Facebook Ads CPA

Meta publishes no official CPA benchmark. Ads Manager reports cost per result, where the result is whatever objective you chose: a lead, a purchase, a view. Two accounts' "CPAs" are only comparable if their objectives, events and attribution settings match.

Pricing: auction-based; cost-per-result varies by objective, audience, placement and ad quality. meta recommends starting with at least usd 5 budget over 6+ days; no rate card.▸Meta · checked 2026 The $5 figure is a starting-budget floor, not a rate. Definitionally, cost per result is total spend divided by results▸Meta Help · checked 2026. That is the whole first-party numeric universe: a floor and a formula. Verified 2026-08-01.

• Purchase objective with a purchase event: cost per result is a real CPA.
• Lead objective: cost per result is a CPL; multiply through your close rate for effective CPA (the math).
• Engagement or traffic objectives: cost per result prices actions with no acquisition claim at all (action vs acquisition).

Published Facebook-ads benchmark surveys exist (WordStream has historically published one), but the widely-quoted per-industry cost-per-action tables date to pre-iOS-14.5 collection and are stale enough that we do not reprint their numbers here. The current LocaliQ/WordStream benchmark publication with a 2026 vintage covers search advertising, not Meta▸LocaliQ · 2026 survey. When a fresh vintage-labelled Meta survey appears, it enters the register and this page. Until then, your own account is the only trustworthy Meta benchmark.

App Tracking Transparency cut event coverage; Meta backfills with modelled conversions. Consequence: reported cost per result can move because modelling changed, not because performance did. Anchor on blended reality: total Meta spend against deduplicated new customers in your own backend, the blended CPA discipline. Decision aids: Google vs Meta and Meta vs TikTok.