How We Structure a Full-Funnel Facebook Ads Account in 2026

With Meta's Advantage+ shopping campaigns maturing, the old TOF/MOF/BOF campaign-per-stage structure is largely obsolete. We now run a simplified two-layer structure: one broad Advantage+ campaign for scale, and one tightly controlled retargeting campaign for warm audiences.

This reduces account complexity, gives the algorithm more signal density per campaign, and — counterintuitively — improves overall ROAS versus fragmented legacy structures.

Why the Old Funnel Structure Broke

The classic TOF/MOF/BOF setup was built for a world where advertisers, not the algorithm, decided who saw which ad. Splitting budget across five or six campaigns made sense when targeting was manual and precise. It stops making sense once the algorithm needs volume — enough conversion events per campaign per week — to optimize effectively.

Fragment the account into six thin campaigns and you fragment the signal six ways too. Each one re-enters a partial learning phase every time its small budget fluctuates, and none of them ever accumulates enough data to exit it cleanly. Consolidation isn't a shortcut — it's what the current algorithm actually needs to perform.

Layer 1: The Broad Advantage+ Campaign

This is where the bulk of the budget lives — typically 70–80% of total spend. Targeting is left broad by design; Meta's own signal-matching consistently outperforms manual interest stacking once there's enough conversion volume feeding it.

Our job in this layer isn't targeting — it's guardrails and creative supply. We set cost caps or bid caps to keep efficiency in check, exclude existing customers where relevant, and feed the campaign a wide spread of creative concepts rather than variations on a single angle, since Advantage+ leans on creative diversity to find pockets of audience the algorithm wouldn't otherwise surface.

Layer 2: The Retargeting Layer

The second campaign stays small, deliberately. It covers warm audiences on tight windows — typically 7-day engagers, 14-day site visitors, and cart or checkout abandoners — with a hard frequency cap to avoid hammering the same few thousand people.

Creative here does different work than the broad campaign: testimonials, urgency-driven offers, and direct objection-handling rather than top-of-funnel storytelling. Someone who already visited the product page doesn't need to be reintroduced to the brand — they need a reason to finish the purchase.

Feeding Both Layers With Creative

A two-layer structure only works if creative supply keeps up with it. Advantage+ campaigns are hungry — we aim for a steady rotation of new concepts weekly, mixing UGC-style and polished formats, rather than optimizing one "hero" ad indefinitely. The retargeting layer needs far less volume, but it needs to be refreshed the moment frequency starts climbing.

Scaling Without Resetting Learning

The same discipline that applies to search advertising applies here: budget increases in the broad campaign are staged in increments rather than doubled overnight, and we avoid touching bid strategy or audience settings in the same week we're scaling spend. The retargeting campaign's budget moves in lockstep with top-of-funnel volume — scale it faster than the broad campaign feeds it warm traffic, and you're just bidding up the same small pool.

Measurement in a Two-Layer World

Ongoing platform-level signal loss makes the Conversions API a requirement, not an optimization — both layers need server-side events deduplicated against the pixel to give Meta's models the fullest picture available. We also treat reported ROAS as directional rather than absolute, and validate it periodically against holdout or incrementality testing rather than trusting platform attribution in isolation.

The result is an account that's simpler to manage, easier to diagnose when something breaks, and — because every dollar reinforces the same two signal pools instead of six fragmented ones — consistently more efficient than the funnel-stage structures most accounts were still running two years ago.