Two winning ads aren’t a System. It’s a coin flip.

The 3 pillars 7-figure brands use to keep finding winners, plus why Meta’s new algorithm quietly killed the old way.

Today’s Playbook

Top 7-to-9-figure DTC brands all share one boring trait. They are not better at picking winning ads. They are better at running a system that keeps producing them. Three pillars do the work.

Pillar 1: Volume. When you test 10 creatives, 8 will fail. Not “8 might fail.” Eight will. The other 2 become your winners, which you scale and iterate on. That math is brutal but it is the math. A brand testing 40 creatives a month builds 80-100 winning concepts a year. A brand testing 5 has the same two winners they had in March and a worried founder.

The people posting “looking for help with [your service]” aren’t cold prospects. They’re warm leads waving a flag.

Pillar 2: Diversity. Five years ago you could ride one or two winning angles for months. Not anymore. Andromeda, Meta’s new retrieval engine, pre-screens tens of millions of active ads and narrows the pool to roughly 1,000 candidates per auction. It actively filters out ads that look too similar to each other. Translation: minor variations of the same creative now compete with themselves and lose. The algorithm rewards diversity across three layers.

• Format: UGC, founder-led, product demo, testimonial, lifestyle.
• Angle: pain, aspiration, social proof, education.
• Messaging: the actual hook, script, and copy that deliver the angle.

Meta’s own data science team says creative quality now drives 56% of campaign outcomes, more than targeting, budget, placement, and timing combined. Effective ad lifespan also shrank from 6-8 weeks to 2-4. So the volume problem isn’t theoretical. The clock is faster too.

Pillar 3: Analysis. This is where most brands quietly bleed. They kill ads too early because day one looked flat, or they keep losers running for weeks. The fix is two layers of metrics, read in the right order.

• Soft metrics first (does the creative work as a piece of content):
  • Hook rate (3-second video plays ÷ impressions). Strong is 30%+. Top 10% performers hit 45%. Below 25% the creative isn’t earning attention.
  • Hold rate (ThruPlays ÷ 3-second plays). Target 25%+.
  • CTR. Above 2% is strong for most consumer categories.
• Hard metrics second (does the creative make money):
  • CPA and ROAS against your target CAC. If an ad has burned 2-3x your target CAC and isn’t converting, kill it or diagnose. Don’t pray.

The order matters. Soft metrics tell you why a creative is winning or failing. Hard metrics tell you whether to scale or kill. Skip the soft layer and you’ll kill ads that have a great hook and a broken offer, when the fix is the offer.

One more thing the seed missed but I want to flag. ​Creative fatigue starts showing in soft metrics about 5-7 days before CPA spikes​. Watch hook rate weekly, not monthly, or you’ll always be one cycle behind.

The Deploy

Here’s a one-week sprint you can run with whatever team you have. Call it a Weekly Creative Sprint.

1. Monday: brief 10 ads. Two formats, two angles, mixed messaging. Write the briefs in a single doc with hook + script + visual cue. Founder or marketer signs off, no committee.
2. Tuesday-Wednesday: produce 10 ads. UGC creator, in-house editor, AI tool, doesn’t matter. Get 10 finished assets in 48 hours. Quality is “shippable,” not “perfect.” Perfect is the enemy of volume.
3. Thursday: launch all 10 in one ad set with broad targeting and a small daily budget (1-2x your target CAC per ad). Let Andromeda sort them.
4. Following Monday: pull soft metrics for all 10. Anything below 25% hook rate, kill. Anything 30%+ with promising hold rate, push more budget into.
5. Wednesday after launch: pull hard metrics on survivors. Anything past 2-3x target CAC with no conversions, kill. Winners (the 1-2 that worked) move into your iteration backlog: same hook, three new variations next week.

Pass or fail metric for the sprint itself: out of 10 creatives, you should expect 1-2 winners and 2-3 promising-but-need-work. If you get zero winners three sprints in a row, the problem isn’t volume. It’s that all 10 briefs are still the same idea wearing different costumes. Force diversity at the brief stage.

Operator Note

I’ve been on the buyer side of this conversation for a while. The honest tension every founder hits: the system above is correct, and almost nobody runs it themselves. Briefing 10 ads a week, producing them, reading soft metrics on Monday, killing fast, iterating winners. That is a full role, not a side task.

One account we run is a clean proof of how much that costs when it isn’t fixed. They came in shipping 4 to 5 creatives a week, all the same angle wearing different costumes. We pushed it to 20 a week with fresh angles layered in, then kept scaling spend behind the winners. ROAS moved from 2.2 to 7, and we are still scaling.

The brands that grow this year are not the ones who read this and nod. They are the ones who pick whether they’re going to run it in-house, hire for it, or outsource it, and then actually do that by Friday.

The Problem isn’t Getting Leads. It’s Getting them to Close.

That last paragraph is the real reason for today’s plug, so I won’t be cute about it.

The volume Andromeda now demands is the exact gap our ​lead-flow system closes.

The piece that matters for today’s issue: AI agents draft dozens of ad angles in the time most shops test two, a human picks what runs (nothing spends your money without sign-off), and the market kills the dull ones fast.

You get the creative diversity the algorithm rewards without burning out a founder or a one-person marketing team.

If your ads are running and the calendar still isn’t filling with people who can buy, start with a free ​leak audit​. No pitch, no obligation. Just the leak.

Find your growth leak -> ​ScaleOnSteroids​

🎯 NEXT STEPS

• Open your ad account and count last month’s live creatives. Under 20 and you don’t have a system, you have a habit.
• Brief 5 new ads for next week using two formats and two angles you have not tried. Not the full sprint, just the on-ramp.
• Book a 15-minute Monday slot to read hook rates on every live ad. Anything below 25%, kill.​​

Stay weird,

Vaibhav

P.S. Next week: the small set of forms and routing tweaks that move “qualified enquiry” rate by more than any new ad creative will. Cheap, fast, almost nobody does it.