In a recent ASA ruling, BT were taken to task about their estimates.
As part of a statement, Openreach indicated "the speed estimate ranged from the 80th to 20th percentile for similar phone lines, therefore, 80% of end-users could achieve the quoted speeds."
The basic estimate has always included some allowance for crosstalk; the belief at the time an "impacted" estimate was introduced was that this was for semi-faulty lines, rather than crosstalk-impacted.
Elsewhere, I have seen that the speed estimates are identical for all lines fed from one DP (eg one pole). And elsewhere still, I've seen that estimates get updated when people actually get the service - which seems to be the case for my property.
Perhaps someone else on the DP took FTTC with a different result; perhaps this (or something else) caused BT to have a better idea of which "similar" lines made the best comparison for yours.