Product preview · the board pack
The people update your board actually reads.
One data export in. The whole story out.
Headcount, composition, movement, performance — the update every board asks for, assembled automatically from a catalog of tested story cards. Every claim carries its evidence; steady states are reported as findings, not filler; and the pack ends the way a good analyst does — what next, what else — with the next unit of certainty priced.
the people update · demo data · every claim computed + evidenced
0 · the watch — this period vs last
| Metric | Last period | This period | Δ | read |
|---|---|---|---|---|
| Regretted-exit share | 38% | 47% | ▲ +24% | watch |
| Internal fill rate | 48% | 41% | ▼ -15% | watch |
| Voluntary exit rate | 11.2% | 12.6% | ▲ +13% | watch |
| Engagement favorable | 64% | 71% | ▲ +11% | improving |
| Avg performance rating | 3.5 | 3.6 | ▲ +1% | steady |
| Headcount | 353 | 355 | ▲ +1% | steady |
Movers ranked by consequence, steady states reported as findings. The watch points the meeting at what moved the wrong way — the sections below give each its evidence.
1 · workforce — headcount
The story is: no story
Headcount is flat across the window — the wiggles never clear the certainty cutoff. The honest reading is stability.
What to do: Say 'stable' out loud in the review, set an alert at the cutoff so you'll know when that changes, and spend the meeting on a metric that's actually moving.
- ✓ best-fit form is Flat / stable (fit 39%)
2 · workforce — composition by department
The mix is holding steady
Across 4 periods, the composition of workforce by department has barely moved — total mix shift of 0.3pp across 4 categories.
What to do: Report it as one line — 'the mix is stable' — and free the meeting for the metrics that are moving. Set a threshold alert (≥3pp shift) so composition earns attention only when it changes.
- ✓ best-fit composition is Mix holding steady (fit 95%)
- ✓ total mix shift 0.3pp (needs ≤3pp)
3 · movement — exits by reason
The change has an address
The total for exits by reason moved +88.9% across 4 periods — and Better opportunity accounts for 100% of that movement (+8).
What to do: Re-state the headline with the address attached — 'the change is Better opportunity' — and direct the diagnosis there first. Decompose before you theorize.
- ✓ total changed +88.9% first→last (needs ≥5%)
- ✓ Better opportunity: 100% of the movement vs 33% of the base (+67pp disproportion; needs ≥55% and ≥+15pp)
4 · movement — voluntary exit rate by department (vs benchmark)
Some of you are over the line
1 of 4 groups sit above Biotech voluntary-exit benchmark (12.0) on voluntary exit rate % by department — the org-level average hides exactly who is over the line.
What to do: Report the over-the-line list by name with each group's gap to 12.0; assign owners per group, not an org-wide program — the reference was crossed locally, fix it locally.
- ✓ 1/4 groups materially above Biotech voluntary-exit benchmark 12.0 (by ≥10%) (needs ≥15% of groups)
5 · performance — rating distribution
The system isn't differentiating
performance ratings is compressed: 240 observations huddle around 3.6 with a relative spread of just 6.9%. The measure barely separates anyone from anyone.
What to do: Before trusting any decision built on this measure, fix the differentiation: forced-distribution review of a sample, calibration sessions, or a scale redesign. And audit what's currently being decided on these scores — it's being decided by noise.
- ✓ best-fit shape is Compressed — barely differentiated (fit 83%)
- ✓ relative spread (CV) is 6.9% (needs ≤12%)
what next? what else?
That’s the period in five findings — two worth acting on, three honest steady-states. More certainty is priced, not promised: segment drills on the exit hotspot, another period of history, a survey wave for the engagement section, or the driver analysis that says why. Each step is a known cost against a known decision.
the full product: one HRIS export in → this pack out, every period · posted price
This preview runs on demo data shaped like a real engagement. The product takes one HRIS export (roster + exits; survey export optional) and produces this pack fresh every period — same evidence discipline, your numbers. Want a metric interrogated by hand first? The Story Finder is the same engine, one question at a time.