borghei/Claude-Skills / project-management/execution/decision-questionnaire
decision-questionnaire
VerifiedTurn a blocked decision into an async questionnaire a stakeholder can answer without a meeting. Use when you need someone else's input to decide, a meeting would be overkill, or answers from several people must be consolidated.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 8.5 KB |
| assets/decision_record_template.md | markdown | 1.2 KB |
| assets/questionnaire_template.md | markdown | 2.0 KB |
| assets/sample_answers.json | config | 642 B |
| assets/sample_answers_conflict.json | config | 304 B |
| assets/sample_questionnaire_clean.json | config | 3.1 KB |
| assets/sample_questionnaire_flawed.json | config | 1.6 KB |
| references/question-design.md | markdown | 11.1 KB |
| references/sending-and-consolidating.md | markdown | 8.3 KB |
| scripts/answer_consolidator.py | script | 8.7 KB |
| scripts/questionnaire_linter.py | script | 11.6 KB |
references/sending-and-consolidating.md
View on GitHub# Sending, Following Up and Consolidating Answers
Writing the questionnaire is half the work. This reference covers the send itself, variants for more than one recipient, the follow-up cadence, turning answers into a decision record, and when to abandon async and meet.
## 1. Choosing async over a meeting
Use a questionnaire when all of these hold:
- The decision needs **specific knowledge** from one to five people, not a negotiation among them.
- The questions can be **stated precisely now**. If you cannot yet word the question, you need a conversation to find it, not a form.
- Respondents can answer **independently** — one person's answer does not depend on hearing another's first.
- A delay of several working days is acceptable.
Prefer a live conversation when:
| Signal | Why async fails |
|---|---|
| The recipients disagree with each other and know it | Independent answers just document the conflict; someone has to broker it |
| The decision is emotionally loaded (reorgs, cuts, people) | Written questions read as cold; answers get guarded |
| You cannot write the options table | You are still in discovery; a questionnaire would fix the frame too early |
| The answer is needed today | Async realistically needs two working days minimum |
A useful hybrid: send the questionnaire anyway and use it as the agenda for a short meeting. Respondents who fill it in beforehand make the meeting fifteen minutes instead of an hour.
## 2. The send message
The questionnaire travels inside a short message (email, chat, ticket comment). That message has four lines and nothing else:
1. What you need: "I need your input on where the analytics add-on is hosted for EU customers."
2. Why them: "You know what spend is committed for Q1, which I don't."
3. Cost and deadline: "About 12 minutes, by Thursday 20 November."
4. The default: "If I hear nothing, I'll go ahead with Frankfurt."
Do not paste the questions into the message body *and* attach the document — respondents answer whichever they see first, and you end up merging two partial answer sets.
### Where answers go
Decide one place for answers before sending: the document itself (inline under each question), a form, or a reply. Mixed channels are the most common source of lost answers. If respondents will see each other's answers, say so in "how your answers will be used" — it changes what people write.
## 3. Multiple recipients
| Situation | Variant | Notes |
|---|---|---|
| Same knowledge, need a cross-check | **Same questionnaire to each**, answered independently | Do not let respondents see each other's answers before submitting; anchoring makes agreement meaningless |
| Different knowledge per person | **One core plus per-recipient sections** | Keep the critical decision question identical for all; add role-specific questions only to the person who can answer them |
| One decider, several advisers | **Advisers first, decider second** | Send advisers the questionnaire, consolidate, then send the decider a two-question follow-up with the consolidated view |
| Large group (6+) | **Not a questionnaire** — use a survey or a decision meeting | Above about five, consolidation becomes polling and the "decision" becomes a vote nobody owns |
When the same question goes to several people, the consolidator compares answers:
- **pick-one / yes-no:** any difference is a conflict (yes-no compares the verdict, ignoring the stated condition).
- **ranked:** a conflict only when first choices differ; differences lower in the ranking are reported but not blocking.
- **number:** a conflict when the spread exceeds half the median (e.g. 1,500 vs 4,000).
## 4. Follow-up cadence
| When | Action |
|---|---|
| Send day | Send; put the deadline in your own calendar |
| Midpoint | One reminder to non-respondents only, restating time needed and the default |
| Deadline −1 working day | Direct message: "Tomorrow I'll go ahead with the default unless you'd like to weigh in." |
| Deadline | Run the consolidator. Record the decision, or follow up on blocking questions only |
| After deadline | Apply the default if critical questions are unanswered; tell non-respondents what was decided |
One reminder and one last-call is the ceiling. More than that trains people that your deadlines are soft.
### Follow-up on a partial answer
Do not resend the whole questionnaire. Send only the unanswered critical questions, or the conflicting ones, with what you already know: "Security picked A, you picked C — what would need to be true for A to work for you?" That is a different, sharper question than the original, and it often resolves the conflict in one reply.
## 5. Reading the consolidator outcome
| Outcome | Meaning | Next step |
|---|---|---|
| `ready-to-record` | Every critical question has a usable, consistent answer | Write the decision record; resolve non-critical conflicts at leisure |
| `needs-follow-up` | A critical question is unanswered before the deadline, or answered with a conflict | Targeted follow-up on the listed questions only (exit code 1) |
| `default-applied` | Deadline passed with critical questions unanswered and no conflict blocking | Record the decision as "decided by default" and notify recipients |
A conflict on a critical question blocks even after the deadline: the default exists to replace silence, not to overrule an explicit objection. When a respondent has said "not A", proceeding with A by default is a decision someone must own, so the tool keeps it in follow-up.
"Don't know" answers are recorded separately, never treated as votes. Three "don't know" responses to a critical question with no real answer is functionally unanswered — and a signal you asked the wrong people.
## 6. Writing the decision record
Use `assets/decision_record_template.md`. The record exists so that, six months later, someone can see what was decided, on what evidence, and what would reopen it. Fill it in this order:
1. **Outcome first**, stated as an action. If the default applied, say so in the first line.
2. **What the answers established** — one row per question that influenced the outcome, with the answer per respondent and how it was used. Questions that turned out not to matter can be omitted; note that they were asked.
3. **Disagreements** — every conflict, how it was settled, by whom, and why. A record that hides a disagreement invites the losing side to reopen the decision later.
4. **Open items** — useful "don't know" answers and catch-all points that need their own decision, each with an owner.
5. **Revisit if** — the concrete condition that would reopen the decision.
Then close the loop: send the record to every recipient, including those who did not answer. People who answered see their input used; people who did not learn what was decided without them, which is the strongest incentive to answer the next one.
## 7. When the questionnaire itself was wrong
Signs the questionnaire asked the wrong thing:
- Most answers land in "none of these" or the catch-all.
- Respondents answer a different question from the one asked.
- A respondent replies "I'm not the right person for this".
- The answers are consistent but do not let you decide.
Do not patch the questionnaire and resend to the same people. Go back to the gap analysis: either the recipient does not hold the knowledge you thought, or the decision is not yet framed well enough for questions. A short conversation is usually the faster repair.
## 8. Worked example (from the sample files)
The sample questionnaire asks a Security Lead and a Finance Controller to help choose the EU hosting region for an analytics add-on.
- **Clean run, day before the deadline** (`sample_answers.json`, `--today 2026-11-19`): both pick A, cost ceilings 4,000 and 3,500 (spread under half the median, so no conflict). They disagree on whether backups must stay in the EU — but that question is not critical. Outcome `ready-to-record`; the backup question goes into "Disagreements" with a follow-up owner.
- **Conflict run** (`sample_answers_conflict.json`): Security picks A, Finance picks C. Q1 is critical, so the outcome is `needs-follow-up` (exit 1). The follow-up is one question to Finance: what would make A acceptable?
- **Silence after the deadline** (an empty response list, `--today 2026-11-21`): outcome `default-applied`; the record states that Frankfurt was chosen without Security and Finance input.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |