From what you said, I see some issues might come from:
1. Your structure wasn't organized enough to allow each variable bucket (with an assumption behind it) to be constantly changed as the interviewer questions your assumptions
2. Your main variable was not mathematically broken down in the right way (at one or more branches of your issue tree)
3. You did not have enough depth when breaking down your main variable (not enough levels in your issue tree), so the interviewer expected more detail in your assumptions
4. Or, your communication of the structure was not clear enough and, even if you did alright on paper, you were perceived as unstructured to the interviewer
Remember: market sizing / estimation questions are not about getting the right number. It's about having the right process, reasonable assumptions, sharp math and assuring your results make sense.
Please, feel free to reach out if you have any other questions