Paid Ads Campaign Architect
Designs paid account structure and reduces internal bidding and paid-organic overlap.
The skill treats exports as untrusted input, requires confirmation before saving, and frames ad APIs as optional rather than required; shared security material describes non-persistent keys, memory controls, and read-only or explicit-mutation boundaries. Deducted 8 points because this skill does not spell out the full data flow for memory, hot-cache/open-loops promotion, sensitive ad-data handling, rollback, or confirmation for each external effect; publisher identity is also unverified.
The contract, done criteria, scope guard, and missing-input behavior are reasonably explicit, including NEEDS_INPUT when placement data is absent. Deducted 2 points because static review cannot reproduce key paths; the selected skill has no dedicated tests, abnormal-input examples, or diagnosable failure messages, and referenced shared components were not validated as part of this target.
Triggers, inputs, outputs, non-goals, and three typed profiles are clearly stated, with bilingual metadata and a keyless Tier-1 manual-export path. Deducted 4 points because geo relevance is marked low and mainland-China platform/network reachability is not addressed; platform-specific boundaries and non-fit cases such as missing exports or PMax-only data are under-specified.
Frontmatter, Quick Start, Skill Contract, Data Sources, Instructions, Scope guard, and Next Best Skill provide useful progressive structure; version, Apache-2.0 licensing, compatibility, and shared references are documented. Deducted 5 points because the file lacks representative outputs, FAQs, troubleshooting, named maintenance responsibility, and a clear update/change path; the security policy's supported-version statement may be inconsistent with skill version 18.0.0.
The intended outputs are concrete: campaign-type choice, grouping, match types, negatives, exclusions, paid-organic overlap, and an A-dimension score, which could produce a usable planning draft. Deducted 9 points because the supplied material contains no real output or independently verified result for this skill; plans still require human validation of platform policy, data quality, and incrementality, while budget, full RQS, and execution are delegated elsewhere.
The skill cites the ROAS benchmark, security policy, connector guidance, and shared contract, while repository CI and tests provide limited auditable context. Deducted 5 points because the supplied tests do not cover campaign-architect key paths or representative artifacts, and there is no third-party execution evidence, sample input/output, or corroborating source set; static review cannot establish runtime behavior.
- Do not treat the output as a validated account design; manually review platform policy, search-term quality, placement data, negative-keyword completeness, and true paid incrementality.
- Before using GA4, ad exports, or optional connectors, confirm data authorization, sensitive-field handling, and the exact confirmation and recovery process before memory writes.
- For mainland-China users, separately verify reachability of target ad platforms, transparency libraries, and any relevant connectors.
- The skill is version 18.0.0, while SECURITY.md still identifies 17.x as the current supported line; confirm maintenance and security-support status.
What does this skill do, and when should you use it?
Campaign Architect is for designing or restructuring a paid advertising account before launch. It selects among campaign types such as Search, PMax, and broad match, then maps ad groups or asset groups, targeting, match types, negatives, and exclusions. It also reviews paid-organic intent overlap and scores the ROAS A (Audience) dimension plus account structure. It does not compute the final RQS or allocate budget.
Reads the account goal, campaign and search-terms exports, audience or placement reports, and available GA4 traffic-acquisition exports; selects campaign types based on the goal, intent maturity, and creative or feed readiness; organizes ad groups or asset groups around single intent themes; defines match types, audience signals, negative keywords, cross-campaign negatives, placement exclusions, and audience exclusions; compares paid query themes with organic landing pages to flag possible cannibalization; confirms a direct-response, prospecting, or incremental-profit profile and emits the corresponding ROAS A-dimension score with structure notes; after user confirmation, saves results under memory/ad/campaign-architect/.
- A marketer is preparing a Google Ads or other paid account and must choose Search, PMax, or broad match.
- An ads manager has campaign and search-terms exports and needs to restructure ad groups, match types, and negative keywords.
- A growth team suspects multiple campaigns are bidding against the same intent and needs cross-campaign exclusions.
- SEO and paid teams need to check whether paid queries overlap with pages that already rank organically.
- A media lead wants a structure and ROAS A-dimension diagnostic before handing the account to the full auditor.
What are this skill's strengths and limitations?
- Covers campaign-type selection, ad-group or asset-group layout, targeting, match types, negatives, and exclusions.
- Clearly separates paid account structure from organic site architecture and checks paid-organic cannibalization.
- Supports three typed ROAS profiles and emits an A-dimension score with a standard handoff summary.
- Can operate from user-provided exports without requiring keyed advertising APIs.
- Scores only the ROAS A dimension and structure; it does not compute final RQS or perform the full account audit.
- Does not allocate budget; budget allocation belongs to budget-optimizer.
- A missing placement report prevents a default A1 pass and requires NEEDS_INPUT.
- The source does not provide an independent test suite, verified platform coverage, or live campaign results for this skill.
How do you install this skill?
Install the repository, then use this skill. In Claude Code: /plugin marketplace add aaron-he-zhu/aaron-marketing-skills, followed by /plugin install aaron-marketing@aaron. On an Agent Skills-compatible host: npx skills add aaron-he-zhu/aaron-marketing-skills. You may also run git clone https://github.com/aaron-he-zhu/aaron-marketing-skills.
How do you use this skill?
Example trigger: Plan the paid account structure for [goal] on [platforms]. Here is my exported campaign + search-terms report: [paste/path]. Another supported prompt is: Should this be Search, PMax, or broad match? Lay out ad groups and the negative-keyword list for [themes]. Provide the goal, themes, and current structure when available; if the placement report is missing, A1 must be marked NEEDS_INPUT.
How does this skill compare with similar options?
Unlike ad-account-auditor, this skill handles preliminary account structure and the A dimension, while the auditor handles full RQS and ROAS veto items. Unlike budget-optimizer, it does not allocate spend. Unlike site-structure-optimizer, it designs paid advertising structure rather than organic site architecture.