Feedback Reconciliation Triage
Sort conflicting reviewer comments on a draft into the few that are real objections and the many that are taste, then resolve each with a named owner, so a piece that took an hour to write does not take four days to clear.
Version 1.0 · Updated 15 September 2026
What it is
Most delay in communications approval is reconciliation, not writing. A press release takes an hour to draft and days to clear, because three reviewers leave comments that pull in different directions and nobody is asked to say which of them matter. This template turns a pile of comments into a short table: each comment classified as a substantive objection, a factual correction, a stylistic preference or a question, with a proposed resolution and the person who decides. AI does the first pass of sorting and drafting the reconciled text; you make the calls.
By the end you have a single reconciliation note you can send back to every reviewer, showing what was taken, what was not, and why. It is most useful on pieces with three or more reviewers, and on any draft that has been through more than two rounds.
When to use it
Use this template when:
- Two or more reviewers have commented on the same draft and their edits conflict
- A draft is on its third round and nobody can say what is still wrong with it
- Reviewers are rewriting for taste and the writer needs a way to hold the line
- Legal or compliance comments need separating from executive preference
- You want a record of why an edit was declined, for the next time it comes up
Don’t use this template if:
- There is a single reviewer and the comments are clear
- The comments are all factual corrections with no disagreement between them
- The piece has a single owner with final say who is available to decide now
Inputs needed
Before starting, gather:
- The draft, with version number
- All comments, verbatim, labelled with the reviewer’s name and function
- The approval map for this piece, or a note of who holds final sign-off
- The agreed key messages and any fixed wording (legal lines, regulated claims, product names)
- The deadline, and what happens if it is missed
The template
Reconciliation note
Piece: [Title, version] Reviewers: [Names and functions] Round: [e.g. 3 of what should have been 2] Decision holder: [Role and name] Deadline: [Date and time]
Comment triage
| # | Reviewer | Comment (summarised) | Type | Conflicts with | Proposed resolution | Decided by | Status |
|---|---|---|---|---|---|---|---|
| 1 | [Name, Legal] | [What they asked for] | Objection / Correction / Preference / Question | [# of any conflicting comment] | [Accept / Decline / Amend, with the wording] | [Role] | Open / Resolved |
| 2 | |||||||
| 3 |
Type definitions
- Objection: the reviewer says the text is wrong, unsafe, non-compliant or off-message. Must be resolved before sign-off.
- Correction: a fact, name, number or date is wrong. Accept unless the reviewer is mistaken, in which case show the source.
- Preference: the reviewer would have written it differently. The writer decides, within the style guide; the reviewer is told.
- Question: the reviewer wants to understand something. Answer it; it may become an objection or go away.
Conflict resolution
For each pair of comments that contradict each other:
| Conflict | Comment A | Comment B | What is at stake | Resolution | Who decided |
|---|---|---|---|---|---|
| [e.g. #1 v #4] | [Legal wants the caveat] | [CEO wants it cut] | [Regulatory exposure v readability] | [Keep the caveat, move it to the final paragraph] | [Head of Comms] |
What goes back to reviewers
One message to all reviewers, in this order:
- Objections and corrections accepted, with the new wording
- Conflicts and how they were resolved, naming who decided
- Preferences not taken, with one line of reasoning each
- Questions answered
- What is needed from whom, by when, to close
AI prompt
Base prompt
I am reconciling reviewer comments on a draft [TYPE OF COMMUNICATION]. Below are the draft and every comment, labelled by reviewer and function.
Draft:
[PASTE DRAFT]
Comments:
[PASTE COMMENTS, EACH LABELLED "Reviewer name (function): comment"]
Fixed constraints that cannot change:
[e.g. agreed key messages, legal wording, regulated claims, product names]
Final decision holder: [ROLE]
Do the following:
1. List every comment in a table with: number, reviewer, a one-line summary, and a type from exactly these four: Objection (text is wrong, unsafe, non-compliant or off-message), Correction (a fact, name, number or date is wrong), Preference (the reviewer would have phrased it differently), Question.
2. Identify every pair of comments that conflict with each other, and say in one line what is at stake in each conflict.
3. For each Objection and Correction, propose the amended wording.
4. For each Preference, say whether the change improves clarity or accuracy. If it does not, propose declining it and give one sentence of reasoning I can send to the reviewer.
5. For each conflict, recommend a resolution and name the role that should decide it, using the decision holder above where the conflict is between a substantive objection and a preference.
6. Produce a short reconciliation message to all reviewers in this order: accepted changes with new wording; conflicts and resolutions; preferences not taken with reasoning; questions answered; what is still needed and from whom.
Be direct about which comments are matters of taste. Do not soften a Preference into an Objection to avoid disagreement. Where a comment is ambiguous, classify it as a Question and say what needs clarifying.
Prompt variations
Variation 1 – Legal against executive:
The main conflict is between legal review and an executive's edits on [PIECE]. Legal comments: [PASTE]. Executive comments: [PASTE]. For each point where they disagree, tell me: whether the legal comment is a compliance requirement or a caution; whether the executive's version can be achieved while meeting the requirement; and a third wording that does both where one exists. Flag anything where no compromise is possible and the decision has to go up.
Variation 2 – Rounds are not converging:
This draft has had [N] rounds of comments. Here is the comment history by round: [PASTE]. Identify comments that have been raised more than once, comments that reverse an earlier accepted change, and reviewers whose comments in later rounds are on text they approved earlier. Tell me the single issue most likely to be holding up sign-off and how to close it.
Variation 3 – Protecting a writer from rewrites:
A senior reviewer has rewritten large parts of [PIECE] for style. Original: [PASTE]. Rewritten: [PASTE]. Our style guide says: [PASTE KEY RULES]. Compare the two against the style guide and the agreed key messages, and tell me which rewrites improve the piece, which are neutral, and which weaken it. For each weakening rewrite give me a polite, specific sentence explaining why the original stands.
Variation 4 – Under deadline:
We have [X hours] before [PIECE] must go. Comments: [PASTE]. Sort them into: must resolve before publishing; can be resolved in a later version; can be declined now with a note. Draft the note for each declined comment. Assume the decision holder is [ROLE] and is available for one ten-minute call.
Human review checklist
- Is every Objection resolved, and does the reviewer who raised it agree it is resolved?
- Has every Correction been checked against a source, not accepted on authority?
- Have Preferences been decided by the writer or editor, within the style guide, rather than by whoever is most senior?
- Does every conflict name the person who decided it?
- Are declined comments explained in one sentence each, with no defensiveness?
- Has the fixed wording (legal lines, regulated claims) survived every accepted edit?
- Would a reviewer reading the reconciliation note understand what happened to their comment without asking?
- Has the final decision holder seen the conflicts, not only the resolved draft?
- Is the next round scoped to closing open items only, with no reopening of resolved ones?
Example output
Anonymised and shortened. Real outputs will vary based on your inputs.
Piece: Partnership announcement, v4 Reviewers: Anna (Legal), Dev (CEO), Priya (Brand) Round: 4 Decision holder: Head of Communications Deadline: Thursday 10:00
| # | Reviewer | Comment | Type | Conflicts with | Proposed resolution | Decided by | Status |
|---|---|---|---|---|---|---|---|
| 1 | Anna (Legal) | Add “subject to regulatory approval” to the first paragraph | Objection | 3 | Accept; place at the end of paragraph two rather than the lead | Head of Comms | Resolved |
| 2 | Anna (Legal) | Partner’s registered name is wrong | Correction | Accept | Writer | Resolved | |
| 3 | Dev (CEO) | Cut the regulatory caveat, it kills the lead | Preference | 1 | Decline the cut; accept moving it out of the lead | Head of Comms | Resolved |
| 4 | Dev (CEO) | “Delighted” should be “thrilled” | Preference | Decline; style guide avoids both, quote reworded to lead with the outcome | Writer | Resolved | |
| 5 | Priya (Brand) | Second paragraph uses the old product name | Correction | Accept | Writer | Resolved | |
| 6 | Priya (Brand) | Is the partner’s logo cleared for use? | Question | Asked Legal; answer due Wednesday | Legal | Open |
Message to reviewers: Three changes taken (registered name, product name, caveat added at the end of paragraph two). One conflict, caveat placement, decided by the Head of Comms: the caveat stays, moved out of the lead. One preference not taken: “thrilled”, because the style guide avoids emotional adjectives in executive quotes and the quote now leads with the outcome. One open item: logo clearance, with Legal, due Wednesday. Sign-off requested Thursday 09:00.
Tips for success
Classify before you resolve. The temptation is to start editing. Sort every comment first; half the conflicts disappear once it is clear which ones are taste.
Name the decider, every time. A conflict with no owner becomes a fourth round. The approval map should already say who holds final say; if it does not, fix that with the Approval Workflow Mapper.
Send one note, not a thread. Reviewers reply to what they can see. A single reconciliation note showing every comment’s fate stops the same point being raised again by someone who missed the answer.
Keep the declined preferences. A short record of what was declined and why is the fastest way to stop the same stylistic argument recurring on the next piece.
Use the AI pass as a second opinion, not a verdict. It is good at spotting that two comments contradict each other and at drafting neutral reasoning. It cannot know that the CEO’s preference is a hill worth dying on this week. That is your judgement.
Common pitfalls
Treating seniority as substance. An executive’s stylistic preference is still a preference. Classify it honestly, then decide whether to accept it anyway; that is a choice, not a category.
Letting a Question stay a Question. Unanswered questions become objections in the next round. Answer them in the reconciliation note.
Merging edits without checking the fixed wording. Accepting five reasonable edits can quietly remove a regulated caveat. Check the constraints line after every merge.
Reopening resolved items. Once a conflict has been decided by the named owner, the next round is for open items only. Say so in the note.
Pasting confidential comments into a tool without checking the policy. Reviewer comments can contain legal advice or unannounced information. Follow your AI Use Policy on what can go into which tool, and strip anything that should not travel.