Facebook ads for supplements: the rules, the setup and how to measure them

Meta's Advertising Standards treat supplements as a restricted category, not a banned one. The Health and Wellness policy sets three rules that matter most:

• Age targeting. Ads promoting dietary, health, weight loss or weight gain products must be targeted to people at least 18 years old. Set this in the ad set before you submit, or the ad can be rejected.
• No attacks on appearance. Ads can't contain terms, descriptions or questions that are negative about a person's appearance, specific body parts or hygiene. "Tired of your belly fat?" is the kind of line that breaks it.
• Weight loss imagery. Weight loss ads can't show a close-up of someone pinching fat. Ads aimed at adults can show people using the product and its effect, but must say clearly how long noticeable results took.

Meta also says it can reject or remove any ad at its discretion, and its policies change without notice. Read the current policy page before a launch rather than relying on last year's approvals.

Meta's rules decide whether an ad runs. US law decides whether its claims are allowed. The FTC's Health Products Compliance Guidance says claims about the health benefits or safety of dietary supplements generally need competent and reliable scientific evidence, and the kind of evidence depends on the claim.

The FTC also says anyone taking part in deceptive marketing can be liable, including ad agencies, retailers and endorsers, not only the brand. So the agency writing your hooks and the creator reading them both need to know what the product can and can't claim.

FDA labeling rules for structure/function claims are a separate system. The FTC says those labeling provisions don't govern how it judges the same claim in an ad. If your label can say something, check that your evidence also supports saying it in an ad.

A setup that passes review and leaves you something to measure:

1. Pick where buyers will purchase: your own store, your Amazon listing, or both. Most of what follows depends on it.
2. Set the audience to 18 and older in every ad set, including retargeting and lookalike audiences.
3. Write each claim from your evidence file, and keep that file with the ad. Leave out lines that put down how people look.
4. Put an Amazon Attribution tag on any link that goes to Amazon. The Facebook ads for Amazon guide covers tags and the Brand Referral Bonus.
5. Before scaling, size a holdout test with the holdout test calculator, so you know how long a test would need to run.

Supplement brands hit two blind spots at once.

• Amazon. Meta can't see purchases on Amazon. A shopper who sees your ad and buys on Amazon a week later shows up nowhere in Ads Manager, and only in Amazon Attribution if they clicked a tagged link.
• Reorders. For a product people reorder, the first order can be a small part of what a customer buys. Ads Manager reports the first purchase in its window; the reorders months later belong to no campaign. The subscription LTV calculator puts a number on them.

Reports can also over-credit. Retargeting ads get credit for loyal customers who were already coming back. The way to separate the two is a geo holdout test: pause Meta ads in some US states, keep them on in the rest, and compare orders by ship-to state.

lift = (ads-on orders ÷ holdout orders, during) ÷ (ads-on orders ÷ holdout orders, before) − 1

The full method, including the uncertainty range, is on how lift is calculated.

A made-up supplement brand sells on its own store and on Amazon. It spends $3,000 a month on Meta ads, and Ads Manager reports $2,400 of store purchases: a ROAS of 0.8. On that number, the ads lose money.

A geo holdout test finds the ads caused about $2,000 of store sales and $3,400 of Amazon sales, $5,400 in all. That's an incremental ROAS of 1.8. Most of the effect was on Amazon, where Meta couldn't see it.

Whether 1.8 pays depends on margin. With a 50% contribution margin before ads, the break-even ROAS is 2.0 on first orders, so this brand needs reorders to make the ads pay. That's a question the LTV numbers answer, not the ad report.

• Leaving age targeting open. Supplement ads must target people 18 or older, retargeting included.
• Borrowing claims from the label or a competitor. Each ad claim needs its own evidence.
• Judging the ads on Ads Manager alone. It can't see Amazon purchases or later reorders.
• Cutting spend because Amazon sales didn't jump overnight. Sales move with season, price and stock. Without a comparison group, a drop or a rise says little about the ads.

Spillmark runs geo holdout tests for brands that advertise on Meta and sell on Amazon. It plans the test from your own order history, locks the design before it starts, and measures the Amazon sales your Meta ads cause, with an uncertainty range. It takes off Amazon fees, your product costs and ad spend, so you see profit, not just ROAS.