Stop the Roller-Coaster: The Jeremy Haynes Playbook for Crushing Volatility and Scaling Paid Ads to the Moon

Earnings Disclaimer: You have a .1% probability of hitting million dollar months according to the US Bureau of Labor Statistics. As stated by law, we can not and do not make any guarantees about your own ability to get results or earn any money with our ideas, information, programs or strategies. We don’t know you and, besides, your results in life are up to you. We’re here to help by giving you our greatest strategies to move you forward, faster. However, nothing on this page or any of our websites or emails is a promise or guarantee of future earnings. Any financial numbers referenced here, or on any of our sites or emails, are simply estimates or projections or past results, and should not be considered exact, actual or as a promise of potential earnings – all numbers are illustrative only.

---

Watch the full video breakdown on this topic here.

---

Key Takeaways (Read These, Tattoo Them, Live Them)

1. Every campaign is a living, breathing machine-learning model—treat it with respect or watch it flatline.
2. “Up-and-to-the-right” is the only metric that matters. Feed the model consistent, high-quality data or it dies.
3. Waves and buckets explain 99% of your turbulence. Understand them and you’ll sleep at night.
4. Budget isn’t just money; it’s throttle. The faster you burn, the faster you blast through buckets—good or bad.
5. Conversion API is legal mind-control. Withhold the wrong data, feed the right, and steer the algorithm where you want it.
6. Account simplification is gospel. Fewer campaigns, fewer ad sets, fewer ads = fatter profit.
7. Creative surgery can kill the patient. Randomly nuking ads, scaling like a maniac, or duplicating ad sets by the dozen fractures the model.
8. Stop blaming the trillion-dollar platform. If your swings are wild, it’s driver error. Learn the craft—or pay someone who did.

---

Table of Contents

1. The Cold Splash of Reality
2. The Bathroom Epiphany at Facebook HQ
3. Campaigns = Machine-Learning Models
4. Waves, Buckets, and Why Quality Tanks
5. Budget: Your Nitrous Oxide Button
6. Weaponizing the Conversion API
7. Creative Discipline & Scaling Sanity
8. Account Simplification: Less Is More Cash
9. Stop Whining—Start Winning
10. Need Speed? Where to Go Next

---

1. The Cold Splash of Reality

You’re burning through $1K, $10K, sometimes $50K+ per day on ads and still riding performance swings like a Six Flags coaster. One day you’re swimming in qualified leads; the next, you’re drowning in broke tire-kickers who ghost your sales team faster than a Tinder match after 11 p.m.

If that inconsistency makes you want to punch drywall, good—welcome to the club. But here’s the truth bomb: the volatility isn’t Facebook’s fault, Google’s fault, or Mercury-in-retrograde’s fault. It’s yours.

Why? Because no one handed you the real manual. They gave you platitudes like “just tweak the creative” or “raise budget slowly” while you watched CPMs spike and ROAS nosedive. Let’s fix that—permanently.

---

2. The Bathroom Epiphany at Facebook HQ

2019, Menlo Park—the mothership. Picture a campus designed by the same genius who built Disneyland: free food, arcades, and enough perks to make Willy Wonka jealous. But the real magic was hiding in the most unlikely place—the men’s bathroom of the engineering wing.

Mid-leak, I glance up and see a sheet of paper taped above the urinal. On it? A crude line graph climbing up and to the right with scribbles about machine learning, campaign performance, and account simplification. One urinal over, another sheet. Same on every stall. Facebook engineers literally wallpapered wisdom where you’d never miss it.

That sheet flipped my universe. It said, in plain English:

• “Every campaign = its own ML model.”
• “Feed it data, keep it simple, it thrives.”
• “Complicate it, starve it, it dies.”

I yanked the page, sprinted out, and dragged Corey—my Industry Ad Expert—into a conference room. (Regular ad reps? Mostly useless. Industry Ad Experts? Golden. Global Partner reps? Reserved for Pepsi and American Express. I’m not there—yet.)

Corey stared at me like, “Dude, you didn’t know?” Clearly, most advertisers didn’t. And that’s why you’re reading this instead of popping champagne over stable seven-figure months.

---

3. Campaigns = Machine-Learning Models

Let’s visualize. Time on the X-axis, Results on the Y-axis. The ideal trajectory is →↗. The model’s mission: collect conversions, learn who converts, find more of them, repeat.

But two laws govern its lifespan:

1. Data Momentum – Each new conversion fuels the next prediction. Starve it and growth stalls.
2. Probability Decay – As the pool of ideal prospects dries up, the model reaches into lower-probability audiences. Performance dips.

Think of the campaign as a shark—it must keep swimming (collecting data) or suffocate.

---

4. Waves, Buckets, and Why Quality Tanks

Corey broke it down with waves; I called them buckets. Same principle.

Wave/Bucket 1 – The Golden Few

At launch, the algorithm attacks a micro-segment of people most likely to convert—the 1% look-alike on steroids. Your cost per result is dreamy. Life is good.

Wave/Bucket 1.2, 1.3…

If conversions slow, it hops sideways to similar micro-segments, testing which one still hits gold.

Wave/Bucket 2 – Hot-Dog Lovers vs. Burger Fanatics

Run out of prime steak? The model grabs the next-best deli meat. These prospects share maybe 80 % of the original data points. They might love your offer—or they might be hot-dog junkies when you sell Wagyu burgers. Quality starts wobbling.

Wave/Bucket 3+ – The Fringe

Further out, the resemblance fades. Costs rise, unqualified leads proliferate, and your sales team threatens mutiny.

Lesson: When you notice lead quality dropping, you’re sliding down the wave. Panicking, slashing budgets, or nuking ads only drives the shark deeper into guppy territory.

---

5. Budget: Your Nitrous Oxide Button

Budget dictates how fast and how far the model barrels through buckets.

• $100/day to a 10 M audience – You touch maybe 5 K humans before frequency climbs past 2. The model lingers in Bucket 1 longer—great for testing, slow for scale.
• $10 K/day to the same 10 M – You blast a million eyeballs before hitting frequency 2. Fast scale, but you chew through Bucket 1 like a wood-chipper and hit Bucket 2 within days.

Neither is “wrong.” Just remember: big spend accelerates wave progression. If you don’t control data quality (next section), you’ll pay BMW prices for go-kart leads.

---

6. Weaponizing the Conversion API

Here’s where we go from competent to diabolical.

Problem:

The model thinks hot-dog lovers are perfect because they technically “convert” (they filled your lead form), but your closers know they’re broke time-wasters.

Solution:

Withhold bad data. Stop firing that standard pixel event the moment the form submits. Instead:

1. Strip the pixel event from the thank-you page.
2. Pipe conversions into the Conversion API only when the prospect shows up and passes your sales-qualified checklist.
3. Feed the model pure caviar data so it abandons the hot-dog crowd and chases burger aficionados again.

You’re not lying—you’re curating. Facebook even gave us the tool. Use it.

The algorithm desperately wants to keep rising ↗. When you starve it of bad signals and drip-feed premium ones, it course-corrects itself into richer veins of traffic.

---

7. Creative Discipline & Scaling Sanity

Three layers, one model:

1. Campaign – The brain.
2. Ad Set – Nerves.
3. Ad – Muscle fibers.

Slash the muscle mid-sprint and the athlete face-plants.

Creative Swaps

Ripping out half your ads and dropping shiny new ones without methodology creates a vertical cliff in the model’s line graph. It loses context, stalls, then nose-dives.

Fix:

• Introduce creatives incrementally.
• Kill losers after winners prove dominance.
• Let the model re-stitch itself before another surgery.

Scaling

Yes, you can 2×, 3×, even 10× budgets in certain scenarios. But spam-clicking the +25 % button every hour? Enjoy your upcoming crash. The model can’t sample new buckets fast enough to maintain efficiency.

Ad-Set Duplication Disease

Duplicating identical ad sets to “hack” delivery is like installing ten steering wheels in one car. Account limits exist because agencies abused duplication until performance cratered. If you duplicate, shut off the parent—or you’re splitting data and strangling both children.

---

8. Account Simplification: Less Is More Cash

Why does Dynamic Creative cap you at 10 visuals, 5 headlines, 5 primary texts—even if you drop half-a-mil a day? Because the probability of coherent learning plunges when you juggle 57 hooks, 32 videos, and a partridge in a pear tree.

Simplification Rules of Thumb

1. One objective per campaign.
2. As few ad sets as you can sleep at night with.
3. Stay within dynamic-creative limits if you’re manual—treat those limits as gospel.
4. Duplicate sparingly, then kill the source.
5. Audit monthly—purge legacy baggage.

Trim the fat and the model climbs like a SpaceX booster.

---

9. Stop Whining—Start Winning

Look, blaming a trillion-dollar platform’s “broken algorithm” is easier than admitting you never learned the real game. But here’s the scoreboard:

• Winners master waves, budget, data integrity, and simplification.
• Losers click random buttons, chase hacks, and weep on Twitter about CPMs.

If your performance swings leave bruises, it’s driver error—nothing more, nothing less. Learn, adapt, or keep donating tuition to Zuck.

---

10. Need Speed? Where to Go Next

You just devoured 3,700 words of field-tested strategy—free. Execute and watch volatility shrink, ROAS climb, and your CFO smile.

But maybe you value time more than heroic trial-and-error. Cool. I sell speed. My Inner Circle and Master Internet Marketing programs compress years of scars into weeks of clarity. No fluff, no gaslighting—just the playbooks, Q&A access, and accountability that turn decent operators into million-a-month machines.

Your call. Whether you join or DIY, feed the model, respect the waves, simplify relentlessly, and go get richer.