Compare prediction with outcome
Open Reports → Underwriting Accuracy and select a consistent period. Review where predicted valuation, repairs, offer, or other assumptions can be compared with a later verified result.
A small sample or missing actuals can make percentages misleading. Open the underlying property before deciding that the policy or model was wrong.
Improve the process
Separate data-entry errors from market changes and genuine policy misses. Correct the source record when it is wrong; update the underwriting policy only when repeated evidence supports the change. Record the reason for material policy adjustments so future comparisons remain understandable.