How to Interview Your Best Employee Before They Leave (and Keep What They Know)

Your best operations person just handed in notice. Fourteen years in, and there's a version of your supply chain that exists only in her head: which vendor to call when the usual one is late, how to spot a problem shipment before it's officially flagged, which client actually means it when they say "urgent" versus which one always says that. None of it is written down, because nobody ever needed it written down while she was still there to ask.

You have four weeks before she's gone. What most companies do with that time is a standard exit interview on the last day, covering why she's leaving and how she felt about the job. What almost nobody does, and what actually matters here, is a separate conversation designed specifically to get the operational knowledge out of her head and into something the rest of the team can use.

Why the standard exit interview doesn't do this job

Exit interviews exist to understand attrition: why people leave, what could have kept them, whether there's a pattern HR should worry about. That's a legitimate and useful thing to know. It is not the same conversation as "tell me everything you know that isn't written down anywhere," and trying to do both at once weakens both. Someone processing their own decision to leave is not in the right headspace to methodically walk through fourteen years of operational judgment, and someone trying to extract feedback about company culture isn't going to also get a clean download of vendor escalation logic.

Run these as two separate conversations, ideally with different people leading them if you can manage it. HR or the manager runs the retention-focused exit interview. Someone closer to the actual work, a peer, a successor, or whoever is inheriting the role, runs the knowledge capture conversation.

Timing matters more than most companies realize

The instinct is to schedule this for the last week, when the departure is fully processed and there's finally "time" for it. That's usually too late. By the final days, most people are mentally checked out, going through the motions, and the quality of what you get in a knowledge-capture interview drops off sharply. The window that actually works starts as soon as the resignation is confirmed and runs through the first few weeks after they've physically left, if you can still reach them for a follow-up question or two. Outside that window, in either direction, you get much less.

Practically, that means: don't wait for a "knowledge transfer plan" to get formally organized. Book the first capture conversation within the first week of notice, while the person is still fully engaged and has every reason to want to leave things in good shape (nobody wants to be remembered as the person who left a mess).

What to actually ask

Generic prompts like "tell me what you know" produce almost nothing useful, because tacit knowledge, the stuff that lives in someone's judgment rather than in a manual, doesn't surface on command. It surfaces when you ask about specific situations. A few approaches that work better than an open-ended request:

Ask about exceptions, not the standard process. The standard process is usually already documented somewhere, however imperfectly. What's missing is the "but if X happens, do Y instead" knowledge. Ask directly: "what are the three situations where you'd break from the normal process, and how do you know when to do that?"

Ask them to walk through a recent hard call. "Tell me about the last genuinely difficult decision you made in this role, what made it hard, and how you actually decided." Real, recent, specific cases surface far more useful detail than abstract descriptions of "how things generally work."

Ask who they'd call and why. Most experienced people have an informal network, the vendor rep who actually picks up, the client contact who tells the truth, the colleague in another department who knows the workaround. This list rarely exists on paper and disappears entirely when they leave. Ask for it explicitly, by name.

Ask what they'd tell their replacement in the first week, if they could only say five things. This forces prioritization. People will often surface the genuinely critical judgment calls here that never came up in a more general conversation, because they're being forced to rank what actually matters.

Ask what mistake they see new people make, over and over. This surfaces the gap between what the SOP says and what actually trips people up in practice, which is often the single most valuable thing you can capture.

Handling the awkwardness honestly

There's a real, understandable reluctance to schedule an extra conversation with someone who's already told you they're leaving. It can feel like you're extracting value from someone on their way out, or that it sends a message the company only cares about what it can get before they go. Naming that tension directly, rather than pretending it isn't there, usually resolves it. Frame the ask honestly: "we'd genuinely value your perspective on what you've learned here, both for whoever takes this on next and because we respect what you've built. This is separate from anything about your decision to leave, no pressure either way." Most people respond well to being asked, sincerely, to leave a mark rather than just walk out the door, and the ones who don't want to engage will tell you, which is fine too.

It also helps to be upfront that this isn't a test of loyalty or an attempt to guilt them into staying longer or working harder in their final weeks. It's specifically about capturing judgment for the people staying behind, and framing it that way, rather than as an extension of their job duties, tends to get a more open, generous conversation than treating it as one more task on their offboarding checklist.

Capture it in a form someone can actually use later

A transcript of a ninety-minute conversation is better than nothing, but it's a poor deliverable on its own, because nobody is going to read ninety minutes of loosely structured conversation six months from now when they hit the exact situation it covers. Two things make the capture actually useful afterward. First, record it (video if you can, audio at minimum) rather than relying on notes, because tone and hesitation often carry as much signal as the words, someone pausing before an answer is often marking "this is the part that's actually judgment, not rule." Second, and more important, someone needs to go through it afterward and turn the useful parts into short, findable, specific answers, organized by the situation they apply to, not left as an undifferentiated transcript.

That second step is the one almost every company skips, because it's genuinely tedious to do by hand, and it's exactly the step that determines whether the knowledge capture was worth the four weeks or just a nice gesture on the way out.

This applies beyond resignations too

Everything here applies just as directly to planned retirements, internal transfers, and long leaves, not only resignations. In many ways a retirement is the easier case, because there's usually more advance notice and less emotional complexity than a resignation, which means there's less excuse for waiting until the final week to start. An internal transfer is worth the same treatment even though the person isn't leaving the company: if they're moving from a role where they'd built up years of specific judgment into a different function entirely, that judgment leaves the role just as thoroughly as if they'd resigned, and the same capture conversation applies, just without the time pressure of an external departure.

Working alongside a successor beats a static document

If timing allows even a few days of overlap between the departing person and whoever inherits the role, or the person who'll be fielding those questions going forward, that overlap is worth more than any interview. Watching someone handle three real situations in real time surfaces judgment calls that never come up in a structured conversation, because the departing person often doesn't consciously know they're making a judgment call until they're in the middle of making it. If overlap isn't possible, a well-run capture interview is the next best thing, but don't skip the overlap if you can get even two or three days of it.

Where this becomes something the rest of the team can actually use

The hardest part of this whole exercise usually isn't getting someone to talk, most people want to leave things in decent shape. It's turning what they said into something searchable, specific, and actually findable by the next person who hits that exact situation six months later, rather than a video file nobody remembers exists.

This is the exact gap Decisionlore's boss-interview workshop was built to close, extended to any departing expert, not just the founder. A structured, facilitated conversation extracts the exceptions, the judgment calls, and the informal network, and turns it into cited, queryable answers the rest of the team can ask against immediately, rather than a recording sitting in a folder. If you've got someone walking out the door with fourteen years of undocumented judgment, it's worth doing this properly rather than relying on a last-day exit interview to catch it. See how the workshop fits into your plan on the pricing page, or get in touch before the notice period runs out.