The measurement problem is real and partly solvable

A prevented withdrawal leaves no record.

Support services cannot easily demonstrate their effect because the outcome is an absence. The student who stayed appears in the data as an ordinary continuing student, indistinguishable from one who was never in difficulty.

The partial solution is to record the flag rather than only the outcome. Where students identified as at risk are recorded, their subsequent continuation can be compared with a similar group who were not contacted, which is not a controlled comparison and is considerably better than nothing.

The simplest defensible figure is continuation among students who were contacted after a flag, tracked over a year. Where that figure is markedly higher than for flagged students who were not reached, the service has an argument it can put in a budget conversation.

It is also worth recording what the presenting issues actually were, in categories. An institution that discovers half its at-risk contacts concern transport or placement costs has learned something specific and actionable that no satisfaction survey would produce.

The figure that should not be used is service utilisation, which measures demand rather than effect and rises when things are going badly. Reporting it as a success measure creates an incentive that runs against the service's actual purpose.