Free facebook audience targeting guide Topical Map Generator

A Custom Audience is a group built from your first‑party data (customer lists, website events, app events or offline activity) used to target or exclude known users. Use Custom Audiences when you want high intent targeting — retargeting, cross‑sell/up‑sell, or precise exclusion of existing customers to avoid wasted spend.

Lookalikes find people similar to a seed audience by leveraging Meta’s model across behavioral signals; you upload a high‑quality seed (customer lists, converters, high‑value events) and choose the likeness percentage. Seeds can be as small as 100, but best results come from 1,000+ high‑quality records or aggregated event-based seed sets (1,000–50,000) for stable model building.

Use interest or demographic targeting for early funnel testing where you lack first‑party data, for awareness campaigns, or to discover new audience pockets; switch to Custom/Lookalike once you have conversion data to scale and improve CPA. Interest targeting is also useful for niche creative tests where lookalikes are not yet available.

ATT and browser restrictions reduced deterministic match rates and attribution visibility, so email/phone hashed lists and server‑side (Conversions API) event ingestion are now essential to maintain match rates. Implement CAPI, prioritize high‑quality hashed PII seeds, diversify signals (offline, CRM, app events), and focus measurement on incrementality rather than raw last‑touch metrics.

Track CPA/ROAS for conversion audiences, relative lift versus a control (holdout) group for incrementality, and secondary metrics like frequency, CPM, and match rate. Use A/B tests with identical creative and budget across audiences plus a statistical holdout to isolate audience impact.

Use narrowly defined Custom Audiences for mid‑ and bottom‑funnel (site visitors by intent, cart abandoners, purchasers), lookalikes and broad interest sets for scale/acquisition, and layered exclusions (past purchasers, engaged converters) to prevent overlap. Maintain separate ad sets by audience type to control bidding and attribution.

Refresh frequency depends on volatility: transactional lists (purchases/refunds) typically refresh weekly to monthly, while high‑churn segments (cart abandoners, app active users) refresh daily or near‑real time via CAPI/webhooks. Rebuild lookalike seeds whenever you improve data quality or change your definition of high‑value users.

Hash PII locally (Facebook accepts hashed email/phone), secure user consent aligned with your privacy policy, maintain lawful basis for processing, and document data mapping and deletion processes for audits. For regulated verticals, use minimized data sets, encrypt transfers, and consult legal to avoid prohibited targeting (health, finance) and ensure proper user opt‑outs.

Effective sizes vary by objective: bottom‑funnel Custom Audiences can be small (thousands) for retargeting, while acquisition audiences should be hundreds of thousands to millions to avoid delivery bottlenecks; 1% lookalikes in a given country often produce audiences in the 100k–1M range depending on population. Monitor delivery and expand or narrow the percent/criteria if CPMs spike or reach falls.

Yes — you can export audience definitions to other ad systems when using hashed PII or shared user IDs, and you can replicate seed generation and lookalike strategies on Google or TikTok; however, model performance will vary by platform signals so validate with platform‑specific tests. Use a centralized audience catalog and consistent naming to ensure cross‑channel governance.