An electrician building their own house did this:
"a switch fuse in my meter box with an 80A fuse in it to discriminate from the suppliers 100A fuse.
You won't get meaningful selectivity between 80A and 100A as they are both subject to something like +/-10% tolerance, etc, and the total let-through I2t for the 80A clearing is less than the pre-arcing I2t of the 100A.
Typically you need to have a 1.6:1 ration for BS88 and other European fuses (i.e. two sizes, so 63A & 100A or 50A & 80A, etc) for that.
But really selectivity is only needed if you expect to need it. If you are putting in a switch-fuse for the long tails it is largely down to the regulations that require both a means of isolation (the switch) and OCPD that is matching the sub-main characteristics (i.e. 5s disconnect for fault in or at the end of the cable).
The regulations do allow the DNO fuse
provided they agree to it being used. Realistically it is easier to fit a switched-fuse!
Where you need selectivity (and so fusing down a bit from the DNO) is when you have several circuits coming off the DNO fuse. And by that I mean several sub-mains, for example at a Ryefield board or where tails have been split to allow two CU or some massive load to be added and you reasonably expect a fault on one sub-main not to take out the lot.
For the typical case of long meter tails / sub-main you really ought to have it designed so that it won't be overloading the DNO supply anyway (diversity allowing) so having your own choice of fuse is down to it independently providing protection to your choice of cable and supply Ze, etc.
In that case you might be putting in 80A, even though on a hard fault it probably won't save the DNO fuse, because it then meets your 5s requirement and is enough for the planned loads.