Separate how customers feel from the actual failures to fix

A draft of actionable failure themes with quoted evidence and the emotional context kept separate.

Choose one of three approaches. The steps for that option appear below.

Choose an option above. On a phone, swipe the tabs left or right to see all options →

Claude: separates emotion from failures and drafts

Claude sorts emotional statements from service claims and drafts the themes. You check every row and every statement.

  1. You

    Prepare the material

  2. Claude

    Sort statements, draft themes

  3. You

    Check and finish

  1. Have ready: the feedback text and coding guide

    Use: You

    Take credentials and payment details out of the supplied material, keep source IDs unchanged, list the approved rules, scope and as-of dates, set originals aside.

    Check

    Check no password, authentication code, session token or payment detail remains, and nothing the task needs was removed.

  2. Sort statements, then draft the themes

    Use: Claude

    In one Claude chat, ask for the evidence worksheet as a chat table, then the draft failure themes with quoted evidence, both as chat text.

    Details

    What to paste into the tool

    Task: Evidence worksheet table in the chat reply with row IDs, source references, original values, status and unresolved gaps. Attached are the supplied files. Build one row per item: stable row ID, source file and location, exact quote or original value, applicable rule or version, confirmed or reported status, unresolved gap. Separate emotional statements from observable service claims and code the claims with the coding guide. Keep counterexamples. Do not infer motives or demographic traits. List anything missing, contradictory or unchecked as unresolved; do not fill gaps by inference. Return the worksheet as a table in the chat reply; do not create or return a file and do not run code. Task: Draft actionable failure themes with quoted evidence and emotional context; Explicit unresolved questions and a proposed next human decision, not an executed action. Using only the corrected worksheet, draft the failure themes packet with quoted evidence and emotional context. Keep counterexamples, separate observed patterns from causes, and do not treat emotion as proof of severity. State unresolved questions and a proposed next human decision. Return the draft as chat text; do not create or return a file and do not run code.

  3. Check every row and finish

    Use: You

    Check every worksheet row against the original files, correct in the same chat, then check the draft and close it as preparation only.

    Check

    Check the draft stays inside the checked worksheet rows and treats emotion as context rather than proof. Do not resolve missing or contradictory facts by inference. Settle recipient scope and authority separately.

Before you paste into any AI tool, remove any confidential information. Put it back at the end.

Tools in this workflow

Deciding between ChatGPT and Claude for this? See the comparison.

Final check

Never paste credentials or full payment details.

Reviewer checks every row and statement against the supplied originals.

No password, code, token or payment detail is entered or reproduced.

Missing or contradictory facts stay marked as unresolved.

Nothing is sent, filed, assigned or committed; any later action needs separate approval.

Emotion is not treated as proof of severity or of customer traits; counterexamples are kept.

Suggestions and preparation only. Do not send, publish, change records or accounts, assign work, execute code, initiate a purchase, or commit a remedy or deadline. A human separately approves any downstream action.

Check the result against your original inputs. Approve it before sharing or acting.

What still needs checking
  • Coding guide edge cases the supplied guide does not settle
  • Items where emotion and claim stay intertwined after review

Share this page