Meta ad targeting for a store with no data yet
No pixel history, no customer list, no lookalikes. Here is how to build your first Meta targeting so it produces usable answers instead of noise.
Every guide to Meta targeting assumes you already have a pixel with months of history, a customer list to build lookalikes from, and enough conversion volume to test against. On day one you have none of that.
That is a real constraint, but it is not the disadvantage it feels like. Here is how to build targeting that works from zero.
Broad targeting is a legitimate answer now
The advice to layer narrow interests comes from an earlier era of the platform. Meta’s delivery system has since become much better at finding buyers inside a large pool, and detailed targeting has become correspondingly less necessary — and often actively worse, because narrow audiences cost more per impression and exhaust faster.
For a new store, a defensible starting audience is genuinely wide:
- Country, chosen because you can ship there affordably and your copy is in the right language.
- Age 18–65+, unless your product has a real legal or physical constraint.
- All genders, unless the product is genuinely gendered.
- No detailed targeting at all, or one broad interest.
Then let the creative do the targeting. This is the part people underestimate: an ad showing a specific product to a specific use case is a targeting mechanism. People self-select. The algorithm learns from who responds and goes to find more of them.
When to use interests anyway
Broad works well when your product has mass-market appeal and your creative clearly communicates who it is for. It works badly when your product is niche enough that a random sample of your country contains almost no buyers.
If you sell specialist climbing equipment, “everyone in Italy aged 18-65” is mostly wasted impressions and you should narrow. If you sell a nice water bottle, broad is fine.
When you do use interests, two rules save you money:
Use interests that exist. Meta’s interest list is finite and specific. “Sustainable living” may not be a targetable interest even though it sounds like one; typing a plausible phrase into the box and picking whatever autocompletes closest is how people end up targeting something unrelated. Check what you actually selected.
Do not stack them narrow. Combining three interests with AND logic produces an audience so small that your CPM triples and your ad set never leaves learning. Use OR logic across related interests and keep the resulting audience above roughly 500,000 people for most European markets.
Build the audiences you will want in three months, now
The assets that make targeting easy later have to be started early, and they cost nothing.
Install the pixel properly before you spend a euro. Not just page views — the standard events for view content, add to cart, initiate checkout and purchase. Every day you run ads without them is a day of data you cannot get back.
Set up the Conversions API as well as the pixel. Browser-side tracking loses a meaningful share of events to ad blockers and privacy settings. Server-side recovers a good part of it, and Shopify makes this comparatively easy to enable.
Start collecting emails immediately, even before you have anything to send. A customer list is the seed for the highest-quality lookalike you will ever build, and 200 real purchasers make a better seed than 20,000 site visitors.
The exclusions people forget
Two exclusions matter from the very beginning:
Exclude existing purchasers from prospecting campaigns, unless the product is genuinely repeat-purchase. Otherwise you are paying prospecting prices to reach people who already bought.
Exclude your retargeting pool from your prospecting ad sets. If you do not, your prospecting campaign and your retargeting campaign bid against each other in the same auction — and you pay the premium for the privilege.
What “not working” actually means at this stage
With no data, your first campaign is not really trying to make a profit. It is trying to answer questions, in this order:
- Does anyone click? (creative)
- Do the clickers add to cart? (product and page)
- Do the add-to-carts convert? (checkout and price)
A campaign that produces no sales but a 1.4% CTR and a healthy add-to-cart rate has told you something valuable: the ads work and the problem is downstream. A campaign with a 0.3% CTR has told you something different and equally valuable.
Neither is a failure. The failure is spending three weeks and being unable to answer any of the three questions, which is what happens when the budget was too thin to leave learning or the structure was too fragmented to read.
A word on protected attributes
Meta prohibits targeting based on protected characteristics, and certain categories — housing, employment, credit, and social or political issues — sit under Special Ad Categories with sharply restricted targeting: no detailed targeting on many attributes, limited geographic granularity, and no lookalikes of the usual kind.
If your product touches health, finance, alcohol, gambling or politics, check the category rules before you build the audience, not after your ads are rejected.
AdPilot’s campaign generator suggests interests that exist in Meta’s actual taxonomy, applies exclusions by default, and flags special ad categories automatically when the product falls into one.