Try to base assumptions on simple reference points instead of isolated guesses. Population ranges, household size, usage frequency, and comparable categories are all examples of anchors that can make your logic feel much more credible.

It also helps to state assumptions in a chain. If the interviewer can follow how one assumption leads to the next, the estimate feels structured rather than random.

Remember that the exact number matters less than whether your estimate is reasonable and easy to audit.

I started using more reference points from daily life and it made a huge difference.

Same. Once you explain the logic behind the assumption, it sounds way less random.