Copy This Simple Meta Ads Strategy, It'll Blow Up Your Business

Core problem (why accounts “stop working”)

• Launching new Meta ads resets ad-level learning: new ads start with zero history/data, so Meta serves them to the warmest, most engaged audiences to build signal.
• While learning, existing ads lose access to those same warm audiences, so their performance can look worse temporarily.
• Many marketers mislabel this as ad fatigue; the speaker argues it’s usually an internal measurement artifact caused by competition for the same warm pool during learning.

The “doom cycle”

1. Operators see performance drop and turn off ads (often the ones sustaining learning/audience building).
2. Budget shifts to lower-funnel/less incremental placements, adding complexity and reducing stability.
3. Result: repeated weekly “day trading” and declining ability to scale.

Creative diversity vs. complexity

• The claim: “Creative diversity” should be higher-confidence delivery, not more volume, with less complexity.
• Meta performs better when the account provides a clear structure so it can reliably match message → person → timing.

---

Platform mechanics the system depends on

• Every feed post/ad competes as a “web page” for attention, with a score Meta calls Estimated Action Rate (EAR):
  • Higher EAR ⇒ more delivery, lower cost
  • Lower EAR ⇒ less delivery, higher cost
• Since new ads start at zero, EAR/learning behavior drives the observed performance swings and the doom cycle.

Personalization interpretation (CEO/CMO framing)

• Personalization is framed as a sequence of messaging across touchpoints, not making infinite creative variations for individuals.

---

Framework: “Olympic Rings” (5 concepts / 5 jobs)

Designed as a sequenced messaging system (team sport), not five random ads.

Rings (prospecting → closing)

• Ring 1 (Hello #1): Upper funnel to cold audiences
  • 1 angle, 1 way in, 1 reason to care (e.g., UGC, founder video, bold claim)
• Ring 2 (Hello #2): Upper funnel, different concept/format/angle
  • Captures those who didn’t respond to Ring 1
• Ring 3 (Hello #3): Upper funnel, third distinct concept
  • e.g., testimonial or demo
• Ring 4 (Close #1): Lower funnel logical second touch
  • Static images; direct offer, objection handling, urgency
• Ring 5 (Close #2): Lower funnel second close
  • Same objective as Ring 4, but aimed at a different warm segment/time in the sequence

Key principle

• Each ring has a job and is designed to reach different people at different journey points, so rings don’t cannibalize each other.

---

Playbook to implement within Meta: “3-2-2” Flexible Ads (“322”)

Instead of testing many standalone ads, the system uses one flexible ad containing multiple variants that share a learning pool.

322 build rules

• Choose Flexible Ad (replaces dynamic creatives for this workflow).
• Composition:
  • 3 creatives (all same format and same aspect ratio; either all videos or all images)
  • 2 primary texts (meaningfully different hooks/angles)
  • 2 headlines (meaningfully different value framing)
• Output: 12 combinations inside one ad (3×2×2).

How performance is evaluated

• You don’t see performance per permutation (intentional).
• The “unit” is the flexible ad as a whole: it either scales or it doesn’t.
• “Unicorn” signals:
  • If one permutation dominates, you’ll see it via inbox/comments and disproportionate engagement/reactions (described as a harvest signal).

Scale math (claimed)

• 5 total 322 ads (one per ring) ≈ creative equivalent of 60 ads
• But consolidated into just 5 data buckets → shared learning, less chaos, faster better decisions.

---

Complete account architecture: “Andromeda 1”

Campaign structure

• 1 campaign (CBO) using Campaign Budget Optimization
  • Contains 3 ad sets to reduce unnecessary complexity:
    • control
    • test ad set 1
    • test ad set 2

Ad sets

1. Control ad set (stable environment)
   • Starts with top 6–8 existing ads (per speaker)
   • Evolves over time into 4–6 unicorn post IDs covering all five rings
2. Test ad set 1
   • Runs ring 322 tests (find something better than control)
3. Test ad set 2
   • Parallel proving ground to run comparative testing (A/B-like structure for learning paths)

Ad-level setup

• Control ad set: up to 6 proven post IDs covering all 5 rings
• Test ad sets: 322 flexible ads in each
  • described as 12 combinations per 322 flexible ad
• Total active testing: up to 24 active combinations across two test ad sets, continuously compared against control.
• If the account can’t exit learning with multiple ad sets: collapse into 1 ad set with 4–8 ads total (same principle, fewer containers).

---

Operational process over time (3-step loop)

Step 1: Testing (no interference)

• Run the two test ad sets.
• Do not launch new ads; let the system converge and signals improve day by day.

Step 2: Harvest (optional, every ~3–4 months)

Triggered when:

• One 322 permutation becomes an “alpha predator” (disproportionate engagement/reactions)

Action:

• Find the winning ad in inbox/comments
• Copy the post ID from the URL timestamp
• Create a new ad in the control ad set using that post ID

Important guardrails:

• Do not turn off the 322 being harvested from.
• Expect the post ID not to perform identically standalone, because 322 benefits from shared data/diversity.

Step 3: Scale (budget increases based on a profitability rule)

Scale when:

• Your campaign is performing better than your target CPA.

Example rule:

• If Target CPA = $50 and last 7 days you’re at $45 (10% better),
• Scale by 5% or less.

Budget change mechanics:

• Changes happen at the campaign level, not ad set level.
• Temporary CPA fluctuations are expected due to ramping/insufficient incremental conversions right away.

---

Diagnostic rule to prevent random changes

Two questions that determine what to do when performance drops:

1. “Can I spend more money and stay profitable right now?”
   • If YES: do not touch anything else (no new ads, no re-structure, no extra testing). Just scale budget.
2. If NO, then find the weakest link:
   • Either:
     • a specific ring (1/2/3 prospecting) ad consuming disproportionate spend with worst results, or
     • an ad set across rings consuming the most spend but performing worst vs target
   • Fix: build new 322 ads only for that ring’s job, run them in the test ad sets.
   • One change, one test (scientific method).

Key warning:

• Most people kill tests too early—before the system stabilizes—so performance appears worse than it is.

“Only question that matters”

• Repeated KPI framing: “Can I spend more money tomorrow?”
• ROAS/CTR/hook rate are acknowledged, but the primary decision driver is the ability to scale profitably.

---

Concrete actionable recommendations (what to do next)

• Build the 5-ring Olympic Rings sequence (3 prospecting hellos → 2 closing touches).
• Implement each ring using one 322 flexible ad (3 creatives + 2 primaries + 2 headlines).
• Use Andromeda 1 structure:
  • 1 CBO campaign
  • 3 ad sets (control + test1 + test2)
  • control holds proven winners (unicorn post IDs)
• Run a strict operational loop:
  • Test without adding complexity
  • Harvest optionally every 3–4 months
  • Scale only when above target CPA using the 5% or less guideline
• When performance drops:
  • apply the two-question diagnostic
  • improve the single weakest ring via new 322s, not a full rebuild

---

Sources / presenters

• Alex Schultz (Meta CMO) — cited regarding personalization being a delivery/sequence problem rather than creative volume.
• Zuckerberg — referenced for personalization framing across feed touchpoints.
• Presenter/Speaker: the video’s host (not explicitly named in the provided subtitles).