This Counter64 counts responses received by the managed firewall from the specific URL filtering server represented by this row that arrived after the original request had already been pulled from the pending queue, counted from the last reboot of the managed device. An admin can use this per-server figure to identify exactly which configured filtering server is the source of wasted late replies. A row with a high late-response count relative to its total responses received would suggest that particular server needs a longer timeout or is simply too slow.