A sponsor wants a date for a programme with several unknowns. Refusing to estimate is not an option. How do you handle it?

Give a range with the reasons, not a point

A single date implies precision you do not have and will be treated as a commitment. A range — with the specific unknowns that would move it to each end — is both honest and more useful, because it tells the sponsor what to worry about.

"Nine to eighteen months. Nine if the legacy data is as clean as the sample suggests and the partner can test in Q3. Eighteen if either is wrong."

Name what would collapse the range, and go and find out

Each unknown has a cheap probe:

• Data quality — profile the full volume rather than a sample. Days.
• Partner readiness — ask them, in writing, now.
• Unknown integrations — query the logs.

A two-week discovery that halves the range is worth more to the sponsor than two weeks of building, and framing it that way is usually what buys the time.

Commit to the near term, forecast the rest

Commit firmly to what is knowable — the first phase, where the unknowns do not bite. Forecast beyond it with the range, and re-forecast at each phase boundary with what has been learned.

That is a different promise from a single date and it is one you can keep.

Record the assumptions as assumptions

Every estimate rests on them. Written into the RAID log with a validation date and an owner, they become trackable; left in your head, they become the reason the date slipped and nobody saw it coming.

What not to do

Do not pad silently. A hidden buffer gets consumed and then you have no buffer and no credibility. State the range and the reasoning, and let the sponsor decide how much risk to carry.