What $1 Million in Facebook Ads Taught Us About Selling Courses

Everybody assumes a million dollars of ad spend buys you a secret setting. It doesn't. What actually works in Facebook ads for course creators is cheap testing, asking Facebook for buyers instead of signups, better creative than targeting, retargeting the warm leads you already paid for, and a funnel that finishes the job after the click.

The seven rules, and what ignoring each one costs you

Over two years our ad accounts paid Facebook close to a million dollars, and we have spent more than $3 million across every major platform since. Here is the whole list, so you can grade your own account before you read another word.

The rule	What most people do instead	What ignoring it costs
Test everything, cheaply	Run the version you like best	The ad you were sure about is usually the loser. The one we backed would have cost half the purchases.
Optimize for purchases, not leads	Optimize for leads because the number looks better	Cheaper leads, fewer sales. The worst trade in advertising.
Pick your budget level on purpose	Let Facebook allocate and hope	You never learn which audience actually worked.
Creative beats strategy	Rebuild the targeting for the fourth time	A perfect setup around a boring ad still loses.
Ignore "learning limited"	Panic and raise the budget	You burn a small audience out to chase a yellow dot.
Retarget warm leads	Send everyone to the same cold ad forever	You pay to find people, then let them walk.
Don't always blame the ad	Kill the campaign, blame Facebook	You delete the one thing that was working.

The $100 headline test that beat the ad we were certain about

There are do's and don'ts everywhere. Don't scale too fast, don't go too broad, use a lookalike, hook them in three seconds. What nobody tells you is that once you've done all of it correctly, you're still in a room throwing things at a wall. The polite word is testing, and that room is where the money actually is.

Here is the test that cured us of our own opinions. We ran the same ad with three headlines. After a week, headline two had the lowest cost per click, the most lead conversions and the most purchases. Headline one, the one we were confident about, would have brought in less than half the purchases. The entire test cost about a hundred dollars.

A hundred dollars to avoid running a campaign at half strength for a year. Change one thing at a time, give it a week, keep the winner, start again on the next variable. Headlines, hooks, thumbnails, images, copy, offer. We have been doing this for years and we are still wrong constantly.

The reason most people skip testing is not laziness, it's that building three landing pages to match three ads sounds like a weekend. It isn't in here. The funnel templates are already built inside your account, so duplicating a page and changing one headline takes about four minutes and your test budget goes to the test instead of to setup.

If you want the long version of how we structure a campaign from scratch, that's the seven-step breakdown. And the free one-hour training we made walks through the entire funnel these ads pointed at, top to bottom, at no cost.

Ask Facebook for buyers, not signups

This is still controversial with people running lead-gen ads, and we tested it until we were blue in the face. Every ad we run is optimized for purchases now.

Facebook knows people's shopping habits and it is extremely good at doing exactly what you ask. Tell it to find people who will sign up for a free webinar and it finds you the world's most enthusiastic free-webinar signers, at the lowest price on earth. Those are not the same humans who open a wallet. Across the board, optimizing for leads got us cheaper leads. Optimizing for purchases got us more sales. You pay more per lead for a lead worth having, which is the whole point.

We got an expensive reminder in 2026. We scaled Meta campaigns optimized to a checkout event rather than a completed purchase, and they grew to almost all of our spend. Trial volume roughly doubled and it looked like our best month ever.

Then the charges came due. Those trials converted to a paying customer at about 1 in 20, against roughly half for everything else in the account, and our cost per paying customer ran into the thousands on plans that cost a hundred and change a month. Non-Facebook traffic in the same weeks converted normally, so nothing was broken. We had asked for the wrong event and Facebook delivered it perfectly.

Pick the metric closest to money. We run our account on cost per lead, cost per new trial and cost per converted trial. Only the last one is real. If your software can't tell you which ad produced a customer who paid you twice, you're flying on the first number by default. Ours sit in the same account as the pages and the checkout, which is the change that moved our numbers most.

Campaign budget or ad set budget, and why we still choose by hand

Both work, and the gurus won't tell you that. The only real difference is how hands-on you want to be.

Say you have three audiences in three ad sets. With campaign budget optimization you set one daily number and Facebook decides who gets the money. Hands off, and a genuinely good way to test. What we can't live with is not knowing whether an ad set underperformed because the audience was bad or because Facebook never showed it to anyone.

So we use ad set budgets, spend the same on each audience, read the results ourselves, then move money on purpose. We ran a few isolated tests on this, nothing exhaustive, and the ad set Facebook chose to fund was also the winner in our manual splits. The machine is right often enough to trust with scale, and the reading you do by hand is what teaches you your own audience. Do it manually until you know your numbers cold, then let it run.

No targeting setting will save a boring ad

Creative beats strategy. In plain terms: make better ads. However much of a Facebook scientist you become, you hit a ceiling fast if the ad itself is no good, and almost everyone spends their week on the wrong side of that line.

The number one flaw we see in our students' ads is a weak hook. Facebook is an attention business. You get about three seconds, usually with no sound, to stop a thumb. If nobody stops, it does not matter how good the rest of the ad is or how clever your lookalike was.

We put a number on this in 2026. We spent $10,000 testing user-generated ads: seven hired creators, the identical funnel and lead magnet, nothing changing but the creative. Our own in-house ads were bringing trials in around $200 each. The best creator ads came in at $88. All seven combined averaged about $252, so as a group they lost to the ads we already had. That is the shape of ad testing. Most of what you make is worse than what you have, and the one that wins pays for all of it.

The cheapest creative you own is the content that already worked. We took an Instagram video people were commenting on organically, ran it as a Facebook ad, and got leads about 50% cheaper than the ad before it. Your audience already voted. Put money behind the winner. Meta's documentation on how ad delivery works is worth reading once, then never again. The rest of your time belongs to the ad.

"Learning limited" is not a diagnosis

You know the yellow dot. It means your ad set didn't get 50 conversions inside its learning week. Facebook's advice is to get more conversions, which mostly means give Facebook more money.

If you sell a high-ticket program, or you advertise to one city, 50 purchases in seven days is not happening, and spending more just exhausts a small audience faster. So do nothing. Your ad is still delivered the same way it would be with a green dot. Some of our best ads never went active at all and were returning 10 to 1. We aim for 3 or 4 to 1 on paid traffic, so a permanently yellow ad set at 10x was one of the best things in the account.

The only question worth asking is whether the ad is making you more money than it costs. Everything else on that screen is Facebook talking to itself.

The warm leads you already paid for

The rule of seven says the average buyer needs to see you around seven times. Most course creators spend their entire budget on impression number one.

We ran Facebook ads for years before we built a retargeting system that worked, and when we finally did it doubled our monthly revenue. Nothing about the cold ads changed. The mechanic is simple: take the warmest people, the ones who watched most of the video, opted in, DM'd a keyword or hit the sales page, and show them a small discount. Small enough that it doesn't eat your margin, big enough that passing it up feels dumb.

Where people stall is the plumbing. That offer needs a coupon, a checkout, an audience that updates itself and an email sequence behind it: four subscriptions in most people's stack, and four places for the handoff to break. On Premium and Elite it's one account, so the tag that says "watched 75% of the ad" is the same tag that triggers the email and the same list you push back to Facebook. Still the most under-used thing we teach, and the one that pays fastest.

When the ad is fine and the funnel is broken

Students come to us saying their Facebook ads aren't working. Usually the ad or the campaign setup is the culprit and we fix it there. Sometimes it isn't, and that is the case nobody enjoys hearing.

If your cost per click and cost per lead look good and no one is buying, the ad has done its job. Once someone clicks away from Facebook, it's your funnel's turn. Go walk it yourself, phone and desktop. Is the offer clear in one sentence. Is the value stack worth the price. Can a stranger find the button. We do this after $40 to $50 of spend on any new campaign, and the number of dead buttons and broken URLs we have found in our own funnels is embarrassing.

The two failures we see most are invisible from Ads Manager. First, tracking: the pixel fires on the form and calls it a purchase, so the numbers you're optimizing against are fiction. Second, email: people launch ads before the sending domain is verified, every lead they buy goes to spam, and the ads take the blame. That is why the onboarding call included with Premium and Elite sets up your domain, DNS and email deliverability before you spend a dollar, and why we wrote up what deliverability setup actually involves. If you'd rather someone else build it, our vetted experts do it for a flat fee.

Your first $500, in the order we would spend it

If you have never run an ad, don't start by trying to be clever. Start by buying information.

1. $0. Build the funnel from a template and walk it yourself. Buy your own product with a test card. Confirm the email lands in an inbox, not spam.
2. $100. One audience, one creative, three headlines, one week. Optimize for the purchase or the trial, never the opt-in. Keep the winner.
3. $150. Same winning headline, three different hooks in the first three seconds. This is where the biggest swings live.
4. $150. Turn on retargeting for anyone who watched, clicked or opted in, with a small discount.
5. $100. Put the rest behind whatever won, and start the next test.

That's five hundred dollars and about a month, and at the end you own a proven ad, a proven page and a retargeting loop. If Facebook is your first channel, read how it compares to YouTube ads before you scale, because the two reward completely different creative. If you want a human to look at your funnel first, book a walkthrough with our team and bring your ad account.

The free training goes deeper on the funnel side, and it's the full seven-figure funnel with nothing held back. Watch it, build the thing, then go buy your hundred dollars of information.

Frequently asked questions

How much should a course creator spend to test a Facebook ad?

About $100 per test, on one variable at a time, over roughly a week. That was enough for the headline test that doubled our purchases, and it is enough for you. Troubleshoot any new campaign after $40 to $50 of spend rather than waiting for the budget to run out.

Should I optimize Facebook ads for leads or purchases?

Purchases, almost always. Optimizing for leads gets you cheaper leads and fewer sales, because Facebook finds the people most willing to do exactly what you asked for. If your product is high ticket and you cannot get enough purchase events, optimize for the deepest event you can measure, then judge every campaign on cost per customer anyway.

Does "learning limited" mean my Facebook ad is broken?

No. It means the ad set didn't hit 50 conversions in a week, which is normal for high-ticket offers and small geographic audiences. Some of our best-returning ads never left that status. Judge the ad on whether it makes more than it costs.

Why are my Facebook ads getting leads but no sales?

Nine times out of ten the ad is fine and the funnel behind it isn't. Check the offer, the price, the page on a phone, and whether your emails land in inboxes. Broken buttons, unverified sending domains and a pixel counting form fills as purchases are the three we find most often.

Is retargeting worth it for a small course business?

It's the highest-return thing in this entire article. Retargeting the warmest audience with a small discount doubled our monthly revenue, and you have already paid to find those people once. If your budget is tiny, run retargeting before you run anything else.

Do I need a separate tool for landing pages, email and checkout to run ads?

No, and having them separate is where most ad money leaks. Course Creator 360 keeps the pages, the checkout, the CRM, the email and the automations in one account, so you can see which ad produced a customer instead of guessing across four dashboards. Premium and Elite come with a 30-day trial and an onboarding call that sets up your domain and email before you spend on traffic, so start there and get the funnel right first.

Related

• How to Scale Ads to Seven Figures: My 7-Step Loop
• Facebook Ads vs YouTube Ads for Course Creators: Which Wins?
• The Facebook Ads Tip That Changed Our Results
• GoHighLevel Email Deliverability vs Course Creator 360: Why Your Emails Land in Spam