How to Run a Process Interview That Gets Past the Official Version

The process interview questions that work are the ones about a specific recent instance rather than the process in general, because asking someone how a process works gets you the version they would explain to a new starter, and asking what happened last Tuesday gets you the version that actually ran.

Everything else in this piece is a variation on that one move.

A process interview is a twenty minute conversation with someone who does a job, aimed at finding out how the work really moves. It is not an audit. It is not a performance conversation. And it is the only reliable way to learn things that were never written down, which in most companies is the majority of what matters.

Why the general question fails

Ask "how does onboarding work here" and you will get a clean answer. Four or five steps, in order, delivered confidently.

That answer is not a lie. It is a summary, and summaries are built by averaging. The person has run the process forty times, and the version in their head is the median one with the awkward bits sanded off. They have stopped noticing the spreadsheet. They no longer register the two days of waiting because waiting does not feel like work. The workaround they invented in 2024 has been absorbed into their sense of what the process simply is.

So you get an accurate description of intent and a poor description of reality, and you have no way of telling which parts are which.

Start with the last real one

"Walk me through the last time you did this. Not in general. The specific one."

This single reframe does most of the work. Memory of a particular Tuesday is concrete in a way that memory of a process is not. People remember the client who was difficult. They remember opening the wrong tab. They remember that the approval took until Thursday because someone was off.

If they cannot recall a recent one, have them open it. The actual emails, the actual record, the actual folder. Reading their own artefacts back to themselves recovers detail that no amount of questioning will.

You will lose some generality this way. One instance is one instance, and the last one may have been unusual. That is a real cost and it is worth paying, because a specific true story beats a general false one, and you can always ask whether that run was typical afterwards.

Follow the hesitation

The most valuable sentence in any process interview begins "and then I just".

"And then I just copy it into the sheet." "And then I just message Dan." "And then I just wait until it clears." Every one of those is a step that would never appear in a document, described by someone who has stopped considering it a step at all.

When you hear it, stop. Ask what that involves. Ask how long it takes. Ask what happens if it does not happen, and who else knows how to do it.

Hesitation works the same way. When somebody pauses, or backtracks, or says "actually, before that", the thing they are recovering is usually the part that makes the process work. People do not stumble over the parts that are simple.

This is the part of the work abi. FDE runs as interviews, by voice or by text, following the workaround instead of moving on to the next question. They run in the browser on a private link, with no account and no install, and most people finish in under twenty minutes. The map itself is free on any account. The interviews are the part that costs anything to run.

The four questions worth memorising

Once the story is out, four questions fill most of the remaining gaps.

What did you have to wait for? Waits are absent from almost every process document ever written, because nobody thinks of them as steps, and they are usually where the elapsed time actually goes.

What happens when it goes wrong? The exception path is not a footnote. In plenty of processes it is a third of the volume and most of the effort, and it is undocumented precisely because it does not feel like the process.

Who do you ask when you are not sure? The answer names your real approval structure, which is often not the one on the org chart.

Is there a faster way you use when you are busy? Ask this warmly and mean it. The answer is the workaround, and the workaround is what runs on a normal week. The official route is what happens when someone is watching.

What not to do

Do not interview people in groups. The most senior person in the room sets the version of events, and everyone else adjusts toward it. You lose the disagreements, and the disagreements are the most useful output you were going to get.

Do not correct anybody. If what they describe is not what you believed, write down what they said. You are collecting accounts, not adjudicating between them. Two people describing the same handoff differently is a finding, not a mistake to be resolved in the room.

Do not start with the org chart. Start with a process and follow it wherever it goes, including out of the department you thought owned it. Work does not respect reporting lines and the interview should not either.

And do not take notes in a way that looks like scoring. A visible tally next to someone's name changes what they tell you within about four minutes.

The part that costs you something

Interviews have a genuine weakness and it is worth naming rather than talking around.

People are unreliable narrators of their own time. They round up on the tasks they resent, forget the ones they enjoy, and reconstruct the order of events to be tidier than it was. What you get is directionally right, not precise. If you need precision to the hour, you need instrumentation, and if the work genuinely lives inside one system that logs everything, read the logs. They will beat any interview.

The second cost is that this only works if it is safe. If people suspect the exercise is a headcount review with a friendly name, you get the official version of everything, which is worse than asking nothing, because now you have a document that looks like evidence. Say what it is for before you start, and then behave in a way that matches for the following six months.

Where to start

Pick the process that would hurt most if the person who runs it left. Interview them, then interview the person who receives their output, then the person who hands them their input.

Three conversations, an hour in total, and you will have found something you did not know. Everybody does. That is not a promise about the method, it is a statement about how much of a business is undocumented.