Sales needs a delivery date for something you have not finished discovery on. What do you commit to?

The date is standing in for a decision

Nobody wants a date. They want to make a decision that depends on one, and until you know which decision you cannot tell what kind of answer would help. The three common cases need entirely different responses. A salesperson closing a deal needs something they can put in front of a customer without being wrong in writing. A marketing lead planning a launch needs a lead time for work that has its own commitments attached, and mostly needs to know the earliest date they should start spending. An executive tracking whether a bet is on course needs confidence, not precision, and is usually satisfied by a quarter.

So the first move is a question, and it should be a specific one rather than a stall. "What are you going to do with the date, and when do you have to do it?" tells you whether you are being asked for a commitment, a forecast, or reassurance, and those have different accuracy requirements. A candidate who starts negotiating a number without asking this has skipped the only step that makes the rest tractable.

The second thing to establish is the cost of being wrong in each direction. If a missed date means a customer walks and a contract is void, you are in commitment territory and should commit to very little. If a missed date means a blog post is published a fortnight late, a forecast with a stated confidence is plenty. Treating those two identically — usually by refusing both, on principle — is how PMs acquire a reputation for being an obstacle.

Commit to what you control

You cannot honestly date scope you have not defined, but there are always three things you can commit to, and offering them is the difference between a refusal and an answer.

The first is a timeboxed end to discovery. Not "we'll know when we know", but "discovery closes on the 12th and you will have a scoped estimate on the 14th". That is entirely within your control, it gives sales something to say, and it converts an open question into a two-week wait.

The second is a date for the date, which is the same idea generalised, and it is worth naming explicitly because it is the answer most people fail to think of. You are committing to the moment the uncertainty resolves rather than to the outcome.

The third is a narrow slice you already understand. Very often the deal does not need the full capability, it needs the part the customer will use in the first quarter, and that part is frequently well enough understood to date. Splitting the ask is often the only genuine way to satisfy both sides, and it also protects you: a small committed slice with a large uncommitted remainder is a much safer contractual position than a vague whole.

What you actually say

The wording matters more here than in almost any other execution scenario, because the answer will be forwarded to someone who was not in the conversation. Two versions of the same content land very differently.

Weak	Stronger
"I can't give you a date yet."	"I can give you a date on the 14th. Until then, here is what I can commit to."
"Probably end of Q3."	"The part they need for pilot, by 30 September. The rest is unscoped and I will not date it yet."
"It's about six weeks of work."	"Six to eleven weeks once scope is fixed, and scope is not fixed. Treat eleven as the planning number."
"Yes, we can do that."	"Yes to the reporting export. Not the scheduled delivery, which I have not sized."

Sample phrasing to a customer-facing colleague, which you should be able to produce almost verbatim in an interview: "Discovery closes on the 12th. On the 14th I will give you a scoped date for the pilot capability with a confidence level, and a separate view on the rest. Before then, the only thing I would put in writing is that we are actively building in this area. If you need something in writing this week, tell me what the minimum is and I will tell you whether I can stand behind it."

Ranges, and the honest asymmetry

If you must give a number before scope is settled, give a range and say which end to plan on. Ranges get compressed by whoever receives them — the low end is the only number that survives the retelling — so name the planning number explicitly rather than trusting the range to carry the message. Stating that the uncertainty is asymmetric also helps: unscoped work almost never comes in early, because the discoveries that remain are discoveries of additional work, so the range should be skewed upwards rather than centred on your best guess.

Where this scenario is actually lost

The failure mode is not giving a bad date. It is giving a date informally, in a chat message or a corridor conversation, and then discovering that it has become a commitment in a proposal you never saw. Anything you say about timing to a customer-facing colleague should be assumed to be quoted externally, so it needs to be written down in a form you would be content to have forwarded, with the conditions attached in the same message rather than in the following one.

The related failure is quieter and worse: agreeing to a date you privately do not believe because refusing was uncomfortable, and then spending the following weeks managing the consequences of a commitment you knew was wrong when you made it. An interviewer is listening for whether you will hold a position under mild social pressure, which is why the follow-up about the CEO asking publicly is not a courtesy question.

Commit to the resolution of the uncertainty and to the slice you understand; anything you say about timing will be quoted, so say it in a form you would be happy to see forwarded.