# First 90 Days — Cycle Review Sheet

> The checkpoints belong to the pair; this review belongs to the practice. Use with `first-90-days-guide.md`. The day-90 conversation is the compressed version; this sheet is the full one, run by HR or the manager after every completed cycle.

## How to run this review

- Fill it after every cycle that ends — not annually. One cycle, one page, in the same folder as the pulse slips.
- Review the loop, not the person: did the checkpoints happen on schedule, was the pulse read, did the day-30 change actually change?
- Aggregate across hires quarterly: the pattern is the onboarding product's bug report.

## Read the four checkpoints

1. **The slipped checkpoint** — a missed day-30 is not a calendar problem; it is the honest reading of what the manager protects when the sprint hits. Fix the slot, or admit the cycle is optional.
2. **The pulse that flatlined** — confidence high at week 1, invisible after. Either the survey died in an inbox or the answers stopped being safe. Both are system failures with different fixes.
3. **The change that drifted** — day 30 named it, day 60 forgot it. A promise kept late still counts; a promise quietly dropped teaches the new hire the local currency of words.
4. **The surprise exit** — someone left at month four and the sheets show nothing. Re-read the watch-out signals against the record: quietness, cancelled 1:1s, no questions. The paper usually saw it first.

## The six review questions

1. **The schedule truth.** Which checkpoint slipped most across your hires this year — and what does its slot actually compete with? → Targets: calendar honesty; a recurring meeting nobody defends is a rumor.
2. **The two-way grade.** What did new hires score lowest on the onboarding questions — and would you have believed them without the anonymous sheet? → Targets: the blind spot of the greeter; the pulse exists because exit interviews arrive too late.
3. **The kept promise.** Pick one day-30 "changed" item per cycle: kept, drifted, or improved? → Targets: word-to-action ratio; this is the number the new hire is really watching.
4. **The fit call timing.** When did the record first show "wrong seat" — and how many weeks were lost to optimism? → Targets: early honesty; a day-30 fit call is a favor, a month-four resignation is a surprise nobody deserved.
5. **The aggregate read.** Across cycles, where does confidence dip for everyone — week 2? day 60? → Targets: system repair; a dip shared by five hires is a process, not a person.
6. **The next-cycle change.** One change to the loop for the next hire, with an owner and a date. → Targets: durability; onboarding that never changes is just the founder's first week, copied forever.

## Failure modes of the review

| Failure mode | Why it happens | The rescue |
|---|---|---|
| Archive theater | Sheets filled, never re-read | The day-90 conversation starts from the previous cycle's review, out loud |
| Blaming the leaver | The record shows signals, ego rewrites them | The watch-out card is read against the record, not against memory |
| The anonymous lie | Pulse "anonymous" but answers traceable | Anonymity is architecture or it is decoration — fix the tool, not the culture deck |
| One-size cycles | Every role runs the identical loop | Role-specific blanks on the deck exist for a reason; use them |

## What this is not

This is not an HR audit form and not a productivity report. It is the practice's own ninety-day pulse — the evidence that the onboarding loop either learns or repeats. The new hire's experience is the signal; this sheet is where the signal stops being anecdote.