Key finding
The sample playbook requires the order number and a damage photo. The rep attaches both to the case and sends it to Claims, which decides the remedy.
Jump to a section
“A customer received a damaged item. What should I collect?” is a useful pilot question because the answer changes what the rep does while the customer is still available. Leave out the photo, and someone may have to ask the customer to come back.
Write the answer key before opening chat
Our one-page sample support playbook calls for the order number and a photo of the damage. The rep attaches the evidence to the case and routes the completed case to Claims. Those details make a small, testable answer key: two pieces of evidence, one destination.
Ask the question in the LuxonLink demo and open the cited PDF. Read the passage out loud. A reply like “document the issue and escalate” sounds familiar, but the rep still needs to know what to collect and who receives it.
Change the customer’s request
Now ask “Can I promise a refund while Claims reviews the case?” Check the playbook for authority to make that promise. The sample workflow explains evidence and routing. It does not give a rep a blanket approval to issue a refund. A useful answer preserves that distinction and leaves the decision with the appropriate owner.
The same exercise works with a real support playbook. Pick a question where the wrong answer creates rework or commits the business to something. Ask a senior rep to mark the exact sentence that supports the next action. If nobody can find that sentence, you have found a documentation question before it becomes a chatbot question.
Use the pilot scorecard to record the customer’s wording, the missing or correct detail, and the cited page. The next reviewer can repeat the test against the same evidence.