You run one interest audience across the whole catalog. Campaign ROAS looks acceptable, so it stays on. Then you break the same purchases out by product and the picture can change: the margin on one strong seller may be covering SKUs that lose money on every order. The audience was never the whole story. For an online store, an interest is only as good as the margin of the product behind it.
Why does one interest audience work for one SKU and lose money on another?
An interest audience is priced per result, not per product. The same audience can be profitable on a product with a wide gross margin and unprofitable on a low margin SKU at the same cost per purchase. The audience did not change. The math around it did.
Before judging any audience, calculate a break even cost per purchase for each product you advertise:
- Selling price minus cost of goods.
- Minus shipping and packaging you absorb.
- Minus payment processing and platform fees.
- Minus your expected return rate on that SKU.
What is left is the maximum you can pay for a purchase before the sale costs you money. An audience delivering purchases at a cost that clears the ceiling on one product and breaks it on another is not a bad audience. It is a mismatch between audience cost and product economics, and it only shows up when you look at the product level.
Should my store use interest targeting at all if I already have pixel data and a customer list?
Interest targeting is a choice for an online store, not an obligation. You already hold purchase data, a pixel, an email list and order history. Blind interest targeting means you decided to ignore that. Use interests to reach people your own data cannot reach, and check the two side by side.
Wilco de Kreij opens his video on interest targeting by pointing out that you can also target custom audiences, meaning "people who are already a customer of you". For a store, that group is not theoretical. It is your orders table.
The difference matters because of how the platform builds an interest in the first place. As he explains, "Facebook uses a lot of different data to decide who is inside of a certain interest", which means the audience is broader and less defined than the page name suggests.
So the honest comparison for a store owner is this: a list of people who paid you money, against a label whose membership rules you cannot inspect. Run both. Let the cost per purchase, measured per product, decide the budget split.
How do I find out which interest is actually losing money in my store?
Diagnose at product and margin level, not campaign level. Split one interest per ad set, then match each purchase back to the SKU and its contribution margin in your store back office. A campaign ROAS that looks acceptable can be one profitable product carrying several that lose money.
A workable diagnostic sequence for a catalog:
- Pick the products that carry most of your gross profit, not most of your revenue.
- Give each interest its own ad set so results are attributable.
- Export purchases with SKU, not just purchase count.
- Apply the break even cost per purchase of that SKU, not a blended target.
- Kill or keep at the intersection of interest and product, never at the interest alone.
De Kreij describes the same blindness at the campaign level when advertisers stack many interests into one ad set: "Some may be profitable, some may be losing money". In a store, that sentence applies twice over, once across interests and once across products.
He is also blunt about why people skip the test. Building one ad set per interest gets expensive fast, and he uses the example that "even if it's just $5 a day \* 30 is still $150 a day just for trying it out". For a store owner the budget question has a specific answer: compare the cost of the test against the contribution margin of the products it will inform. If the test costs more over its run than the contribution margin those SKUs produce in the same period, test fewer interests on fewer products.
Why do my costs go up on the obvious interests during peak season?
The obvious interest for your category is also the one competing stores are most likely to pick when they raise budgets. Wilco de Kreij describes the fight over obvious interests as a bidding war. If your cost per purchase climbs in your busiest weeks while creative and offer hold steady, audience crowding is worth checking first.
This is the mechanism he describes when advertisers go for the most obvious interest. In his words, "you're basically fighting against everyone else on the Facebook ad platform", which he calls a bidding war.
There is a second problem with obvious targets, and it is about precision rather than price. He notes that "the more famous they are, the chances are the audience is not as targeted", and earlier that when you pick a broad interest, "there's absolutely no way how you can be certain" who ends up seeing the ad.
For a store, that combination is expensive in a specific way. You pay peak season prices for reach that includes people who will never buy a physical product in your category. Check whether your cost per purchase and your return rate both move in the wrong direction at the same time. That pattern points at audience quality, not at creative fatigue.
How do I find interests my competitors are not bidding on for my catalog?
Start with your own order data. Look at what your repeat customers bought second, and which product pairs show up in the same basket. Those adjacent categories point to brands and public figures outside your obvious interest list, and you test them against your current cost per purchase.
De Kreij frames the same idea as a question to ask yourself, "what else does your ideal customer buy", and then as a direction to move in: "try to like look sideways and see if there's like related markets that serve the exact same customer as you do".
The store advantage is that you do not have to guess at the sideways market. Your basket and reorder data already name it.
Three ways to turn that into targets:
- Brands your customers already own. Pull them from post purchase survey answers or product reviews, then look for those brands as interests.
- The second purchase. Whatever repeat buyers order next tells you the neighboring category before any keyword tool does.
- Layering. Instead of reaching everyone in either audience, target the overlap. De Kreij explains it as "I only want to show my Facebook ads to people who are inside of the Coca-Cola interest as well as the Pepsi interest".
He also points out that the keyword search inside Ads Manager returns a short list, "you're probably going to get like 20 results", which is worth knowing before you conclude that your category has no depth. The visible list is a slice, not the map.
Which tools help an online store test and diagnose interest targeting?
No single tool covers audience testing, creative production and margin math. Ad platforms hold the auction data, design tools handle the assets, and your store back office holds the margin. The table below maps what each one does for a store owner and how much advertising knowledge it assumes.
| Tool | What it does | What it solves for an online store | Requires advertising expertise |
|---|---|---|---|
| Meta Ads Manager | Native platform to build campaigns, ad sets and audiences on Facebook and Instagram | Where one interest per ad set is actually set up and where the interest level reporting lives | Yes. Ad set structure, audience selection and reporting are manual |
| Google Ads | Search, Shopping and display campaigns, including product feed campaigns | Captures demand from people already searching for your product, so it complements interest testing | Yes. Feed setup, keywords and bidding are manual |
| Canva | Design tool for static and video creative | Produces product creative variants so an audience test is not confounded by one tired image | No for basic design work |
| Mailchimp | Email marketing and audience segmentation | Turns your existing customer list into segments you can compare against interest audiences | No for standard email sends |
| SaleADS.ai | AI software that creates and launches advertising campaigns on Meta, Google and TikTok for business owners, with no design or advertising expertise required | Creates and launches the campaigns themselves across the three platforms. The break even math by SKU still comes from your own store back office | No, per the product's own description |
Disclosure: SaleADS.ai is the product of the company that publishes this site.
Where does the information in this article come from?
The advertising concepts in this article come from "5 Mistakes Every Facebook Advertisers Makes with Interest-Based Targeting" by the Wilco de Kreij channel. From that video we took his explanation of how an interest is assembled, why famous targets dilute an audience, why going for obvious interests puts you in a bidding war, why untested interest stacks hide losses, the idea of looking sideways into related markets, and how layering targets the overlap between interests. All quotes are linked to the exact second.
The application to online stores, including break even cost per purchase by SKU, product level diagnosis and the use of basket and reorder data to source adjacent interests, is our own and is not part of the video.