs, and evidence about reactions over time. Its context is government services. The guidance asks for several users and favors services spanning weeks or months. This shorter fictional episode teaches the worksheet. A real consolidated map still needs research across relevant users.

Every record in the worked example is fictional. The map demonstrates a method. It reports no customer research or delivery results.

Define the actor and the episode

Choose one person pursuing one outcome. Write where the episode starts and what evidence would establish its finish.

For this example, a purchasing coordinator uses a documentation support service. The coordinator discovers two versions of a supplier review checklist. They need the correct version before preparing an internal review pack.

The episode starts when the coordinator notices the mismatch. It ends when they confirm that the replacement checklist opens and matches the requested version. Sending an attachment alone does not meet that example's finish condition.

The channels are a resource page, email, a phone conversation, and a document link. Contract negotiation, procurement approval, and the provider's internal review process sit outside this map.

Write those exclusions above the worksheet. Otherwise a team can keep expanding the map until it contains every relationship surrounding the request.

The actor is a role with a task. Avoid a fictional biography full of assumed attitudes. Their budget, seniority, and general comfort with technology stay unknown unless those details matter and have support.

If the question concerns the first weeks of delivery, use the separate client onboarding framework. Here the task is narrower: understand one request from the customer's perspective.

Label what each note can establish

Use four explicit labels: observed action, reported statement, interpretation, and unknown.

An observed action records what a researcher sees or an artifact documents. A sent message can establish its wording and timestamp. It cannot establish that the recipient reads it.

A reported statement records what a participant says about the episode. Keep the statement tied to the person, question, and moment. A later recollection can be useful without becoming a direct observation of the original event.

An interpretation names a researcher's proposed explanation. Write the evidence it draws on and the alternative still open. Treat it as a question to examine.

Unknown means the available record does not answer the question. Leave that word visible. An empty cell is easy to mistake for a step that nobody needs to investigate.

For the fictional episode, observation O1 records the coordinator opening two checklist pages. Statement S1 records their account that they cannot identify the current version. Interpretation I1 proposes that inconsistent labels contribute to the uncertainty. The coordinator's feelings during the wait remain unknown.

Do not convert the timestamp between messages into a frustration score. Do not draw an emotional low because a team member expects one.

The customer interview workflow develops the related discipline of separating research notes from editorial conclusions. Carry that evidence boundary into the map.

Build an evidence register before the grid

Give each usable note a stable identifier. Record its type, origin, scope, and access location.

An artifact is a record format, not a fifth evidence label. Classify what it establishes as an observation, report, interpretation, or unknown. An email can document transmission while carrying a reported statement about file usability.

For a real map, reference the controlled research record. Keep private messages, names, and confidential documents out of broadly shared diagrams. A useful reference explains what the evidence supports without copying everything into the cell.

The example uses this synthetic register:

Reference Type Fictional record Limit
O1 Observed action Coordinator opens two checklist pages in a staged walkthrough Does not establish what other customers encounter
S1 Reported statement Coordinator says the page labels do not identify the current version Does not establish the cause of the labeling mismatch
A1 Observed action (email artifact) Sent request asks which checklist applies Does not prove the provider reads it
A2 Observed action (inbox artifact) Acknowledgment in the coordinator's inbox gives a request reference Does not establish that the coordinator reads it
S2 Reported statement Coordinator says they phone to ask what happens next Does not establish whether they first seek an answer elsewhere
A3 Observed action (inbox artifact) Reply in the coordinator's inbox supplies a replacement link and version label Does not prove the file opens
A4 Reported statement (email artifact) Coordinator reports that the replacement opens and matches their request Does not independently verify file use
I1 Interpretation Researcher proposes that inconsistent labels contribute to S1 uncertainty Cause remains untested; other explanations remain open

These are invented records for teaching. They are not anonymized customer evidence.

Keep conflicting notes together. If a participant remembers receiving a phone update but the record contains no call note, retain both facts. Missing documentation does not prove the conversation never happens.

Copy the service experience map worksheet

Use a row for each meaningful stage. Keep actions concrete enough to check against a note.

Stage Customer goal Action Channel Evidence reference Obstacle Open question
[Stage in the episode] [Outcome the actor seeks] [Observed or reported action] [Where the interaction occurs] [Record ID and evidence type] [Supported barrier or unknown] [Question the evidence cannot answer]
[Next stage] [Specific goal] [Action with its evidence label] [Channel or unknown] [Record ID or missing] [Barrier or unknown] [Next research question]

Put the map owner, episode boundary, research dates, and revision date above the table. Add a note explaining the evidence labels.

Customer goals below are inputs to the fictional scenario. In a research map, cite support for each goal. Label inferred goals as interpretations.

Name stages after the customer's work: identify the correct checklist, ask for clarification, check progress, verify the replacement. Internal queue names can obscure what the person is trying to do.

Keep the customer goal distinct from the action. Opening an email is an action. Knowing whether the request needs another input is a goal.

A channel identifies where the encounter occurs. It does not establish that the channel is preferred. Email use in one episode supplies no basis for declaring that all customers want email.

An obstacle needs evidence too. If nobody can say why a stage stalls, write the observed condition and leave its cause open.

Walk through the fictional request

The fictional register supports this draft. Its gaps remain part of the map.

Stage Customer goal Action Channel Evidence reference Obstacle Open question
Identify Find the applicable checklist Opens two versions Resource page O1, observed; S1, reported Current version is unclear to this actor Which label does the actor use to choose?
Ask Get a version decision Sends clarification request Email A1, observed No answer yet What response does the actor expect?
Acknowledge Know the request exists Receives reference number Email A2, observed Next step is unspecified Does the acknowledgment explain enough?
Check Understand what happens next Phones support Phone S2, reported Reason for choosing phone is incomplete What does the actor try before calling?
Receive Obtain the requested version Receives replacement link Email and document link A3, observed File usability remains unverified Can the actor open the file?
Confirm Prepare the review pack Confirms file and version Email A4, reported No further obstacle recorded Does another task begin after confirmation?

There is no invented satisfaction curve. The table shows an unanswered progress question and a documented customer report. File use remains self-reported.

The provider may know exactly which specialist holds the request. That knowledge does not establish what the coordinator sees. Keep internal assignment and review activity in a separate service blueprint.

A confirmation closes this scoped episode. It does not prove the entire purchasing process succeeds or that the customer prefers the service.

Split journeys when the sequence changes

A consolidated map needs a coherent sequence. GOV.UK recommends separate maps when user groups follow different stages and materially different steps. Its guidance also recommends reviewing the draft with people who observe research or deliver the service.

Consider a second fictional coordinator who cannot access the resource page. They start with a phone request and receive a permitted document through a different channel. Their episode includes an access decision before version selection.

Do not insert an access stage into every row of the first map. Keep the two sequences separate until research establishes a useful common pattern.

A third fictional requester reports that a checklist contains a factual error. That task requires validation of content, rather than choosing between two existing versions. It deserves its own scope and evidence.

Separate actors when they hold different responsibilities within the same episode. The requester prepares the pack. Their approver decides whether to accept it. Combining their actions can imply that one person sees or controls both steps.

Record why you split a map. The reason should name a task or sequence difference. A demographic label alone does not explain a service requirement.

One fictional episode cannot establish how frequently any path occurs. Leave prevalence unmeasured.

Turn a gap into a research question

The check-progress row has a useful gap: what does the coordinator try before calling?

Write a follow-up question that invites the person's sequence. Ask them to walk through the period after the acknowledgment. Ask which information they seek and where they look.

Avoid asking whether an unclear email frustrates them. That wording supplies both an explanation and a reaction before the person describes either.

The proposed research task is to review another permitted episode with a requester. Record the acknowledgment they receive, their next action, and their account of that choice. Keep the outcome unknown until the research happens.

If the person says they phone because a deadline approaches, update the reported-statement field. That answer does not establish that every caller needs a faster response.

If the artifact includes a next-step explanation that the person overlooks, preserve the difference. The next question may concern language, placement, or relevance. The map alone cannot choose among those explanations.

Give the research question an owner and a review point. A visible uncertainty without a next action can remain in the diagram indefinitely.

Review the map against the records

Bring the draft to someone who captures the evidence and someone who understands the service touchpoints. Ask each participant which statements they can support.

Work across the table from left to right. Check that the actor remains the same, each action has a reference, and the finish condition matches the scope.

Then read only the unknowns. Decide which question affects the next service decision. A map can contain many gaps while only one requires immediate research.

Maintain a short change note when new evidence revises a stage. Record the changed claim and the supporting reference. Keep the previous interpretation distinguishable from the current evidence.

Use buyer journey content mapping when the decision concerns which article or resource a buyer needs. This service map instead records how a person encounters the service. A missing explanation may later inform content work, but the map does not establish search demand.

Share a summary with its scope and evidence limits intact. Remove private detail without removing the uncertainty that changes the decision.

Keep the unknowns visible.

Sources

Topic map: Customer and usability research

Articles in this cluster, then the 3 clusters that share the most topics with it.

This cluster: Customer and usability research

Neighbor: Topical authority and keyword strategy

Neighbor: Content production with AI

Neighbor: Technical SEO, crawling, and migrations