I Let Claude Run My Meta Ads For a Month and It Outperformed Every Experts Advice (The System)

How I filled my very first webinar with 335 registrants for $700

I have lit money on fire with Meta ads. More than once.

I did the thing you are probably doing. I boosted posts. I built these elaborate interest audiences stacked ten deep because a guru on YouTube said the money was in the targeting. I split my tiny budget across five ad sets so each “angle” had its own little home, then I sat there refreshing Ads Manager and turning things off the second they had a bad morning. I paid a lot to learn nothing, and I quietly decided ads just did not work for someone at my level.

This time I did something different. I let Claude run them. I stopped arguing with it, I stopped “improving” things at midnight, and I did exactly what it told me to do for a month.

Here is what happened.

The receipt

I ran ads for my free live class, the one where we build your custom Claude business dashboard together on the call. I spent about $700.

That $700 brought in 335 registrations.

That is roughly two dollars a registration. Two dollars. Every “realistic” number I had ever seen put a webinar registration somewhere between six and nine dollars, and that is on a good day with a warm audience. I had no big list feeding this. It was cold traffic and a two dollar seat.

I am telling you the number first because I want you to know the rest of this actually worked, and it worked by doing less, not more. Let me walk you through the five things Claude had me do that go against almost everything the experts sell you.

1. It refused to let me target anyone

The experts want you buried in audience settings. Interests, behaviors, lookalikes stacked on lookalikes. It feels productive. It feels like the work.

Claude let me set two things: age and gender. That is it. Women, the age range that fits my buyer, my country. No interests. No behaviors. Nothing clever.

The reasoning is boring and correct. When you have no list and a small budget, the algorithm is a better targeter than you are. Meta already knows who books calls and who buys courses. Your job is not to guess who those people are. Your job is to hand the algorithm a good signal and get out of its way. Every interest I would have added would have made the pool smaller and the learning slower. So we skipped it, and the algorithm went and found my people for two dollars a head.

2. It tested creative cheap before spending a cent on conversions

This is the move I wish someone had shown me two years and several hundred wasted dollars ago.

Everyone tells you to optimize for the conversion from day one. Run your ads for registrations, right out of the gate. The problem is registrations are “expensive” events, so Meta spends slowly and carefully while it figures out your creative, and you burn a week of budget just learning which video people will even stop scrolling for.

Claude flipped the order. First campaign, we optimized for landing page views. Cheap, fast, tons of data. Within a few days it was obvious which creative was pulling people in and which was dead weight, and I had barely spent anything to find out. Only then did we take the winners into a second campaign optimized for the actual registration. So I was never paying premium conversion prices to test a hook. I tested the hook for pennies, then spent the real money only on the ads that had already proven they work.

Test creative on cheap traffic. Scale winners on the real objective. Two campaigns, in that order.

3. It used ad set budgets, not campaign budgets

Small technical thing, big consequence.

The default advice, and honestly the setting Meta nudges you toward, is to put your budget at the campaign level and let it decide how to split the money. It sounds smart. In practice, when you are testing, it means the algorithm quietly starves the ads you actually want data on and dumps everything into one early favorite that might not be your real winner.

Claude had me set the budget at the ad set level while testing, so I controlled exactly how much each thing got to spend. Every creative got a fair, equal shot to prove itself instead of getting choked out on day one. When a clear winner showed up, I could feed it deliberately instead of hoping the campaign math landed there on its own. You want your hand on the dial while you are still learning what works.

4. It scaled the winner slowly, on purpose

Once I had an ad that was printing two dollar registrations, every instinct in my body said pour the whole budget in right now.

That instinct is how you kill a good ad. Big budget jumps throw the ad back into the learning phase, and it comes out the other side more expensive and worse. So we scaled it the boring way: raise the budget about twenty percent every two or three days, and leave it alone in between. Slow enough that the algorithm never panics, steady enough that the thing keeps growing. The ad that started at two dollars stayed near two dollars while it scaled, because I did not spook it.

Patience is the actual skill here. Not cleverness. Patience.

5. It told me to skip retargeting, and it was right

This is the one that surprised me most, because retargeting is supposed to be the easy money. Warm people, cheap clicks, free sales, that is the whole pitch.

I ran it. It did nothing. My retargeting registrations were not cheaper than my cold ones in any way that justified the setup, the extra campaign, or the attention. On a small budget, retargeting splits your money one more way and asks a tiny warm audience to somehow produce enough volume to matter. It cannot.

So we turned it off. And here is the better move Claude pointed me to: retarget with email instead. The people who hit my page and did not sign up, the ones who registered and went quiet, those are follow up emails, not ad dollars. Email is free, it is warmer, and it does the exact job people are overpaying Meta to do. Keep your ad budget on cold traffic where it actually earns, and let your inbox do the chasing.

Why this beat the experts

Here is the thing nobody wants to say out loud. Claude did not beat the experts because it knew some secret setting. It beat them because it would not let me do the expensive, complicated, impressive looking things that make you feel like a real marketer and quietly drain your account.

No fancy audiences. No day one conversion tax. No budget setting that steals from your own test. No scaling too fast. No retargeting a crowd too small to matter. It kept the whole thing consolidated and calm and let one good structure learn. That is it. That is the secret the million dollar accounts are actually running, sized down to a budget you and I can afford.

You do not need an agency. You do not need to be smarter. You need the structure, and the discipline to stop touching it.

Get the playbook

I put all of it into one thing: The Small-Budget Meta Ads Playbook. The exact campaign structure, the honest math, the numbers that tell you when to scale and when to kill an ad, and the three week runway I ran, all in a few pages you can follow start to finish.

It is free. Download it here.

Go get your seats filled.

x, Brielle