Paid Social Audience Builder

This drill builds targeted audiences on LinkedIn and Meta for problem-aware paid social campaigns. Problem-aware prospects know they have a pain point but have not started evaluating solutions. Your audiences must reflect this: target by job function, seniority, and industry (proxy for having the problem), not by intent signals like competitor website visits (that is solution-aware).

Prerequisites

• ICP defined (run icp-definition drill if needed)
• LinkedIn Campaign Manager with Marketing API access
• Meta Business Manager with Marketing API access
• Clay account for enrichment (optional but recommended)
• Attio CRM with at least some closed-won customer data

Input

• ICP document: firmographic criteria (company size, industry, funding stage) and buyer persona (title, seniority, department)
• Closed-won customer list from Attio (for lookalike audiences on Meta)
• Any existing website visitor data in PostHog

Steps

1. Extract audience signals from your ICP

Pull your ICP criteria from the icp-definition drill output. Translate each criterion into platform-specific targeting parameters:

| ICP Criterion | LinkedIn Targeting | Meta Targeting | |---|---|---| | Company size 20-200 | companySize: SIZE_11_50, SIZE_51_200 | Interest-based (no direct firmographic) | | SaaS / Software industry | industry: SOFTWARE, INTERNET | Interest: "SaaS", "Software as a service" | | VP/Director Engineering | jobFunction: ENGINEERING, seniority: VP, DIRECTOR | Custom audience from Clay export | | Series A-C funded | Not directly available; use company size as proxy | Not available; use lookalike from customer list |

LinkedIn is stronger for B2B firmographic targeting. Meta is stronger for retargeting and lookalikes.

2. Build LinkedIn audiences

Using the linkedin-ads-audience-targeting fundamental, create 3 audience segments:

Segment A — Core ICP (tight):

• Job function + seniority + industry + company size
• Expected size: 20,000-80,000
• This is your highest-quality segment; bid highest here

Segment B — Adjacent ICP (broader):

• Same job function and seniority, broader industry or company size range
• Expected size: 80,000-200,000
• Tests whether adjacent markets respond

Segment C — Title-based (experimental):

• Specific job titles that your best customers hold (pull from Attio)
• Cross-reference with Clay using clay-people-search to validate these titles exist in volume on LinkedIn
• Expected size: varies

For all segments, exclude: your company's employees, existing customers (upload from Attio using attio-lists), and competitors.

3. Build Meta audiences

Using the meta-ads-audiences fundamental, create 3 audience types:

Custom Audience — Website visitors:

• Create from Meta Pixel data: visitors to your blog, landing pages, or product pages in the last 30 days who did NOT convert
• These people have shown interest but need another touch

Lookalike Audience — Best customers:

• Export your top 20% customers (by deal value or retention) from Attio
• Upload to Meta as a source audience (hash emails with SHA-256)
• Create 1% lookalike for quality, 3% for reach
• Apply geographic and age filters to refine

Interest-based Audience:

• Target interests related to the problem your product solves (not the solution category)
• Example: if you sell deploy automation, target "DevOps", "Continuous Integration", "Site Reliability Engineering" — these indicate someone who deals with the problem space
• Layer with demographics: age 25-55, manager+ roles

4. Score and validate audience quality

Before spending ad budget, validate that your audiences actually contain ICP-matching people:

• Using clay-people-search, sample 50 profiles from your LinkedIn audience criteria. Score them against your ICP using clay-scoring. If fewer than 60% score as ICP matches, tighten the audience.
• For Meta lookalikes, you cannot preview members. Instead, run a $50 test for 3 days and check the quality of leads that come in against your ICP criteria.

5. Set up audience exclusions

Critical to avoid wasted spend:

• Exclude current customers (Attio export uploaded to both platforms)
• Exclude recent converters (people who submitted a lead form in the last 30 days)
• Exclude disqualified leads (from Attio, if tagged)
• On LinkedIn: exclude your own employees and competitor employees by company name
• On Meta: exclude your custom audience of converters

Refresh exclusion lists weekly via n8n automation.

6. Plan audience rotation

Audiences fatigue. Plan a rotation schedule:

• Week 1-2: Run Segment A (core ICP) and Meta Lookalike 1%
• Week 3-4: Add Segment B (adjacent) and Meta Lookalike 3%
• Week 5-6: Test Segment C (title-based) and Meta interest-based
• After 6 weeks: retire exhausted audiences, build new ones based on what converted

Track audience-level performance in PostHog to identify which segments produce the best leads, not just the most clicks.

Output

• 3 LinkedIn audience segments configured in Campaign Manager
• 3 Meta audience types configured in Ads Manager
• Exclusion lists uploaded and scheduled for weekly refresh
• Audience rotation calendar for 6 weeks
• Validation data: ICP match rate per audience segment