"I Don't Know Yet"
Say you do not know, say what you will do to find out and by when, and give the best current guess with its confidence — so the room keeps trusting you instead of trusting whoever guessed loudest.
The ask, the obvious response, and how it goes wrong
Every lesson starts where the work starts: someone asked for something, and the first response that comes to mind has a problem.
When you are asked something you cannot answer yet, what do you say so that people trust you more afterwards, not less?
Finance, in the quarterly planning meeting, looking at you: "If we switch to the new tax provider, how much will it cost us per order?"
Give a number. Planning needs numbers, and "I don't know" in front of finance sounds like you have not done your homework. Estimate something reasonable — "a few cents per order" — and refine it later.
The "few cents" goes into finance's spreadsheet as a fact. When the real figure turns out to depend on which regions we ship to and is several times higher for cross-border orders, the plan is wrong and the person who is remembered for it is you.
- The "few cents" goes into finance's spreadsheet as a fact. When the real figure turns out to depend on which regions we ship to and is several times higher for cross-border orders, the plan is wrong and the person who is remembered for it is you.
- Everyone in the room who knew you were guessing now discounts everything else you said in the meeting.
- The opposite failure is just as common: "I don't know" and nothing else. The question is left hanging, finance fills the gap with their own guess, and the decision is made without engineering.
- Neither bluffing nor stopping at "I don't know" commits you to anything. The answer that builds trust is the one that commits you to finding out.
What is really going on
- A question in a meeting is usually a request for a commitment, not a fact. Finance does not need the number this second; they need to know when they will have it and how much to trust it. You can answer that even when you cannot answer the number.
- Bluffing works exactly once per person. The first time a confident number turns out invented, every future number you give carries a hidden discount. Saying "I don't know yet" is how you keep your numbers at face value.
- The word "yet" does the work. "I don't know" is a dead end; "I don't know yet, I will by Wednesday" is a plan with a date, and a plan is what the room needed.
- Most unknowns are partly known. You know the per-call price; you do not know how many calls an order makes for cross-border shipments. Separating what you know from what you do not is often more useful than either alone.
Why the bluff costs more than the gap
A confident guess feels like competence while you are saying it. Its cost comes later, and it is not the wrong number — numbers are wrong all the time. The cost is that everyone who later learns it was a guess retroactively re-prices every other thing you said.
The gap costs much less than it feels like, because the person asking rarely needs the answer this second. What they need is to know that someone is getting it, when it will arrive, and how much to trust what they have in the meantime.
"A few cents per order, roughly. It is a per-call API, so it should be cheap."
"I don't know yet. It is priced per tax calculation, and I do not know how many calculations a cross-border order makes. I will run last month's orders through their sandbox and have a per-order figure in the planning doc by Wednesday. If you need a placeholder, it is somewhere between a few cents and a few tens of cents, and cross-border volume decides which."
The bluff is quoted in the plan as fact. The second answer is also quoted — as a range with a date — and when Wednesday's number arrives, it replaces the placeholder instead of contradicting a promise.
The four parts
Know, do not know, how I will find out, when. It sounds mechanical, and after a few uses it stops being noticeable to anyone but you. What people notice is that your uncertainty always comes with a date.
Checkout errors spiked ten minutes ago. The head of product joins the incident channel: "Is this affecting all orders? Do I need to tell the CEO?"
"Looking into it, will update soon."
"Not sure yet. What we know: card payments are failing for some customers since 14:05; PayPal orders are going through. What we do not know is whether it is all card orders or one card network. I will have that in ten minutes from the provider's status page and our logs, and I will post here at 14:30 either way. If you need to say something now: 'Some card payments are failing, alternative payment methods work, next update 14:30.'"
Ranges, and what drives them
Sometimes the room genuinely needs something now. Then a range is the honest answer — but a range alone invites the listener to pick the end they like. Always say what moves the answer within the range.
"A few cents to a few tens of cents, depending on cross-border volume" is more useful than a tighter range without a driver, because finance can look at their own cross-border share and place themselves in it. It also tells them which question to ask next.
- Give the range in the unit they will use: per order, per month, per customer — not per API call.
- Name the one factor that most moves it.
- Say whether the decision changes across the range. If it does not, nobody needs to wait for the real number.
Coming back is the part that keeps the room
The trust is not built when you say "I don't know yet". It is built on Wednesday, when the number appears where you said it would. Every kept follow-up makes the next "I don't know yet" cheaper, until people start taking it as a sign you are being careful rather than a sign you are behind.
If Wednesday arrives and you do not have it, post before the deadline, not after: "The sandbox is rate-limited; I have half the orders done. Domestic is at the bottom of my range. Full answer Friday." That is still a kept promise, because you managed the date instead of missing it.
How to do it
Most important first.
- Use four parts, in this order: what you know, what you do not, what you will do to find out, and when you will come back. "The base price is per tax calculation. What I do not know is how many calculations a cross-border order makes. I will run last month's orders through their sandbox and have a number by Wednesday."
- If the room needs something now, give a range with a confidence and the thing that drives it: "My guess is between a few cents and a few tens of cents per order, and the answer depends almost entirely on cross-border volume."
- Name the date and the channel, then keep it. "Wednesday, in the planning doc" is a promise; missing it costs more than not having made it.
- When you come back, lead with the answer and say how it compares to the guess: "It is higher than my range for cross-border — here is why."
- If you find out you cannot find out by the date, say so before the date, with a new one.
How to explain the decision
The sentences, the order, and what to lead with — for someone who did not make the call.
- In the moment: "I don't know yet, and I would rather not guess a number that ends up in the plan. Here is what I do know…"
- The commitment: "I will run last month's orders through their sandbox tomorrow and post the per-order cost in the planning doc by Wednesday."
- If pushed for a figure now: "If you need something in the spreadsheet today, put in a range — a few cents to a few tens of cents — and mark it as pending. Cross-border volume is what decides where in that range we land."
- When you come back: "The number is in the doc. Domestic orders are at the bottom of my range; cross-border is above it, because each order needs a calculation per shipment leg."
- "Just give me a ballpark." Here is one: a few cents to a few tens of cents per order, and it depends on cross-border volume. Please mark it as a range in the plan — I will replace it with a real number on Wednesday.
- "You are the engineer, you should know this." I know how it works; the cost depends on how our own orders are shaped, which I have not measured yet. Measuring it takes a day.
- "We do not have time to wait until Wednesday." Then let us decide with the range and note which way the answer would change the decision. If it would not change it, we do not need to wait.
What can go wrong
- "I don't know" as a habit: the phrase used so often, without the follow-up, that it starts to mean "not my problem".
- The promise not kept. "I will find out by Wednesday" and nothing on Wednesday is worse than the bluff, because now you are both uncertain and unreliable.
- The range so wide it means nothing: "between zero and a lot". A range is only useful if you say what drives it.
- Saying "I don't know" to a question you could answer in thirty seconds by checking — which reads as not having prepared.
- "Saying I don't know shows weakness." Guessing and being wrong shows weakness. Saying what you will do to find out shows you know where the edge of your knowledge is.
- "I should never give a number I am not sure of." You should give ranges often — the discipline is in labelling them and saying what would move them.
- "This is about meetings." It is equally about written answers: a Slack reply or a PR comment deserves the same four parts when you do not know (Writing for Stakeholders).
Knowing whether it worked
- Finance comes back to you directly for the number instead of asking your manager whether engineering "has an answer yet".
- Your ranges start getting used as-is in plans, marked "per engineering", because the last ones held.
- You can look back over a quarter and find every "I will find out by X" followed by an answer on or before X.
- Other engineers on the team start saying "I don't know yet" in meetings too — a sign it is safe.
- On a small team, not knowing is normal and everyone can see why. In a larger organisation, the people asking are further from the work and more likely to treat any number as a commitment; the four-part answer matters more.
- At 10x the number of stakeholders, the same question gets asked by several people. Posting the answer once, in a known place, with the date you promised, saves you from answering it five times with five slightly different numbers.
- The sentences do not change with scale. What changes is how many people will quote your number, which is why you should be more careful about giving one.
- It feels weaker in the moment. In some rooms, the confident guesser wins the meeting, and you will lose a few of those.
- It creates a follow-up obligation. Every "I don't know yet" is a task with a deadline you set yourself in public.
- It can slow a decision. If the plan genuinely needs a number today, a range with a confidence may not be enough, and someone will make the call without it.
Where this applies
Product advice is context-sensitive. These labels say what each claim is specific to, and where a different stage, team or product would differ.
- GENERALAnywhere someone asks for a number or an answer you have not established. It differs in an active incident, where "I don't know yet" needs a much shorter "by when" — minutes, not days.
- TEAM-SPECIFICIn teams where saying "I don't know" has been punished, the first few times cost more and need a manager's cover; in teams where it is normal, the risk flips to overusing it without the follow-up.
Where the depth lives
This domain teaches the product-side judgement and hands the mechanism off.