01 — Theory
What has to be in place before anyone is interviewed
Five questions decide whether the capture in step 6 runs in days or in weeks. None of them needs a tool; all of them need a decision.
What do you need before the first interview?
Four things, in this order. The documents that describe the process as designed: a handbook chapter, a work instruction, a policy, an old diagram. The org chart of the units involved, so that lanes have names before the first conversation. The list of systems and the data objects that travel through the process: forms, reports, tickets, credit notes. And access: a booked room or call, consent for recording where you record, and a person per unit who has agreed to talk. Read the documents before the interview, not after it; the differences between the document and what people say are the most valuable part of the interview, and you only see them if you know the document.
Whom do you interview, in what order, and how many per role?
Start with the person who sees the process end to end, usually the team lead of the unit where it begins, and let that first conversation name the others. Then one person per lane who actually does the work, not the person who manages it. Where a lane has several people doing the same work, two interviews are enough to tell routine from personal habit; a third adds little. Interview the hand-over partners in the order of the flow, so that each conversation can test what the previous one said about the hand-over. Leave the approver and the control functions for last: they know their check, but rarely the path that leads to it.
Interview, workshop, documents, shadowing or task mining: when does each fit?
Documents give the designed process and are the right starting point when they exist. Interviews give the process as it runs, one perspective at a time, and are the default for anything with decisions, exceptions and hand-overs. A workshop fits when two units disagree about the same hand-over and the disagreement is the finding; it is expensive in people and should be used for that, not for the first description. Shadowing fits for work that people cannot describe because it is habit: screen work, a desk with many small interruptions. Task mining records screens and delivers click paths; it needs consent, an installation and a works-council agreement, and it does not see phone calls, hallway decisions or the reason behind a step. The comparison of interview and task mining sets out where each one is strong.
What hypotheses should you walk in with?
Three, written down and dated. One about where time is lost, one about where the designed process and the real one differ, one about which decision rule is unclear. A hypothesis is not a conclusion; it tells the interviewer which answer would be surprising, and surprising answers are the ones worth a follow-up question. Without hypotheses the interview records everything equally and the analysis in step 7 starts from zero. With hypotheses that turn out wrong, the capture has still produced a result: the team now knows what it believed and why that was not the case.
Which level of detail and which modelling conventions do you agree up front?
Decide the level of detail by the purpose: a process to be analysed for waiting time needs every hand-over, a process to be documented for new staff needs every step that uses a system, and a process map for the steering committee needs neither. Agree the conventions before the first diagram exists: lanes are organisational units from the org chart, not people; a step is a verb plus an object; a gateway carries a question and its answers carry the rule; systems and documents are attached to the step that uses them; control steps are marked. Write this down on one page, and every interview after the first will produce diagrams that fit together. The BPMN guide covers the notation these conventions rest on.
02 — Better with patchley
What changes in this step, and what stays with the team
Preparation is mostly decisions, and patchley does not take decisions. It removes the reading and the set-up work that usually sits between the decision and the first interview.
Someone reads the handbook by hand before the first interview.
With patchleyUpload the handbook chapter as PDF or DOCX, name the process, and walk into the interview with a first draft diagram to test against what people say.
Lanes and conventions get settled interview by interview, and the diagrams do not fit together.
With patchleyThe org chart and your organisation's templates are set up before you start, so every diagram uses the same lanes and every analysis the same questions.
Questions come only from the prepared list; what the list did not foresee is missed.
With patchleyWith responses switched on, patchley asks follow-up questions as you describe. In the discussion it points out what is unclear or missing in the diagram.
Choosing whom to talk to, and the hypotheses. The tool can show a draft; only the team knows who runs the process and what it suspects about it.
03 — The running case
Tannhof prepares the complaint-handling interviews
Tannhof Armaturen GmbH is a fictional valve maker; every figure in this case is invented. The process is customer complaint handling, and the team wants to know where a complaint spends its time. What exists at the start is the quality handbook, procedure QM-14, several years old.
How Tannhof chose its interviewees
The handbook names five units: Customer Service, the Complaints Team and the Lab in Quality, the Returns Desk in Logistics, Purchasing and Accounts Receivable. The team booked two interviews first: the team lead of Customer Service, where every complaint starts, and the lead of the Complaints Team, which the handbook puts at the centre. Everyone else was to be interviewed only if those two conversations pointed at them.
What thirty minutes of reading changed
Giving patchley the handbook chapter as text, with the focus "customer complaint handling", produced the draft below. Reading the draft raised three questions the handbook does not answer: what happens when the complaint form is incomplete, how long the returned part is on its way, and who decides between credit note and replacement. Those became the hypotheses for the first interview.
04 — Helper tool
Interview guide generator
Three questions this tool answers: what should we ask this person, what should they bring, and which answer would change our hypothesis. The questions come from a written library, not from a model; the generator orders them for the process type, the role and the goal, and says what each answer becomes in the diagram. It runs in your browser and sends nothing anywhere.
Choose the type of process, the role of the person you will talk to and what the session is for. Add the process name if you want it in the questions.
05 — Best-practice outcome
A capture plan with interview guides
One page per process: the sessions, the people, the hypotheses, the conventions. Three checks tell you whether it is done.
Preparation is done when the first interview can start with "walk me through the last case" instead of "who else should we talk to".
06 — Common traps
Where preparation goes wrong
Three mistakes that are cheap to avoid in this step and expensive to repair in the next one.
Only managers get interviewed
Managers describe the process as designed, because that is the version they are accountable for. The workaround, the spreadsheet next to the system and the phone call that replaces a form live with the people who do the work. A plan without them captures the handbook a second time.
The boss sits in with the team
With the manager in the room, nobody mentions the step that is skipped on Fridays. Interview managers and their teams separately, and say so in the invitation; the differences between the two conversations are often the first finding.
"How does it work?" as the opening question
A general question gets the designed process. "What did you do with the last one?" gets the real one, with the system that was actually opened and the colleague who was actually called. The first interview questions are built around that difference.
07 — Ready to move on?
Two questions before step 6
If both answers are yes, the capture can start. If not, another day of preparation is cheaper than a repeated interview.
Are the sessions booked and is consent settled?
Every named person has a date, knows whether the session is recorded, and has agreed. Where a works council or a data-protection officer has a say in recording, that conversation has happened; as general orientation, it is easier before the first session than after it.
Have you read what exists?
The handbook, the old diagram, the policy: read, or turned into a draft and read as a diagram. The interviewer knows what the document claims and can hear where the people disagree with it.