First decide whether the missing value is actually necessary for the question you are solving.

If you can answer without it, state the limitation and continue. If it is essential, ask whether additional information is available.

Only estimate when an approximation is appropriate and make the assumption explicit.

Good exhibit analysis includes knowing what the data does not support. Missing information should reduce confidence, not force fake precision. For more on extracting useful conclusions from incomplete exhibits, see the guide to understanding data in case interviews.

I used to automatically fill gaps. Sometimes the missing point isn't even needed.