| name | pipeline-health-report |
| description | Use when reporting on hiring health or investigating a stalled role — measures funnel conversion and time-in-stage, locates bottlenecks and drop-offs, and turns each finding into a recommended fix with a named owner. |
Pipeline Health Report
A playbook for turning the pipeline's raw activity into a decision-ready read on where hiring is winning and where it's leaking. The failure mode it prevents: a dashboard full of numbers that names no bottleneck, assigns no owner, and changes nothing.
Define the funnel and pull the data
- Fix the stages you'll measure end to end: sourced → replied → screened → advanced → onsite/loop → offered → accepted. Use the same stages across roles so comparisons hold.
- Pull the counts from the logged Tasks. Before analyzing, check completeness — if touches or grades aren't logged, note the gap; a report built on missing records misleads.
Measure conversion and speed
- Compute stage-to-stage conversion: what fraction advances from each stage to the next, per role and in aggregate.
- Compute time-in-stage: median (not just mean) days a candidate waits at each step. Slow stages lose good candidates as surely as low-conversion stages do.
- Compare against a baseline — a prior period, other roles, or a target — so a number reads as good or bad, not just as a number.
Find and diagnose the problems
- Rank bottlenecks by cost in hires, not by how bad the percentage looks. A modest drop at a high-volume stage can outweigh a steep drop at a tiny one.
- Separate signal from small samples — two declines is a coincidence; call a trend only when the count supports it.
- Bucket drop-offs by cause where you can: comp, speed, candidate experience, competing offer, mis-set expectations. Trace declines back to the stage and reason.
Recommend, don't just report
- For each finding, name the owner who can act, the specific change to try, and the metric that would prove it worked.
- Lead the report with the one or two fixes that would recover the most hires; everything else is supporting detail.
Quality bar
- Every headline metric has a baseline for comparison.
- Bottlenecks are ranked by cost in hires and separated from noise.
- Each finding names an owner, a fix, and a success metric.
- Data-quality gaps are stated up front.
Output format
Pipeline health report: funnel table (counts, conversion, time-in-stage vs. baseline) → ranked bottlenecks with cost in hires → drop-off/decline analysis by cause → recommended fixes with owner and success metric → data-quality notes.