Recovering After a Bad Fermi Assumption
How to recover after a bad Fermi assumption by acknowledging it, revising it, propagating the change, and explaining sensitivity.
Candidates who struggle when an interviewer challenges an assumption.
A challenged assumption is normal
Fermi interviews often test whether you can revise assumptions. A correction is not automatically a failure; it can show model control.
Revise one driver
Change the challenged assumption, then propagate that change through the estimate. Avoid rebuilding everything unless the structure was wrong.
Concrete recovery
If usage frequency should be weekly rather than daily, divide that driver by about seven and update the final estimate accordingly.
Explain sensitivity
Say how much the final estimate moved and whether that assumption is the dominant driver. This shows control of the model.
Common mistakes
Candidates often defend a weak assumption too long. It is stronger to revise cleanly when better information appears and propagate the change.
Practice the pattern
Use the LeetQuidity curriculum and calibration to turn this topic into a focused practice plan.