# Trend Reassessment

> Weekly trend reassessment (PRD §4.10.1). For each LIVE store, re-evaluate the trend behind it on current data (signals + GA4 + orders KPIs), re-score dimensions/risk/lifecycle, and write a SUGGEST-ONLY refresh proposal (refresh_proposals). Nothing on a live store changes until the operator approves a proposal.

- Skill: `guillaumemeyer/trend-reassessment` (Agent Skill)
- Install (CLI): `npx skillmds@latest add guillaumemeyer/trend-reassessment`
- Raw SKILL.md: https://api.skillmd.com/api/skills/guillaumemeyer/trend-reassessment/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Product & Planning
- Author: guillaumemeyer (https://skillmd.com/u/guillaumemeyer)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/guillaumemeyer/trend-reassessment

---


# Trend Reassessment (weekly)

## Role

Every week, for each store in status `live` (or `paused`), re-check whether the
trend still holds and propose updates. **Suggest-only**: writing a proposal
never mutates the live store.

## Flow (per store)

1. Pull the store's trend + current state (branding, products, blog, KPIs:
   GA4 events, orders, customers).
2. Re-check signals on current data (SerpApi / Google News / X / external
   events + GA4 + Search Console + orders): re-score the four dimensions,
   `total_score`, lifecycle, and per-category risks (reuse the `scoring`
   skill rules).
3. Write a **refresh proposal** (`refresh_proposals`, status `proposed`):
   - `trend_evolution` summary (what changed vs original report).
   - Suggestions: brand copy/visual tweaks, store page updates, blog topics,
     product additions/retirements, **new products** and **new concept+SKU
     combinations** — each new concept includes a LOW-RES preview (image gen,
     bounded cost); final HD assets only after approval.
4. If the trend has declined sharply or the store's KPIs are flat/bad, say so
   explicitly in the proposal and recommend retirement or pause — still
   suggest-only.

## Persist

- One `refresh_proposals` record per store per cycle (idempotent on
  store+week).
- One `agent_runs` row: `{stage: "analyze", status: "ok", summary, metadata:
  {proposal_id}}`.

## Operator actions (back-office, not you)

- **Approve** → proposal applied (build-check picks it up).
- **Dismiss** → proposal archived, store untouched.

## Completion contract (final output)

```text
skill: trend-reassessment
status: ok|partial|failed
stores_reassessed: N
proposals: N
agent_run_id: <pb_id>
```