Practice
Your estimate for a feature turns out to be badly wrong halfway through. What do you do?
Read first: a sample answer
Fictional sample answer written by JoRoute to show the shape of a strong answer. Do not copy it: replace every detail with what is true for you.
First I would tell my lead the same day, because a late warning is worse than a bad estimate. I would explain why the estimate was off, for example an API that turned out to need a migration. Then I would offer options with numbers: cut the export feature and ship on time, or keep it and add five days. The product owner decides, because it is their trade-off. I would also note the cause so our next estimation and planning includes a spike for unknown APIs. In my last role in Denver I flagged a late estimate like this on day 6 of 15, and we cut one report and still shipped on the agreed Friday.