Change Request Template
A change record proportionate to risk: what is changing, who is affected, how it is reversed, and how you will know it worked — without turning routine work into ceremony.
Reviewed 2026-09-21 · next review 2027-09-21
WHEN TO USE IT
For changes to shared systems where the people affected are not the people making the change — enterprise IT, network, shared platform, database. Routine application deploys through a tested pipeline should NOT use this; making them do so is the single most reliable way to get a change process routed around.
Tier it. A password policy change and a core routing change do not need the same scrutiny, and pretending they do trains everyone to treat the form as paperwork.
HOW TO USE IT
The requester fills it in; a reviewer who is not the requester approves it. The reviewer's job is to find the missing back-out and the unlisted dependency, not to re-derive the work. Approval should take minutes for a tier-3 change and a conversation for a tier-1.
Record the outcome afterwards. A change process with no record of what actually happened cannot tell you your change failure rate, which is the only number that justifies the process existing.
THE TEMPLATE
# Change request: <title>
Tier: 1 (high blast radius) | 2 | 3 (routine)
Requester: <name> Reviewer: <name> Window: <when>
## What is changing
Precise description. For configuration, the exact before and after.
## Why now
What breaks or fails to improve if this waits.
## Who is affected
Systems, teams and users. Name the ones who do not know yet.
## Back-out plan
The exact steps to reverse this, and how long they take.
If the change can remove your own access to the target, say how you
get back in.
## Verification
How you will know it worked — the specific check, not "monitor".
Include the end-to-end check, not only the component's own status.
## Dependencies and sequencing
What must happen before and after, and anything that must NOT run
concurrently.
## Risk if it goes wrong
Realistic worst case, and who needs to be available.
## Outcome (filled in after)
Succeeded | backed out | partial. What actually happened.
ADAPTING IT
In smaller organisations, tier 3 can be a single line in a channel with the back-out named — keep the discipline, drop the document. In regulated settings add the approval record and evidence retention.
The back-out and verification fields are the two that must never be optional at any tier. Everything else can be scaled down. See /releaseops for the release side and /itops for the enterprise-change process this fits into.