Case studies show you what someone can actually do.

Case studies are almost the only real signal you get.

Ask someone how they handle a difficult client and you get a rehearsed paragraph. It will be fluent, it will be flattering, and it will tell you nothing about whether they can actually do it on a Tuesday. People are not lying. They are describing an idealised version of themselves, and after a few interviews they get good at it.

A case study removes the ability to describe. It puts a situation in front of someone and requires them to act inside it. You stop hearing an account of the work and start watching the work.

Don't tell me what you would do. Do the case study, because through that you see so many other things.

The correction that matters most

That correction is the one people slip on constantly, because it sounds like a distinction without a difference. It is not. "Tell me how you would handle a donor who is wavering" is a knowledge question, and knowledge is the smallest part of what you need to know. "I am that donor. Go." is a case study, and it also tells you whether they ask good questions, whether they can hold a room, whether they explain themselves clearly, and what they do when the ground moves.

History matters, but it matters less than it looks

A résumé is fast to read and worth reading. It tells you the size of rooms someone has been in. What it cannot tell you is whether any of it came to bear. Someone can spend fifteen years accumulating experience they never once converted into judgement, and the page looks identical either way.

So take the history quickly, from the page and the job titles, and spend the interview on what is going on today. The past is context. The case study is evidence.

Three dimensions, and you need all three.

A case study is not scored out of ten. It returns three separate readings, and a strong number on one of them does not compensate for a weak number on another.

• How far. How far along in their thinking do they actually get before the time runs out? Some people are still framing the problem when the hour ends.
• How fast. The same distance covered in half the time is a different person. Thoughtful but slow is genuinely worse than thoughtful and fast, and it is easy to forgive slowness because the individual sentences sound good.
• How deep. Pick one branch and keep descending. Everybody sounds competent at the top layer. The reading is the level at which they start to falter.

A case study is a pyramid of boxes.

Every box has sub-boxes, and those have sub-boxes. The structure is both very wide and very deep, and you can never cover all of it. So do not try. Cover enough of one layer to see how someone thinks, then pick a single branch and descend until they run out.

That descent is the part interviewers skip, and it is the part that catches the most expensive mistake in hiring: someone who represents depth across a whole layer and only has it at the top. They are not bluffing. They genuinely know a great deal about all of it. They have just never done one of them.

Tomorrow, fifty people are coming.

This is a case study for an operational role, and it is deliberately built so that a capable person cannot fail it for the wrong reason. There is no specialist knowledge in it, no trick, and nothing to know in advance. Anything that goes wrong is the candidate.

What the follow-up questions actually reveal

You answer honestly, and stay in character. No chairs, no desks. They bring their own laptops. There is one TV, in another room.

The library, from junior to very senior.

One case study, used unchanged across every level of seniority. That is not a shortcut, it is the point: when the brief is the same for everyone, the seniority shows up in the answer rather than in the question, and you get a directly comparable reading across candidates who are years apart.

How the senior conversation actually goes

Designing a case study.

There are two ways in, and which one you use depends on how much practice you have had.

The built case: design it once, then become expert in it

Write the case study out properly, with its options and its permutations mapped: what the answers can be, which follow-up each one opens, where the branches lead. Then use it repeatedly. After a dozen interviews you know the terrain so well that you stop spending attention on running it and can spend all of it on noticing.

• Choose something they cannot fail for the wrong reason. If there is any blocker in the case unrelated to the capability you are measuring, remove it. You want to know for certain that a good candidate can win here.
• Choose something universally known, so the setup costs one sentence.
• Deliver the context well. Your delivery is part of the instrument. If the framing is muddled you have measured your own briefing, not their thinking, and you will not be able to tell the difference afterwards.
• Map the permutations before you use it. A branch you have not thought about is one you will fumble live, and the candidate pays for it.

The borrowed case: take it off your own desk

Once you are comfortable, the best case studies are the ones you are living through. The mess you were dealing with at midnight last night, the client conversation that went sideways this morning. Play the client, hand them the situation, and watch.

It works because you have just been through every nuance yourself, so you can respond in character without preparation and you know precisely which moves were good ones. It is also free: no design time at all.

The rules you say out loud, before you start.

Set these up front, in plain words. They cost thirty seconds and they change what you get for the next hour, because most of what looks like weak performance in a case study is actually somebody guessing at rules nobody told them.

It's the not realising you don't know that gets you in trouble. Because if you realise you don't know, it's very easy to get the knowledge.

Why the pause rule is not a courtesy

Running it: four disciplines.

1. Do not answer for them. Hold the silence.

This is the hardest one and the most commonly broken. The pause gets uncomfortable, you fill it with a hint, and you have just destroyed the measurement. Worse, you will not notice you destroyed it: the interview will feel like it went well, because you and the candidate together produced a good answer. Silence is where you find out where they were going.

2. Notice. Deliberately, and in detail.

Notice what they say and what they do not say. Notice whether they are taking notes while you answer. Notice whether the first move is a question or an assertion. Notice the order they ask things in, which is usually a cleaner signal than the content of the questions. Record the interview, because you cannot both run a case study well and take a full record of it, and afterwards the detail is gone.

3. When they pause, teach. Then go back and re-test.

The pause rule means people will use it. When they do, give them the knowledge, genuinely: "you should ask me what we need, and then work out how you're going to get it." Then return to the case and watch whether they can use what you just handed them.

That second half is the point. The pause tells you they can spot a gap. The re-test tells you they can learn, which is the more valuable of the two and the one that almost never gets measured in an interview.

4. Move on the moment an area is safe.

Once you can see they are comfortable at a level, leave. "That's great, we're in the right direction here. I want to hop to something else. Assume he agreed with you on the major donors, and you've started the work. Now you're planning..." You are not there to have a nice conversation. Every minute spent confirming something you already know is a minute not spent finding an edge.

When to go deeper, and when to stop and start over.

Any case study can be stretched into three hours, and sometimes it should be. The choice you make roughly ten minutes in is between depth and breadth, and it is decided by how they are doing rather than by your schedule.

Going deeper on a strong candidate

Make it harder. Deliberately. If they say the answer is to recruit ambassadors, become the client and ask how. They explain. Ask what they would say to the ambassadors exactly. Keep pressing, and while you press, play a specific character: a client who is afraid to ask.

Then watch for one thing. Do they notice that your endless questions are the symptom, and that the root cause is a client who has not been made comfortable enough to say what he actually wants? A strong candidate stops answering the questions and addresses the person asking them.

Stopping on a weak one

End the case study and start a different one. Do not investigate why the first one went badly, because you do not have the time and the answer will not change what you do next. You are answering a single question: was that once, or is it every time?

Usually you will see the same shape appear across all of them, and that consistency is the finding. Occasionally the second case goes well, and you have just saved yourself from rejecting somebody over a briefing you gave badly.

What you are actually measuring.

Underneath the three readings sit two separate capabilities, and they substitute for each other. Knowing which one you are looking at changes the hire.

• First-principles problem solving. Rare. It shows as the quality and order of the questions, not the answers. Someone with it will engage well in an area they have never touched, and you can watch them do it because the questions they ask will still be sensible ones.
• Domain experience. Common enough to buy. It shows as knowing which two questions to ask first, and as not needing to invent anything.
• Learning speed. Measured only by the teach-then-re-test move. It is the multiplier on both of the others.

Someone can be an obviously strong problem solver and still be shallow here, and both facts can be true in the same interview. The honest read on a good generalist working outside their experience is usually: strong first-principles thinker, no depth in this specific area, so several of the questions could have been better ones. Write that down as two separate findings, not one blended score.

You are not only assessing. You are also selling.

Deciding, with the risks named out loud.

The reading you have at the end is almost never "yes" or "no". It is a shape, with strengths and specific holes in it, and the decision is what you are willing to carry.

You cannot get everything at once

Once you know what you want, you still have to prioritise, and sometimes the right hire is a stepping stone. Someone might take you one level up but not two. That is often still worth doing, and it is a different decision from settling: you are choosing which capability to buy now and which to buy later, deliberately, rather than holding out for a person who does not exist.

Name the risks, and say them to the candidate

Every hire is made with risk in it. The discipline is making the risks explicit to yourself before you decide, and then sharing them with the person, at least at senior level. It changes what the first six months are about, because both of you know where the thin ice is.

Ask for a rich handoff

When someone else runs the first interview and you take the second, do not accept a thumbs up. Ask for five to seven minutes of unstructured voice: what they did, what was good, what was not good, what the concerns are. A written summary compresses out exactly the hesitations you need, and it is the hesitations that predict.