How to answer a question you do not know the answer to

You have prepared for a week. Somebody asks the one thing you did not prepare, and there is a silence in which you can hear the room.

What happens in the next four seconds does more to your credibility than the preceding twenty minutes. That is worth preparing, and almost nobody does — preparation goes into the material, and the material is not where this goes wrong.

First: which kind of question is it?

They look alike and they are not. Getting this wrong is what produces the answer that is technically correct and lands badly.

Type one: you lack the fact

The easiest, and still routinely mishandled. The temptation is to produce an approximate number so as not to look empty-handed. Do not. An estimate given under pressure is remembered as a commitment, and if it is wrong you will spend more credibility correcting it than you saved.

The complete answer has three parts and takes six seconds: I do not have that with me. I will send it by Thursday. Roughly, I would expect it to be in the low teens, but treat that as unverified.

Note the order. The commitment comes before the rough figure, and the rough figure is explicitly labelled. That sequence is the difference between sounding rigorous and sounding evasive.

Type two: you have not considered it

Harder, and the one where people damage themselves most, because admitting you have not thought about something feels like admitting you did not do the work.

But there is a move available. You may not have the answer, and you almost always have the approach — so give that, out loud:

This is not stalling. It demonstrates the thing the questioner actually wants to know — that you can reason about the problem — while being honest that you have not yet. In my experience it recovers more ground than a prepared answer to a different question ever does.

Type three: it is not yours

Short, and the failure mode is guessing on someone else's behalf. Name the owner, offer to get the answer, and stop. "That sits with legal. I have not spoken to them about this specific case — I will, and I will copy you."

Do not add "but I think they will be fine with it." You have just made a promise on behalf of a department that did not authorise you to make it.

Type four: it is not a question

The one that costs the most. "Have you actually done this before?" is not a request for your CV. It is an objection wearing a question mark: I am not sure you can pull this off.

Answer the literal question and you lose, even if the answer is good, because the objection stays untouched and the room heard you dodge it. Answer the objection and the question resolves itself:

"Fair concern. Not at this scale. Here is what I am doing about it — X is running the part I have not done before, and the first checkpoint is in six weeks, early enough to stop."

Naming the concern out loud is the load-bearing part. An unspoken objection stays in the room and gets discussed after you leave, where you cannot answer it.

The four seconds

Whichever type it is, the mechanics of the pause matter as much as the content.

• The silence is shorter than it feels. People systematically overestimate how visible their internal state is — the same illusion of transparency that makes nerves feel obvious. Two seconds of thinking reads as considered. It does not read as a failure.
• Do not fill it with hedges. Research on courtroom testimony found that a speech style dense with hedges, hesitations and qualifiers reduced perceived credibility independently of what was said. "I mean, obviously, it sort of depends" costs you more than the pause would have.
• Say the words plainly. "I do not know" spoken evenly is a strong sentence. Said apologetically, it becomes an admission of a fault. Same words.
• Write it down in front of them. Visible, and it makes the commitment real. Then actually send it, on the day you said.

How to prepare for it

Write the six questions you hope nobody asks. Not the likely ones — the ones that make you uncomfortable. Then answer each aloud in three sentences. There is a free tool for exactly this, which sorts each question into the four types above and gives you the move that works for each.

You will find that two of them have good answers you had simply never assembled, two need a date rather than an answer, and one or two are the objection type — which means they are the real risk in the meeting, and you can now decide whether to raise them yourself before somebody else does.

Raising it yourself is almost always better. It is the same information, arriving as thoroughness rather than as something you were caught not mentioning.