How Meta Actually Works in 2026

Meta is the most powerful advertising platform ever built. It also happens to be the most misunderstood. Most 7 and 8-figure brands are still managing their accounts like it's 2021, making decisions the algorithm has already made for them, and wondering why performance feels so unpredictable.

The reason is simple: Meta changed, and most brands didn't notice. In this episode of the eCommerce Playbook podcast, CTC's VP of Paid Media Tony Chopp walks through the CTC Canon for Meta advertising, the codified framework our profit engineers use every day to structure, manage, and scale Meta accounts in 2026.

Two Systems Running Everything: Andromeda and GEM

The foundation of modern Meta advertising is two relatively new systems that most advertisers have never heard of. The first is Andromeda, Meta's creative-first ad retrieval engine. The second is GEM, the Generative Evaluation Model, Meta's ranking intelligence layer operating at LLM scale.

Together, they process billions of interactions daily and make probabilistic allocation decisions that no human operator can replicate. And their existence explains something that has confused media buyers for years: why creative volume matters so much.

Andromeda evaluates creative attributes, user context, and behavioral signals to build a candidate set of eligible ads for every single impression opportunity. Meta's own engineering team describes it as enabling a 10,000x increase in model capacity, meaning the system can evaluate far more ad candidates per impression than was previously possible. GEM then determines what gets shown from that candidate set.

"These two technological engines are the reason for what is often colloquially referred to as the need for more creative into the Meta Ads ecosystem." — Tony Chopp, VP of Paid Media

The implication is significant. Meta is no longer an audience-first platform. It is a creative-first platform. Your targeting selections matter far less than the creative you put into the system. The algorithm finds the audience. Your job is to give it the best possible signal to work from.

The Breakdown Effect: Why Historical ROAS Is Unreliable

One of the most important and counterintuitive concepts in the CTC Canon is what Tony calls the breakdown effect. The core principle: historical ROAS does not predict future ROAS.

Meta states this explicitly in their own documentation. "The relationship between historical performance and future delivery is not deterministic." The probabilistic model Meta uses for future delivery is fundamentally disconnected from the backward-looking summary statistics you see in your dashboard.

This matters enormously for how you manage an account. When you look at yesterday's ROAS by ad and make decisions based on it, you are operating on incomplete information, using a small sample set of historical outcomes to predict a future that the system models far more accurately than any human can.

The practical consequence: stop making ad-level optimization decisions based on backward-looking dashboards. The system has already done that work. Your job is to set the right constraints and trust the allocation.

How CTC Structures a Meta Account

The CTC Canon uses a three-tier hierarchy that is intentionally simple. At the campaign level, the default is a sales objective with value optimization and a minimum ROAS target, using ASC or CBO to let Meta handle budget allocation with an inflated budget and cost control. At the ad set level, a seven-day click optimization window with broad targeting and acquisition/retention exclusions. ROAS targets live here, not at the campaign level. At the ad level, AI enhancements are used wherever brands are willing to test, including dynamic creative, creative enhancements, and multiple URL testing.

There are two types of campaigns in this structure. Evergreen campaigns run continuously and split between acquisition and retention. Each has ad sets that fill sequentially with new creative up to Meta's current 150-ad limit, plus a dedicated DABA ad set within acquisition. Marketing moment campaigns are time-bound, attached to sales, product launches, or any other event with a defined start and end date.

"Set the constraints, feed the machine, and let it work." — Tony Chopp

Budget Liquidity: Why Spend Variance Is a Feature

One of the most misunderstood concepts in Meta advertising is budget liquidity. The CTC Canon uses inflated budgets with cost controls deliberately, not despite the day-to-day spend variance it creates, but because of it.

Meta calls this liquidity: the system's ability to take advantage of moments of high-value inventory and pull back during expensive ones. By setting an inflated budget and enforcing profitability through minimum ROAS or a cost cap, you allow the algorithm to move freely across the time spectrum, spending aggressively when the market is cheap and naturally throttling when it is expensive. That variance is the system working correctly.

Signal Quality: The Foundation Everything Else Depends On

Before campaign structure, bidding strategy, or creative volume matter, the algorithm needs clean data. Signal quality is not optional infrastructure. It is the foundation.

The key components of high-quality signal in 2026:

• CAPI integration: Server-side events sent directly from your backend to Meta, bypassing ad blockers, cookie restrictions, and incomplete pixel fires.
• Event Match Quality score of 7/10 or higher: This measures how well your server events can be matched to Meta user profiles. Below this threshold, the algorithm is optimizing against degraded data.
• Purchase signal integrity: If Meta is only receiving 80% of your actual purchase events, it is optimizing on incomplete data. Performance suffers and measurement becomes unreliable. Audit this monthly at minimum.
• Proper audience definitions: Acquisition, retention, and existing customer audiences must be correctly configured at the account level. Meta's lifecycle settings at the campaign level are becoming increasingly important as the platform evolves.

For Shopify brands, the native Meta integration handles most of this automatically and is CTC's preferred CAPI setup. Custom or headless builds require dedicated engineering attention to get right.

Solving Underspend: A Four-Phase Framework

When delivery falls short, the CTC Canon has a defined sequence for diagnosing and fixing it. Work through these in order before adjusting anything else.

Phase 1: Creative expansion. Deploy additional creative into evergreen and active marketing moment campaigns. Prioritize diversity of concept, not minor variations. Big differences in creative approach drive delivery movement on Meta. This is almost always the right first answer.

Phase 2: Bid surface expansion. Duplicate evergreen campaigns into lowest cost, bid cap, or cost cap bidding using the same creative pool. A different bidding mechanism accesses a different audience without any additional segmentation.

Phase 3: Underspend ad recovery. Launch a dedicated ASC campaign with a single ad set, value optimized, same ROAS target as evergreen. Populate it with ads from evergreen where spend is less than 3x the CPA target. These are ads that have not had sufficient opportunity to scale. Give them another shot in a dedicated environment before writing them off.

Phase 4: Constraint relaxation. Expand beyond current efficiency limits only when business performance justifies it. This means adjusting your ROAS target or cost cap. This is the last lever, not the first.

The Most Important Principle in the Canon: Never Turn Off an Ad

This is the single most common mistake CTC sees brands make, and it runs counter to decades of media buying orthodoxy.

When you turn off an ad with poor historical ROAS, you are solving a problem the system has already solved. If you are running a CBO with cost controls, the algorithm has already reduced allocation to that ad based on signals you cannot see. Turning it off does not improve outcomes. It removes optionality from the machine.

The three responsibilities of a CTC profit engineer follow directly from everything above. First, ensure correct budget allocation with no limitations at the campaign level. Second, ensure the bid matches the business objective, meaning the ROAS target or cost cap reflects your actual unit economics. Third, identify and launch new creative. Feed the machine. That is the entire job.

Frequently Asked Questions

Ready to Run Meta the Right Way?

CTC's profit engineers manage Meta ad accounts for 7, 8, and 9-figure ecommerce brands using the Canon framework every day. If you want to see what that looks like for your brand, let's talk.

Talk to Us