google-trends-research（Google trends 研究）

Google Trends Research

Follow shared release-shell rules in:

• postplus-shared release-shell rules

Use this skill for Google Trends platform-data work.

Typical requests:

• 看最近哪些关键词或话题在变热
• 比较几个关键词的搜索趋势
• 看某个主题在哪些国家或地区更热
• 找 rising queries 或 related topics
• 给内容研究、选品研究、campaign planning 提供 search-intent signals

Read first:

• postplus-shared research preferences

Core Rule

Treat Google Trends as a search-intent source, not as full demand proof.

Good uses:

• topic discovery
• keyword momentum tracking
• regional interest comparison
• rising-query discovery
• watchlist monitoring

Do not overclaim from Google Trends alone:

• transaction demand
• conversion intent
• marketplace competitiveness
• creator or content execution quality
• merchant-model fit

Those need to be combined with marketplace, content-platform, or business-context evidence.

Task Shapes

Classify the request first:

1. Trending Now Scan

Use when the user asks:

• 最近 Google 上什么在热
• 某个国家今天 / 这几天的热点搜索是什么
• 先给我一批实时热点主题

Preferred hosted collection key:

• google-trends-fast

2. Keyword Momentum Check

Use when the user asks:

• 这个关键词最近有没有变热
• 比较几个词的趋势变化
• 看过去几个月 / 一年里的搜索热度

Preferred hosted collection keys:

• google-trends-fast for the current released fast path
• google-trends-keywords for the older keyword-only path

3. Regional Interest Mapping

Use when the user asks:

• 哪些国家 / 地区对这个主题更敏感
• 这个词在不同市场哪里更热
• 想做 geo priority 判断

Preferred hosted collection key:

• google-trends-fast

4. Related Query Expansion

Use when the user asks:

• 这个主题还有哪些相关搜索词
• 有没有 rising queries 可当 seed
• 帮我扩一组关键词池

Preferred hosted collection key:

• google-trends-fast

Hosted Collection Strategy

Do not expose provider selection to the user.

Current defaults:

• broad keyword analysis: google-trends-fast
• realtime trending feed: google-trends-fast
• older keyword-only path: google-trends-keywords

Choose the narrowest actor that fits the task.

Use google-trends-fast by default when the task needs:

• interest over time
• geo comparison
• realtime trending searches
• regional trend signals

Use google-trends-keywords only when the workflow is already validated against that older keyword-only input shape.

Hosted Collection Input Fields

google-trends-fast

Keyword analysis input:

{
  "enableTrendingSearches": false,
  "keyword": "led skincare device",
  "predefinedTimeframe": "today 12-m",
  "geo": "US",
  "fetchRegionalData": false
}

google-trends-fast defaults to trending-search mode in its hosted input schema. For keyword analysis, enableTrendingSearches must be explicitly false; otherwise keyword fields are ignored by the actor.

Trending-search input:

{
  "enableTrendingSearches": true,
  "trendingSearchesCountry": "US",
  "trendingSearchesTimeframe": "24",
  "trendingSearchesMaxItems": 50
}

Repo-Owned Runner

For product-shell execution, use the repo-owned collection runner:

node skills/20-research/google-trends-research/scripts/collection_actor_run.mjs \
  --collection-key google-trends-fast \
  --input <input.json> \
  --output <raw-output.json>

Recommended Workflow

Use the lightest valid chain:

1. classify the request into one task shape
2. collect a small valid sample
3. extract the trend signals that matter
4. separate observation from inference
5. hand off to other skills if deeper evidence is needed

Release-Shell Execution Contract

• keep query briefs, raw trend payloads, normalized outputs, and watchlist caches under <work-folder>/.postplus/google-trends/
• keep only final user-facing summaries or shortlist exports outside .postplus/
• compile a small keyword or geo brief before the expensive collection step
• start with a bounded first pass:

- one topic cluster - one geo scope - one timeframe comparison

• if hosted capability is unavailable, unauthorized, or returns a stable network error, stop immediately instead of switching to ad hoc shell glue

Good Output

Return:

• keyword or topic set
• observed trend signal
• timeframe
• geo scope
• strongest rising queries or related topics
• provisional implication
• missing layer

Good recommendation shapes:

• search interest is rising, but platform-content proof is still missing
• good US search signal, weak evidence for other markets
• topic is hot now, but looks news-driven rather than durable
• use these rising queries as seeds for TikTok or Instagram content scouting

Handoff

Escalate to platform research skills when search intent should be validated against actual content or commerce evidence:

• TikTok content heat or hook patterns -> skills/20-research/tiktok-research
• Instagram creator, account, or campaign scouting -> skills/20-research/instagram-account-research or skills/20-research/instagram-campaign-scout
• Amazon marketplace demand -> skills/20-research/amazon-research
• marketplace demand -> skills/20-research/amazon-research

Escalate to higher synthesis when the user is making a real business decision:

• cross-source sourcing or selection judgment -> skills/30-strategy/sourcing-selection

Failure Modes To Avoid

Do not:

• treat search spikes as proof that a product will sell
• confuse news-driven spikes with durable category demand
• skip geo and timeframe details when comparing terms
• answer Google Trends requests from generic web articles when a platform-data route is available