AgentSkillsCN

ads-expert

精通Google Ads的专业知识。当用户就Google Ads策略、关键词、出价、广告系列或PPC营销展开讨论时,此技能将自动激活。它会调用经过深度研究的知识库,涵盖五因素关键词模型、质量得分的运作机制、匹配类型、地域定位策略,以及常见的常见误区。

SKILL.md
--- frontmatter
name: ads-expert
description: "Google Ads expert knowledge. Activates when the user discusses Google Ads strategy, keywords, bidding, campaigns, or PPC marketing. Loads the researched knowledge base covering the 5-factor keyword model, Quality Score economics, match types, location strategy, and common mistakes."
user-invocable: false

Google Ads Expert — Background Knowledge

Step 1: Load the knowledge base

Read the file ~/.claude/skills/google-ads/knowledge.md in full. This is a researched, source-backed reference covering:

  • The 5-factor keyword valuation model (intent, QS, unit economics, volume, competition)
  • Quality Score CPC impact data (QS 10 = −50% CPC, QS 1 = +400%)
  • Long-tail keyword conversion data (2.5–5× higher CVR, half the CPC)
  • Match type state of play (2025–2026: exact + broad, skip phrase)
  • Location strategy (campaign-per-city, reverse targeting)
  • Bidding strategy by phase (manual CPC → target CPA → target ROAS)
  • Wedding/local service benchmarks (UK CPCs, CVRs, case studies)
  • Common mistakes ranked by damage
  • Keyword research methodology (free tools)
  • Key source URLs for re-fetching when data is stale

Absorb all of it. This is the foundation for all Google Ads advice.

Step 2: Check staleness

Look at the research date at the top of the knowledge base. If the current date is more than 6 months after that date, fetch these URLs from section 10 of the knowledge file to check for updates:

  • WordStream benchmarks URL (CPC, CVR, CPA data changes yearly)
  • Search Engine Land match type article (match types are actively evolving)
  • Google's own bidding and QS documentation (Google updates these)

Report any significant changes to the user.

Step 3: Apply knowledge

When advising on Google Ads:

  • Never recommend a keyword strategy without grounding it in the 4-question filter (section 8: intent, ad relevance, landing page match, profitability)
  • Always calculate break-even CPA/ROAS before recommending budget or bids
  • Default to exact match at launch, broad match + Smart Bidding after 30 conversions
  • Skip phrase match unless the user has a specific reason
  • Recommend separate campaigns per city, never one mega-campaign with bid modifiers
  • Flag common mistakes proactively (auto-apply recommendations, missing negatives, broken tracking)
  • When uncertain about current Google Ads features or benchmarks, fetch the relevant source URL from section 10 rather than guessing

Available action skills

If the user needs a structured workflow, these skills are available:

  • /ads-plan — full campaign strategy from scratch
  • /ads-keywords — keyword research and evaluation
  • /ads-copy — ad copy, headlines, extensions
  • /ads-audit — campaign performance review