Change Management Checklist for Small Teams (No CAB Required)
Big companies run change advisory boards. Small teams run on the honor system — and the honor system fails the same way every time: someone ships a "quick config tweak" on Friday afternoon, it takes down the thing nobody was watching, and the postmortem opens with "why wasn't this reviewed?" You don't need a committee. You need one ten-minute record and four questions. Here's the checklist.
The 10-minute change record — copy this
- Name the change in one sentence. "Rotate session timeout from 30d to 8h on the billing app." If the sentence is vague, the change is vague — and vague changes are the ones that roll back badly.
- Classify it: standard, normal, or emergency. Standard = done before, scripted, low blast radius (deploys, cert renewals). Normal = everything else. Emergency = production is broken and you're fixing forward. Only normal needs the rest of this list.
- Answer the 4 questions that replace a CAB. What breaks if this goes wrong? How do we know within 15 minutes that it went wrong? What's the rollback — and has anyone actually run it? Who's watching it for the first hour? A change with no answer to #3 doesn't ship.
- Check the freeze rules. Friday after 2pm, the last two days of a quarter, the day before a customer demo, and any window overlapping an on-call handover are freeze windows by default. A change can still ship — but it needs a named approver who isn't the person doing it.
- Write the rollback as commands, not intentions. "Re-enable feature flag X and redeploy tag 4.1" — not "undo the change". The 2am version of you runs commands, not judgment.
- Ship it in a change window with a watcher. Deploy, then fifteen minutes of watching the errors dashboard and the customer-facing status. Unwatched changes are discovered by customers.
- Close the record within 24 hours. Outcome, actual rollback used (yes/no), and one lesson. Ten closed records a quarter beats a process nobody maintains.
The one-page change record
| Field | Record |
|---|---|
| Change | Billing app: session timeout 30d → 8h |
| Class / risk | Normal — auth flows touched, ~400 users affected |
| Rollback | Set SESSION_TTL=30d in config, restart app pool (tested Aug 30) |
| Detection | Error-rate alarm >2% for 5 min; manual login test at +10 min |
| Window / watcher | Wed 14:00-16:00 — on-call engineer watching |
| Outcome | Shipped 14:22, error rate flat, closed same day |
The rules that make the checklist stick
- The requester is never the approver for normal changes. One other pair of eyes, five minutes, async is fine. This single rule catches half of all bad changes.
- Emergency changes get a retro-record, not a pass. The rule isn't "no changes without approval", it's "no changes without a record". Fix-forward first, write the record within 24h, and the process survives contact with reality.
- Standardize until only normal changes remain. Every change that happens twice becomes a script or a runbook — then it re-classifies as standard and stops costing review time. That's the automation path: fewer decisions, same safety.
- Failed rollbacks are finding #1. When a change goes wrong and the rollback also goes wrong, that's two defects — and the rollback one is the expensive lesson. Test rollbacks on slow weeks, not during incidents.
- The record is the audit. Auditors, customers, and insurers don't ask "do you do change management?" — they ask to see records. Twelve dated records answer that question.
---
The Ops Starter Kit Vol. 2 ($27) includes the full change-control pack — change record, freeze calendar, and the rollback log this checklist plugs into — launch week: 30% off any paid kit with code HIVE-LAUNCH30 at checkout.