Marketing teams would never run a funnel without knowing which step loses people. Hiring teams do it constantly — reporting time-to-hire and cost-per-hire, neither of which tells you where candidates are leaving.
The numbers worth having
Four, per stage:
- Started — how many were invited and began
- Completed — how many finished
- Median time to start — how long the invitation sat unopened
- Withdrawn — who left, and after which step
The completion rate on any stage is a direct readout of whether that stage is worth what it asks. A screening step with 55% completion is not filtering for motivation; it is filtering for tolerance.
What the drop-offs usually mean
Low start rate. The invitation is unclear about time required, or arrived from a sender nobody recognised, or the deadline sounded like homework.
High start, low completion. The task is longer than advertised. This is the most common and most fixable finding, and the fix is almost always to make the task shorter rather than to explain it better.
Long time-to-start. Candidates are unsure whether it is worth doing, which usually means the role description and the effort being asked for are out of proportion.
Withdrawal after a scheduled step. Someone else moved faster. Nothing to fix in the step itself — the problem is upstream, in elapsed time.
The one thing to fix first
Whichever stage has the lowest completion rate, cut its length in half and see what happens to quality. In most processes, quality does not move and completion jumps ten to twenty points, which means the extra length was buying nothing and costing candidates.
Ask them
Two questions at the end of any process, win or lose: how long did that take compared to what you expected, and what would you change. Free text, optional, takes thirty seconds.
The answers are more useful than any benchmark report, because they are about your process rather than the average of everyone's.