Estimates are not perfect and cannot be made perfect. Depending on who you believe, more time spent on a estimate will either not improve the estimate, or reduce the accuracy of the estimate. The point of estimating is to provide data to the Product Owner, but everyone should understand that the data is just that, an estimate.

That does not mean we should not look to improve our estimates. The best starting point for analysis is to compare the estimate against whatever the final reality was. With stories we have the estimate. To get the “actual” we simply ask “now that we have done the work, what should our estimate have been”. Did it take longer / shorter than expected?“ If yes, was it more / less risky or complex? What is the number.

Beware of blaming problems on estimation. Sometimes the thinking goes “our plan did not work so we need to improve our estimates”. Our Estimates are Terrible and Why Doesn't Traditional Project Management Work For Software Projects? will provide additional background here.