Meta Ads After Andromeda: The Playbook That Works Now

iOS 14.5 shipped in April 2021. Meta lost the ability to track most users across apps, the Pixel went from reliable to partially blind, and the buying playbook got rewritten in a matter of months. Since then, Meta has kept rewriting it: Advantage+ automation, the removal of detailed targeting exclusions, and a new ad retrieval system called Andromeda.

Yet many accounts are still run with the same creative briefs, audience structures, and reporting they used in 2019. The common patterns: a missing or broken Conversions API, stale audiences, creative that hasn't changed in months, and weekly "results" meetings that review vanity metrics instead of revenue.

The people running those accounts aren't lying to you. They're operating from a playbook that no longer describes the platform they're buying on.

What changed, and why it matters

Four structural shifts changed Meta ads between 2021 and 2026. Understanding them is the prerequisite for everything else.

iOS privacy and signal loss. After iOS 14.5, a large share of conversions became invisible to the browser Pixel. Those conversions didn't disappear. The platform just can't see them. If the platform can't see them, it can't optimize for them. Accounts running without server-side event recovery are optimizing against a partial picture.

Advantage+ and the end of manual targeting. Meta's delivery models improved to the point where broad targeting with strong creative usually beats micromanaged audiences. Meta pushed the point home in 2025 by removing detailed targeting exclusions from new campaigns, saying the change improved cost per conversion in its testing. You can still exclude custom audiences, such as existing clients, but interest-based exclusion layering is gone. The skill shifted from "who you target" to "what you show them."

Andromeda and creative diversity. Andromeda is the retrieval system that decides which ads are even considered for a given person, before the auction runs. It reads the creative itself, and Meta has said creative diversification is now the strongest lever for finding the right audiences. Ten small variations of the same image tend to be treated as one idea and compete with each other. Genuinely different concepts (different angles, formats, hooks, and people) are what widen your reach.

Creative velocity as the primary lever. Stale creative fatigues: frequency climbs, costs creep up, and performance decays. Meta reports this through ad relevance diagnostics, which rank your ad's quality, engagement rate, and conversion rate against ads competing for the same audience. (Quality Score is a Google Ads term. Meta doesn't use it.)

Each of these shifts reduced the value of the 2019 playbook and increased the value of a different set of capabilities.

Regulated industries play by different rules

If you advertise financial products or services, including investment advice, insurance, banking, or credit, Meta requires the Financial Products and Services Special Ad Category for ads that reach US audiences. That category limits targeting: age and gender can't be narrowed, location targeting has a minimum radius, lookalike audiences aren't available, detailed targeting options are restricted, and some customer-list audiences need certification or can't be used. Housing and employment ads have similar rules.

For financial advisors, that means the creative and the landing page do almost all of the targeting work. It also means every ad is marketing under the SEC Marketing Rule or FINRA's communications rules, so compliance reviews the copy, any testimonials, and any performance language before it runs.

Healthcare has its own constraint. Tracking pixels on pages that collect health information can expose protected data. Have compliance review which events fire, and from which pages, before you optimize on them.

The five signs your account is still on the old playbook

The Pixel is the only event source, and the Conversions API is missing or misconfigured. This is the fastest diagnostic. If Events Manager shows incomplete deduplication, or your server events aren't sending properly hashed customer information, your optimization signal is degraded. The algorithm is optimizing against a partial picture, and performance suffers predictably.

Audience strategy is stuck in 2019. Custom audiences are the same lists as six months ago. Lookalikes at 1%, 3%, 5%, and 10% run as separate ad sets. Old interest exclusions linger in legacy ad sets. Today this usually produces higher costs and worse reach than broad targeting with strong creative.

Creative hasn't been refreshed in 30+ days. The old playbook treated creative as a monthly campaign asset. The current platform treats it as a continuous testing variable. Accounts without a steady flow of genuinely new concepts are leaving performance on the table.

Reporting is ROAS-first, without pipeline context. ROAS is useful. But ROAS without knowing what happens to the lead after the click is like measuring a sales team by doors knocked. Lead-to-opportunity rate, close rate, and deal size matter more than the ROAS number in Ads Manager.

CRM handoff is broken or nonexistent. Leads land in a spreadsheet or an unmonitored inbox and sit there. No automated follow-up, no scoring, no routing. The ad spend generates leads that decay before anyone talks to them. That isn't a media buying problem. It's an infrastructure problem that shows up as "our ads stopped working."

What the current playbook looks like

These are the pieces to install first when building or rebuilding a Meta ads operation.

Server-side Conversions API, deduplicated and complete. Every lead, purchase, and key event is sent server-side with proper hashing, customer information parameters, and deduplication. The platform sees much more of the picture, and better signal is often the cheapest performance gain available before any creative or targeting change.

Broad targeting, structured around the objective. Most accounts need one or two prospecting campaigns and, where volume supports it, one retargeting campaign. Audience differentiation happens inside the creative, not the ad set.

A creative production pipeline, not campaign assets. A steady flow of conceptually distinct ads, AI-assisted production, a rule for cutting losers quickly, and a rule for scaling winners. Creative isn't a deliverable. It's a continuous process.

Funnel integration, not ad isolation. The ad is the first touch in a sequence that includes automated follow-up, qualification, CRM routing, and sales handoff. Optimize the metric that matters: cost per qualified opportunity or cost per new client.

Weekly operational reviews. Spend, creative performance, event signal health, and funnel conversion, every week. Monthly reviews are too slow for the platform's optimization cycle.

Read your own numbers

Here is an illustrative example, not a client result. An account spends $10,000 a month at $60 per lead, which produces about 167 leads.

• If 10% become qualified opportunities and 25% of those close, that is about 17 opportunities and 4 new clients.
• If faster, automated follow-up lifts lead-to-opportunity to 20% with the same close rate, that is about 33 opportunities and 8 new clients.

Same spend, same cost per lead, twice the clients. The ads didn't change. The handoff did. Run the same math on your own account before you decide the problem is the media buying.

Why many agencies can't make this shift

The old playbook was manual and relationship-driven. The current one is infrastructure-driven. It requires:

• Server-side integration and event debugging (engineering)
• High-volume, genuinely varied creative (creative operations, not just design)
• CRM and automation wiring (systems integration)
• Funnel analytics that connect spend to revenue (data infrastructure)

Shops built around media buyers with spreadsheets and monthly creative cycles need a different org chart to deliver this, even when they understand it.

Three diagnostics you can run this week

The Conversions API check. Open Events Manager, check your event sources, and review deduplication for your lead or purchase events. If server events are missing, misconfigured, or showing large volumes of undeduplicated events, start there.

The creative check. When did you last launch a genuinely new concept, not a new headline on the same image? If it was more than a few weeks ago, you're operating at a disadvantage.

The follow-up check. Submit a test lead through your own ad. Watch what happens in the next hour, the next day, and the next week. If the sequence is empty or manual, you're losing a meaningful share of the value your spend generates.

If two or more of these show problems, the account needs a rebuild, not optimization.

There's no gentle transition from a broken playbook

Optimizing inside a broken structure produces marginal gains. Rebuilding the event plumbing, the creative pipeline, the CRM handoff, and the reporting is a project, not a tweak. Expect a few weeks to install and a couple of months to optimize into full performance.

The platform changed. The question is whether your operation did.

To see where follow-up and intake are leaking value from your ad spend, take the AI Opportunity Score. If you want a direct review of your Meta operation, book a strategy call. We'll look at your Conversions API setup, creative pipeline, audience structure, and funnel integration, then tell you what's costing you money and what it takes to fix it.