Why narrow targeting now holds Meta prospecting back, and how to set up audiences, creative and budgets so cold audiences turn into new customers.
Most Meta prospecting campaigns, the ones aimed at cold audiences who have never heard of you, fail for three reasons. The targeting is too narrow for Meta's system to find your buyers, the budget is split across so many ad sets that none of them gathers enough data to learn, and the creative is too generic to attract anyone in particular. The fix is broad targeting with a few firm limits, fewer and better-funded ad sets, and ads written for one specific buyer.
This matters because prospecting is where next quarter's customers come from. When cold campaigns disappoint, the usual reaction is to move money into retargeting, which looks efficient for a few weeks and then stalls: you keep paying to reach the same people who already know you, while the flow of new visitors dries up. Your cost per new customer climbs, even when the dashboard looks fine.
This article explains what changed, how to structure prospecting, how creative does the targeting, how to split budget between cold and warm audiences and how to judge the results. The example figures are round illustrations, so swap in your own.
Meta's delivery system decides, auction by auction, who is most likely to take the action you optimise for, such as a purchase or a lead. It bases that prediction on how people use Facebook and Instagram, how they respond to ads, and what your conversion data says about past buyers. The more people it may consider, the more chances it has to find the ones who will actually buy.
Stacked interests made sense when the system had less to go on, but today they often exclude your best prospects by design. Say you sell premium kitchen knives in Austria and combine interests in cooking, a few well-known chefs and high-end cookware brands, leaving an audience of 300,000 people. A 50-year-old who never followed a cooking page, but buys kitchen equipment through Instagram ads every few months, falls outside that audience, and Meta is not allowed to show her your ad.
Two further changes push the same way. Apple's App Tracking Transparency (ATT) rules, introduced with iOS 14.5 in 2021, cut what Meta can see of people's behaviour outside its own apps, so it leans more on conversion data and on how people react to the ads themselves. And with Advantage+ audience, Meta's automated audience option, most targeting inputs are treated as suggestions, while location, minimum age and the custom audiences you exclude stay as firm limits.
Broad targeting is a sensible default, not a rule to follow blindly. Narrow inputs still help in a few situations:
Under Advantage+ audience these inputs act as a head start rather than a fence. The system begins with your best guess and goes further once the conversion data shows it should.
Most cold campaigns need only one or two ad sets. Each extra ad set splits your budget and your conversion data, and Meta's long-standing guidance is that an ad set needs roughly 50 optimisation events in a week to leave the learning phase, the period when delivery is still unstable. Ads Manager marks ad sets that fall short as 'Learning limited'.
Here is how that plays out. Say your average cost per purchase is around €40 and you have €8,000 a month for prospecting. That buys about 200 purchases a month, roughly 46 a week: close to the guideline in a single ad set, but only about nine each if you spread it across five. A sensible starting structure looks like this:
Once two ad sets are running, move budget towards whichever brings new customers more cheaply. Resist adding an ad set for every new audience idea; test new ideas as new creative instead, as the next section explains.
When you stop narrowing the audience, the ad itself does the filtering. Meta reads what an ad says and shows, notices who stops, watches, clicks and buys, and finds more people like them. Generic creative attracts generic traffic; specific creative attracts the people it was written for.
Compare two openings for accounting software aimed at online shops. 'Simple accounting software' could be for anyone, so Meta will show it to anyone who tends to click on ads. 'VAT returns for your webshop in five EU countries' will be ignored by most people, and that is the point: the ones who stop are the ones you want.
On Meta today, your creative brief is your targeting brief. Write each ad for a single buyer and a single problem, and plan several genuinely different ideas rather than many small variations. Ten versions of the same image with different headlines give the system little new to work with; three distinct angles give it three different groups of people to find.
On Meta today, your creative brief is your targeting brief.
Useful angles include the problem in the buyer's own words, the product in use, proof from real customers, a comparison with the usual alternative and the founder explaining why the product exists. Make each angle in at least two formats, such as a short vertical video and a static image, in the language of the market where it runs. Replace ads when results fade and frequency, how often the same person has seen an ad, keeps rising, not on a fixed calendar.
Warm audiences, people who have visited your site, engaged with your posts or given you their email address, usually convert more cheaply, so it is tempting to move money towards them. But a warm audience can only absorb so much. Past a certain point, extra budget mostly shows the same people the same ads more often, while the cold audience that should refill the warm pool goes hungry.
Rather than picking a percentage, size retargeting from the audience you actually have. Say 20,000 people visited your site in the last 30 days and you are happy for each of them to see your ads about twice a week. That is roughly 160,000 impressions a month, which at an illustrative €10 per 1,000 impressions costs about €1,600. Spending €5,000 on the same audience would not buy you three times the sales, only three times the repetition.
The pool you can actually reach is smaller than your visitor count. In the EU many visitors decline advertising cookies, and Meta cannot match every remaining visitor to an account, so use the estimated audience size in Ads Manager rather than your analytics figure. Whatever your warm audience cannot sensibly absorb belongs in prospecting.
Then watch frequency and cost per result together. If retargeting ads reach the same people five or six times a week while cost per purchase rises, the warm pool is saturated and the extra money belongs with cold audiences. Advantage+ campaigns mix cold and warm people, which makes this harder to see; the next section shows how to check.
Prospecting campaigns are often cut, or kept, for the wrong reason. Return on ad spend (ROAS), the revenue Meta attributes to your ads divided by what you spent, includes people who would probably have bought anyway, such as existing customers and people already on their way to you. A prospecting campaign with a flattering ROAS may simply be selling to your regulars.
Say Ads Manager reports 100 purchases from a €5,000 prospecting campaign, a cost of €50 per purchase. If 40 of those buyers were existing customers, you paid €5,000 for 60 new ones, or about €83 each. That is the number to compare with what a new customer is worth to you.
Meta lets you define your existing customers and your engaged audience in the account's advertising settings. For sales campaigns, you can then split results into new, engaged and existing audiences and see how much of your spend reaches genuinely new people. Check the same thing in your own shop data or CRM, the customer relationship management system where you track customers and enquiries, which counts new customers whatever any platform claims.
Give cold campaigns time. People who meet you for the first time often buy days or weeks later, through another channel, so judge prospecting over at least four weeks and look at the trend in new customers across all channels before you decide.