Ask a grants team where a round lost time and they will usually say assessment, because assessment is the part that feels slow. The calendar usually says otherwise. The two reliable sinks are the fortnight before the round opens and the weeks between the panel meeting and the letters going out.
Before opening
A new round is rarely new. It is last year’s round with four changes, and the four changes take a fortnight because the form is rebuilt from a document rather than copied from a configuration. Every rebuild reintroduces the same three mistakes, which are then found by applicants.
Copying a round (form, rules, criteria, templates, timetable) and editing the four things that changed turns a fortnight into an afternoon, and removes the class of error that comes from retyping.
After the panel
The panel decides on a Tuesday. The letters go out three weeks later. In between: reconciling scores into a recommendation, getting the delegate’s approval by email, drafting two letter variants, mail-merging them, and checking that the merge did not put the wrong amount in the wrong letter.
Applicants do not experience your assessment rigour. They experience the wait after it.
Each of these is automatable without changing a single decision: aggregate the scores as they are entered, route the approval to the delegate with the limit that covers it, and generate letters from templates with merge fields drawn from the record rather than from a spreadsheet.
The part worth keeping slow
Moderation. The conversation where a panel resolves a genuine disagreement about merit is the highest-value hour in the whole round, and it is the one thing on this list that should not be compressed. Every other hour recovered is an hour that can go there.