Call out the sensitivity. A market-sizing estimate can still be useful even when one assumption has a large effect, as long as you recognize which assumption is driving the range.

The interviewer is testing judgment as much as arithmetic. If one uncertain variable dominates the result, pretending the final number is precise can be less credible than showing how the estimate changes under reasonable scenarios.

If adoption could plausibly be 10% or 20%, and that doubles the market estimate, you can state a base case and then explain that penetration is the key sensitivity.

Identify the two or three assumptions that drive the result most. Give a reasonable base case, then mention how the answer would move if the most uncertain assumption changes materially.

If one assumption doubles the answer, I would absolutely call that out.