Explain What Your Hiring Funnel Metrics Are Telling You
When to use it
Turn a column of stage-by-stage numbers into a plain-English read on where the funnel is leaking, and why the req is taking as long as it is.
The prompt
Help me interpret this hiring funnel data. ROLE: {{Job Title}}, {{seniority}}, {{location or work model}} STAGE COUNTS: {{e.g. 120 applied, 40 screened, 12 interviewed, 3 final round, 1 offer, 1 accept}} TIME IN EACH STAGE: {{if known}} SOURCE MIX: {{roughly where the top of funnel came from — inbound, sourced, referral, agency}} HOW LONG THE REQ HAS BEEN OPEN: {{weeks}} WHAT I'D EXPECT NORMALLY: {{your own historical rates, if you have them}} Before interpreting anything, tell me whether these numbers are large enough to conclude anything at all, or whether I'm about to read a pattern into noise. Then: 1. Which stage has the biggest drop-off relative to what's typical, and how confident you are that it's genuinely unusual. 2. Whether that drop-off looks like a top-of-funnel quality problem, a process and speed problem, or a bar-calibration problem — and what evidence would distinguish them, since all three look identical in the counts. 3. If this req has been open longer than I'd like, where the time is actually going — candidate supply, my screening speed, scheduling, or the decision itself — and which of those I can change without anyone else's agreement. 4. What I should check in the data before accepting your reading. 5. What these counts can't tell me at all, and what I'd need to start logging to answer it next time. 6. The single next action, stated as something I could actually do this week. Don't give me a benchmark for this role type unless you also say where it comes from and how much it varies.
Tip
Never let it hand you a root cause with confidence. It is reading counts with no view of your process, your market or your interviewers — treat its answer as a hypothesis to check, not a diagnosis.
How to use it
Paste the prompt into ChatGPT, Claude or whichever assistant you use, then replace every {{bracketed}} part with your own detail. The more specific and messier your input, the better the output — a model given raw notes has more to work with than one given a tidy summary you wrote first.
More Reporting & Metrics prompts
Open the full tool for all 87 prompts, the glossary, and the JD decoder.