# Crewhu Surveys

> Crewhu CSAT/NPS survey data: the list/search/get tools and the detractor/promoter sentiment slices, the trend, detractor-follow-up, promoter-recognition and per-tech roll-up workflows, and the edge cases that skew scores — sparse response counts, comment-only feedback, and tenant-local timestamps.

- Skill: `wyre-ai/crewhu-surveys` (Agent Skill)
- Install (CLI): `npx skillmds@latest add wyre-ai/crewhu-surveys`
- Raw SKILL.md: https://api.skillmd.com/api/skills/wyre-ai/crewhu-surveys/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: WYRE-AI (https://skillmd.com/u/wyre-ai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/wyre-ai/crewhu-surveys

---


# Crewhu Surveys (CSAT / NPS)

Surveys are Crewhu's primary product — short CSAT/NPS responses tied
to a ticket close, a specific tech, or a project milestone. This skill
covers listing surveys, finding a specific one, and isolating the two
ends of the curve (detractors and promoters) so MSP managers can react.

## Anti-triggers

- **"User" here is an MSP technician, not an end user.**
  `crewhu_users_*` returns the staff being scored. A customer's staff
  account in a tenant is `m365-users` or `cipp-users`; the customer's
  billing or ticket contact is a PSA contact (`connectwise`,
  `halopsa`, `kaseya/autotask`).
- **Acting on a detractor** — Crewhu records the feedback and stops
  there. Raising the follow-up ticket, logging the call-back, or
  crediting the account happens in the PSA.
- **Compliance or insurance questionnaires** — those are also
  "surveys" in MSP conversation but share nothing with CSAT; use the
  `compliance-pack` or `scalepad` (`scalepad-controlmap`).

## API Tools

### List & Search

| Tool | Purpose |
|------|---------|
| `crewhu_surveys_list` | Paginated list of recent survey responses |
| `crewhu_surveys_search` | Search surveys by keyword / metadata |
| `crewhu_surveys_get` | Pull full detail for one survey response |

### Sentiment Slices

| Tool | Purpose |
|------|---------|
| `crewhu_surveys_detractors` | Detractor responses (low scores / negative comments) |
| `crewhu_surveys_promoters` | Promoter responses (high scores / positive comments) |

## Common Workflows

### Pull recent survey trend

1. Call `crewhu_surveys_list` with a wide enough window (90 days
   minimum).
2. Bucket results by week and compute average score and response count.
3. Surface the trend line, not just the latest week — single weeks are
   noisy.

### Detractor follow-up queue

1. Call `crewhu_surveys_detractors` to get the negative responses.
2. For each, call `crewhu_surveys_get` to retrieve the full comment
   and the responsible technician (the responses include user
   attribution — cross-reference with `crewhu_users_get`).
3. Produce a follow-up list: customer, ticket, tech, score, comment,
   suggested action (call-back, account-manager hand-off, escalation).

### Promoter recognition loop

1. Call `crewhu_surveys_promoters` to find positive responses.
2. Use the responsible tech to drive a recognition workflow:
   - `crewhu_badges_user_recognition` to see whether the tech is
     already trending in recognition.
   - Consider awarding a badge via the `badges` domain (see
     `crewhu_badges_history_list`).

### Per-user roll-up

1. Use `crewhu_users_list` to enumerate techs.
2. For each tech, use `crewhu_surveys_search` (keyed on the user) to
   pull their responses.
3. Compute average score, response count, and detractor rate per tech
   for a manager scorecard.

## Edge Cases

- **Sparse responses** — Some techs have few surveys; flag any
  per-user metrics with N < 10 as low-confidence rather than ranking.
- **Comment-only feedback** — Some responses have a comment without a
  score. Surface those separately; do not coerce them into the score
  average.
- **Time zones** — Survey timestamps are in the tenant's configured
  zone; normalize before comparing across tenants.

## Best Practices

- Always show denominators (response count) alongside averages.
- Pair detractor lists with promoter lists in a manager dashboard so
  the picture is balanced.
- Capture the tech for every response in your output so action items
  are unambiguous.

## Related Skills

- [api-patterns](../api-patterns/SKILL.md) - Auth and pagination