Your team, thoughtfully built.
How to Write a Customer Case Study With AI
To write a customer case study with AI, give it the customer's own words, the before and after, and what you have permission to publish, and ask for a draft that adds nothing to any of them. Quotes stay word for word. Results appear only if they were measured. A case study is believed because it is specific and true, and it stops being both the moment a draft rounds a result upward.
The example below is invented to show the method. It is not a real customer.
Permission and measurement first
A workshop organizer started sending a clearer arrival checklist. They say they now get fewer repeated questions about equipment. They have not counted.
So the story can describe the change and what the organizer noticed. It cannot say "50% fewer questions." Before anything is written, settle what you may use: the name, the logo, the quote. Without permission the draft stays internal and anonymous.
The request
The request: "Draft a case study from the interview and the two checklist versions below. Explain the original problem, the change, how it was used and what we don't know. Keep quoted wording exactly as given, and cite the source for any number. Do not invent a percentage, imply more endorsement than we have permission for, or publish the draft."
What comes back
The old checklist told participants to bring tools. The new one says tools are supplied and to bring a bicycle. The organizer used it in new confirmations and reported fewer repeated equipment questions. No count is available.
With it: the list of exact quotes, a table of each claim and where it came from, and the permission status.
The mistakes to watch for
The number that appears. "Fewer questions" becomes "cut support questions in half" in one edit. Case studies are expected to have a number, and a draft will supply one. If nobody counted, there isn't one.
The quote that got better. Tidying a customer's sentence so it reads well puts your words in their mouth, with their name on it. Use what they said, or ask them to approve the change.
"Eliminated." One customer's good experience doesn't show the checklist caused it, and it shows nothing about your other customers. Say what happened to this customer.
If the numbers arrive
Should the organizer later send counts, check that the two periods are the same length and measure the same thing before a number goes in. Questions are not bookings, and one quiet week is not a trend. Then send the customer the final version before it is published. It is their name on it.
Run it in your Staffmor office
Open the office on the computer running it with its Start Staffmor launcher. If you already have a paired web office, open it here. Paste this request and the source material into the conversation. Use your approved Codex connection, or ask your open Claude Code session to "Check Staffmor for work." A saved request is not a completed result. Review the answer or file when it comes back and reply with corrections. Ask for the claim-and-source table first and approve it before the story is written.
Staffmor structures the evidence and writes the draft. Getting the customer's permission, and publishing, stay with you. New to Staffmor? Set up your office with Codex or Claude Code and an empty folder. Use your own supported AI account and approve the download, office activation and local folder/network access during setup. Claude Code pickup is manual; keep that local session open.
Related: How to write customer interview questions with AI · How to write instructions AI agents actually follow