← Catalog
skill Marketing

Lookalike Audience Plan

lookalike-audience-plan

Designs lookalike audience strategies with source audience selection, percentage tiers, and testing framework. Use when planning paid ad targeting to find new customers similar to your best existing ones.

Add this skill
  1. This skill, packaged and ready to upload. lookalike-audience-plan.zip
  2. In claude.ai or Claude desktop: Customize → Skills (+) → Create skill → Upload a skill, select the zip and toggle it on. Greyed out? Enable code execution under Settings → Capabilities.
  3. It’s live in your chats — no code, no setup. Want every Marketing skill at once? Add the whole plugin from the Marketing page (Customize → Personal plugins → Create plugin → Upload plugin).

When to Use This Skill

Use this skill when you need to:

  • Build a lookalike audience strategy for Facebook, Google, or other ad platforms
  • Select the best source audiences from your existing customer data
  • Plan percentage tiers and testing sequences for lookalike expansion
  • Create a structured rollout plan for scaling ad spend with lookalikes

DO NOT use this skill for interest-based targeting, retargeting setup, or organic audience building. This is specifically for lookalike/similar audience strategies on paid platforms.


Core Principle

THE QUALITY OF A LOOKALIKE AUDIENCE IS ONLY AS GOOD AS THE SOURCE AUDIENCE — START WITH YOUR HIGHEST-VALUE CUSTOMERS, NOT YOUR LARGEST LIST.


Phase 1: Source Audit

Before building lookalikes, identify and evaluate available source audiences.

Required Inputs

Input What to Ask Default
Ad platform "Which platform? (Meta, Google, TikTok, LinkedIn)" Meta (Facebook/Instagram)
Available customer data "What customer lists or pixel data do you have? (email lists, purchasers, leads, website visitors)" No default — must be provided
Average order value / LTV "What is your average customer value or lifetime value?" Unknown
Monthly ad budget "What is your monthly ad spend budget?" $2,000/month
Geographic targets "Which countries or regions are you targeting?" United States

GATE: Do not proceed until the user confirms their available data sources and platform.


Phase 2: Source Audience Strategy

Rank and recommend source audiences based on quality signals.

Source Audience Hierarchy (Best to Weakest)

  1. Top 25% customers by LTV — highest value, clearest signal
  2. All purchasers — proven buyers, strong signal
  3. Repeat purchasers — loyalty signal, smaller but potent
  4. High-intent leads — booked calls, started checkout, requested demos
  5. Email subscribers (engaged) — opened/clicked in last 90 days
  6. All email subscribers — weaker signal, larger pool
  7. Website visitors (key pages) — pricing page, product pages
  8. All website visitors — weakest signal, largest pool

Minimum Source Size Requirements

Platform Minimum Source Recommended Source
Meta 100 people 1,000-5,000
Google 100 people 1,000+
TikTok 100 people 1,000+
LinkedIn 300 people 1,000+

Present recommended source audiences with rationale before proceeding.

GATE: Confirm source audience selections with the user.


Phase 3: Lookalike Build Plan

Design the tiered lookalike strategy with testing framework.

Percentage Tier Strategy

## Lookalike Tiers

### Tier 1: Precision (1-2%)
- Closest match to source audience
- Highest expected conversion rate
- Smallest reach, highest CPM
- Use for: Initial testing, limited budgets

### Tier 2: Balanced (3-5%)
- Good match with broader reach
- Strong conversion potential with scale
- Use for: Scaling after Tier 1 validation

### Tier 3: Expansion (6-10%)
- Broadest reach, weakest signal
- Lower conversion rate but lowest CPM
- Use for: Top-of-funnel awareness, large budgets

Testing Sequence

  1. Start with best source audience at 1% lookalike
  2. Test 1% vs 3% vs 5% of the same source
  3. Test winning percentage across different source audiences
  4. Layer interest targeting on top of broader lookalikes (5%+)
  5. Exclude existing customers and active retargeting audiences

Budget Allocation

Phase Budget Split Duration
Testing 70% Tier 1, 20% Tier 2, 10% Tier 3 2 weeks
Scaling 40% Tier 1, 40% Tier 2, 20% Tier 3 Ongoing

Phase 4: Deliverable

Output the complete lookalike audience plan document.

Plan Format

## Lookalike Audience Plan

**Platform:** [Platform]
**Source Audiences:** [List]
**Geographic Target:** [Regions]
**Monthly Budget:** [Amount]

### Source Audience Details
[For each source: name, size, quality score, upload instructions]

### Lookalike Build Matrix
| Source Audience | 1% | 3% | 5% | 10% |
|----------------|----|----|----|----|
| [Source 1] | Build | Build | Test later | Skip |

### Testing Calendar
Week 1-2: [Specific tests]
Week 3-4: [Optimization actions]
Month 2+: [Scaling plan]

### Exclusion Lists
[Audiences to exclude from each campaign]

### Success Metrics
[KPIs and benchmarks for each tier]

Example: E-commerce Store Selling Skincare

Source audiences available: 3,200 purchasers, 800 repeat purchasers, 12,000 email subscribers, 45,000 monthly website visitors.

Recommendation: Lead with repeat purchasers (800) as primary source at 1% — strongest signal despite smaller size. Test all purchasers (3,200) at 1% simultaneously. Skip email subscribers until purchaser lookalikes are validated.


Anti-Patterns

  • Using "all website visitors" as the primary source — too broad, weak signal. Start with purchasers or high-intent actions.
  • Testing all tiers simultaneously — burns budget without learning. Test sequentially.
  • Ignoring source audience freshness — a 3-year-old email list produces worse lookalikes than a 90-day purchaser list.
  • Skipping exclusions — always exclude existing customers and active retargeting pools.
  • Assuming one source fits all — different products may need different source audiences.

Recovery

  • Source audience too small: If under platform minimums, combine related audiences (e.g., merge purchasers with high-intent leads). Note the quality tradeoff.
  • No purchaser data: Use highest-intent available signal — email engaged subscribers or key page visitors. Recommend building a purchaser list for future use.
  • Multiple products/services: Create separate source audiences per product line. Do not mix unrelated customer types.
  • No existing customer data at all: This skill requires some data. Recommend running interest-based campaigns first to build a pixel audience of 500+ converters, then return to build lookalikes.

View source on GitHub →