Export your top customers by lifetime value, upload them as a custom audience to Meta or LinkedIn, and use the 1% lookalike option so the algorithm finds users resembling your best buyers rather than your full customer base. The narrow size keeps the seed signals tight enough to mirror what actually converts profitably.

1. 1

   Define the audience source: clean the seed list before uploading (CSV or pixel-based for 1% lookalike from high-ltv customers).

2. 2

   Set up a dedicated conversion event in your ad platform's pixel so you can attribute downstream signups, not just clicks.

3. 3

   Build the creative-set: 3-5 variants that share the value prop but differ in hook (problem-first, social-proof, demo-led).

4. 4

   Launch at $X / day with frequency-cap enforced; budget for at least 7 days before reading results.

5. 5

   Read by ROAS / CPL daily; kill bottom-quartile creatives at the 100-impression mark.

6. 6

   Scale winners by 20% / day to avoid breaking the learning phase, and refresh creative every 14 days to fight ad fatigue.

HubSpot publicly attributes meaningful CAC reduction to 1% lookalikes seeded from their highest-LTV cohort rather than broad-firmographic targeting. Mid-market SaaS teams running the same playbook routinely report 30-50% CPL drops vs. broad audiences in the same Meta or LinkedIn account.

• ✗Buying volume before the funnel converts — paid traffic into a broken funnel is just a fast way to burn cash.
• ✗Underbudgeting creative production; the platform is now 10% of the work, the creative is 90%.
• ✗Watch-out from the playbook: Refresh the seed list every 60-90 days as new top customers come in, and exclude existing customers from delivery so you do not pay to re-re

Brands with at least 100-500 paying customers and clean LTV data