What does this skill do, and when should you use it?
Audience Segment Builder solves the pre-campaign question of who to target. It reads a user's own customer, CRM, or GA4 exports and produces trait- or behavior-based seed audiences, value-based lookalike seed lists, suppression segments, and a platform-neutral funnel map. It also relates each bucket to the ROAS Audience dimension under direct-response, prospecting, or incremental-profit profiles. It does not build account structure, campaigns, or match types.
Reads user-supplied customer/CRM CSVs and GA4 audience or demographic exports, including available value or LTV, last-purchase date, plan or tier, source/medium, and fit signals; creates named audiences tied to source columns; ranks rows by the user's own value field to form high-value seed lists; defines exclusions for existing customers, recent purchasers, and bad-fit audiences; maps audiences to cold, engaged, intent, and customer funnel stages; produces a user-facing segment plan and handoff summary; and, after confirmation, saves results under memory/ad/audience-segment-builder/.
- An ecommerce team has a customer CSV and needs repeat-buyer and high-AOV seeds for Google and Meta.
- A growth team preparing prospecting wants a value-based lookalike seed from its highest-value customers.
- A direct-response advertiser needs to suppress existing customers and recent purchasers to reduce wasted impressions.
- A multi-platform team wants one reusable cold-to-hot funnel audience map across Google, Meta, and other platforms.
- A CRM or GA4 export lacks value or fit fields and the team needs explicit NEEDS_INPUT flags instead of invented segments.
How do you install this skill?
- Before processing CRM, customer-value, or GA4 exports, confirm lawful source, purpose limitation, retention, and required de-identification.
- Uploading seeds to platforms such as Google or Meta is external data processing; the skill lacks explicit per-upload confirmation, audit, recovery, or withdrawal procedures.
- Referenced shared files were not included in the supplied evidence, so actual compatibility, handoff format, and memory-write constraints require separate verification.
- The skill declares version 18.0.0 while the security policy supports only 17.x, so maintenance and security coverage should be confirmed.
- Local filesystem
Install the repository on an Agent Skills-compatible host with npx skills add aaron-he-zhu/aaron-marketing-skills. In Claude Code, run /plugin marketplace add aaron-he-zhu/aaron-marketing-skills, then /plugin install aaron-marketing@aaron. The repository also supports git clone https://github.com/aaron-he-zhu/aaron-marketing-skills. The skill is located at ad/research/audience-segment-builder/SKILL.md.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/aaron-he-zhu/aaron-marketing-skills.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/ad/research/audience-segment-builder" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Provide a customer/CRM CSV or GA4 export and state the profile and target platforms. Example: Build audience segments from my customer export: [path]. Goal is DR. Platforms: Google + Meta. Another supported prompt is: Make a value-based lookalike SEED list from my top customers and the exclusion list for people who already bought. [customer CSV] Do not use it to request account structure or match types.
What are this skill's strengths and limitations?
- Uses the user's own customer, CRM, and GA4 data and works by default without paid APIs.
- Covers seed audiences, value-based lookalike seeds, suppression segments, and a cross-platform funnel map.
- Requires segments to be grounded in exported columns and flags missing evidence as NEEDS_INPUT.
- Includes a clear privacy boundary: do not echo raw emails or phone numbers and treat exports as untrusted input.
- Requires a customer, CRM, or GA4 export; without one it can only identify missing inputs.
- Does not build ad accounts, campaigns, ad groups, or match types; those are assigned to campaign-architect.
- Produces the seed audience definition rather than a platform-specific lookalike key.
- The SKILL.md does not specify detailed field mappings or upload implementations for individual ad platforms; API upload is optional convenience.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Paid Ads Audience Segment Builder this page | 56 · Use with care | ★ 3k | 12d ago | Apache-2.0 |
| Ad Fatigue & Frequency Manager | 58 · Recommended | ★ 3k | 12d ago | Apache-2.0 |
| Paid Ads Conversion Signal QA | 56 · Use with care | ★ 3k | 12d ago | Apache-2.0 |
| Paid Search Term Miner | 54 · Use with care | ★ 3k | 12d ago | Apache-2.0 |
| Paid Ads Campaign Architect | 57 · Use with care | ★ 3k | 12d ago | Apache-2.0 |
Its boundary with campaign-architect is explicit: this skill defines the audiences first, while campaign-architect consumes them for account structure and match types. It is also distinct from keyword-research, which reads organic SERP intent rather than paid audience segments.
How did FollowSkills review this skill?
The skill treats exports as untrusted, prohibits echoing raw PII, limits optional platform APIs to uploading finished seeds, and requires user confirmation before saving. Deducted 7 points because it does not specify audience-data compliance grounds, deletion/rollback procedures, retention limits, or per-upload confirmation, and publisher identity is unverified.
Inputs, outputs, completion criteria, numbered steps, and scope boundaries are largely self-consistent; missing fields become NEEDS_INPUT rather than guesses. Deducted 2 points because referenced shared contracts, connectors, and reference files were not fully supplied here, and there is no skill-specific test or abnormal-input evidence; static review caps this at 10.
The audience, trigger phrases, input forms, four output buckets, non-fit boundaries, and platform-neutral requirement are clear, with Chinese-language support. Deducted 4 points for limited boundaries around regional privacy rules, platform differences, audience-size constraints, and GA4 export variants; geo relevance is low, but the core workflow does not depend on unreachable overseas services.
Front matter supplies name, version, license, compatibility, and argument hint; the document uses layered Quick Start, Contract, Instructions, and Handoff sections and identifies a next skill. Deducted 6 points for no skill-specific FAQ, changelog, named maintainer, or update path; the security policy supports 17.x while the skill declares 18.0.0, creating a governance inconsistency.
The skill specifies directly usable audience artifacts, naming rules, value ranking, exclusion windows, and missing-column handling, covering the core task well. Deducted 1 point because no representative real output or third-party outcome evidence is supplied, while cross-platform reuse, seed size, and field quality still require human review; static review caps this at 7.
The files provide auditable procedures, completion criteria, version metadata, and security constraints, while repository tests establish some shared validator and architecture controls. Deducted 1 point because tests do not cover this skill's key paths and there is no independent reproduction, corroborating third-party evidence, or verified sample audience plan; static review caps this at 5.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →