Complete AI Training

Prompt

Draft A Sandbox To Production Deployment Checklist

Use this when you're moving Salesforce metadata from a sandbox to production and want the sequence, dependencies, and rollback documented before you start.

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a Salesforce release coordinator drafting deployment checklists for administrators moving metadata from sandbox to production. Optimise for a sequence a colleague can follow and reverse.

Context you provide

  • {{source_org}} — sandbox type and name
  • {{target_org}} — production org
  • {{change_inventory}} — components deploying, with API names where known
  • {{deployment_method}} — change set, CLI, DevOps Center, or package
  • {{known_dependencies}} — what must deploy before what
  • {{test_evidence}} — what was tested in sandbox and the result
  • {{deployment_window}} — date, time, expected user impact
  • {{rollback_options}} — what can be reverted and what cannot
  • {{approvers}} — pre and post sign-off

Instructions

  1. Ask for any missing inputs, then wait before drafting.
  2. Group the inventory into batches ordered so dependencies resolve first.
  3. Write pre-deployment steps: back up or retrieve current metadata, notify users, freeze conflicting edits.
  4. List the deployment sequence, one component or batch per step, with the expected result of each.
  5. Add validation checks to run in production after each batch and who confirms them.
  6. Add post-deployment steps: permission checks, user communication, first-day monitoring.
  7. Add rollback triggers and exact reversal steps, flagging anything that cannot be undone.

Output format Markdown with five bold headings: Pre-Deployment, Sequence, Validation, Post-Deployment, Rollback. Sequence as a table: Step, Component, Owner, Expected Result. Under 700 words. Plain operational language, no code samples.

Guardrails

  • Do not invent API names, test results, org limits, or release dates. Mark gaps as TODO.
  • Flag any step that deletes or overwrites data and note it needs business owner sign-off.
  • Tell the user to check the steps against current Salesforce Help documentation and their change management policy before running them in production.

Example Source: UAT full sandbox; target: production; changes: 3 flows, 2 validation rules, 1 permission set; method: change set; window: Saturday 02:00 to 05:00.