It is easy to confuse matched inventory with eligible inventory. Forecasting can show plenty of matched requests while a strict frequency cap, a narrow audience rule, or a missing creative size removes most of them before the line item can serve. The campaign then paces behind even though supply looks fine.
A frequency cap is the classic example: a one-impression-per-user-per-week cap can filter the large majority of matched requests, and no amount of extra priority brings those users back. The delivery log usually makes this visible, showing a high matched count but a much lower eligible count with the difference attributed to the cap.
To fix it, separate matched from eligible in the delivery data, confirm the contracted frequency and creative requirements, then correct the cap period or creative association and reforecast the delivery that is actually reachable in the remaining flight.