Interest targeting used to be the main skill in Facebook advertising. Under Meta Andromeda, the ad itself does more of the targeting. This guide explains why broad audiences often beat narrow ones now, which settings still matter, and the cases where narrow targeting is still right.
Ask an experienced Facebook advertiser from a few years ago how they found customers, and they'd show you audiences: interests, behaviours, lookalikes, careful exclusions. Ask the same question today and many will say "broad". This isn't laziness. It reflects how Meta Andromeda and the models around it now work.
Why broad targeting works now
Interest targeting relies on a rough label: someone liked a page or engaged with a topic. Meta's systems now learn from a much richer picture of behaviour and from the content of each ad. Andromeda, in Meta's words, learns higher-order interactions between people and ads, and Meta's April 2025 guidance said creative diversification has replaced niche targeting as the best lever.
In practice, narrow targeting often removes good buyers the system would have found. A yoga mat brand targeting "yoga" misses the physiotherapy patient, the new runner and the person setting up a home gym. Broad targeting plus an ad about "a mat that doesn't slip when you sweat" can find all of them.
What you should still set
- Location. Where you sell, deliver or can serve. For local businesses, a realistic radius.
- Age, if it truly matters. For example 18+ for some products, or a minimum age for loans. Otherwise leave it wide.
- Language, if your ads are in a specific language and the audience spans many.
- Exclusions that save money, such as recent purchasers for a one-time product, or existing customers if you have a separate budget for them.
- Legal limits. Housing, employment and credit ads fall under special ad categories with restricted targeting.
Advantage+ audience: suggestions, not walls
With Advantage+ audience switched on, any interests, custom audiences or lookalikes you add become suggestions. Meta starts there and expands when it finds better results elsewhere. Your hard controls (location, minimum age, exclusions) still apply.
For most advertisers, this is the right default: you give the system a head start with your best customer lists, without capping it.
The creative is the new targeting
If broad targeting reaches "everyone", what decides who actually sees your ad? Largely, the ad. Its words, images, people and offer shape who it's retrieved for and who responds. That's why targeting and creative strategy have merged:
- Want students? Show students, speak to student problems, mention student prices.
- Want premium buyers? Show quality, craftsmanship and the full price confidently.
- Want parents? Show the child, the school bag, the morning rush.
- Want buyers in a city? Mention the city and show recognisable places.
See creative diversification for Andromeda for how to build a set of ads that reach different buyers.
When narrow targeting still makes sense
- Very small niches with low search and engagement signals, such as industrial B2B equipment, where broad delivery wastes early budget.
- Tiny budgets where you need early results from the most likely buyers before the system has data. Start with a lookalike of customers as a suggestion, and broaden as results come in.
- Strict geography, such as a clinic serving a few pin codes.
- Existing customers and [retargeting](/glossary/retargeting) for specific offers, like a launch for past buyers.
- Legal or brand reasons, such as alcohol-related or age-restricted products.
How to test broad against your old targeting
- Keep your current best ad set running.
- Create a new broad ad set with the same ads plus some new concepts, and a similar budget.
- Run both for at least one to two weeks, until each has a fair number of results.
- Compare cost per result and, for stores, delivered orders and ROAS, not just clicks.
- Move budget towards the winner gradually.
Quick quiz · 5 questions
Targeting under Andromeda: check yourself
1. With Advantage+ audience on, what happens to the interests you add?
2. A yoga mat brand targets only the 'yoga' interest. What's the main risk?
3. What mostly decides who sees your ad under broad targeting?
4. Which is a good reason to keep targeting narrow?
5. How should you compare broad against your old targeting?
0 of 5 answered
Targeting for different businesses
- Online stores: India-wide or delivery regions, broad, Advantage+ audience, customer list as a suggestion. See Andromeda for online stores.
- Real estate: the cities where buyers actually live (which may differ from the project location), broad within housing ad rules. See real estate ads in India.
- Clinics and local services: a tight radius, broad within it. See enquiries and calls.
- Coaches and institutes: broad, with creative that clearly speaks to students, parents or professionals. See ads for coaching institutes.
Worked example: a bakery in Pune
An illustrative case. A bakery with two outlets used to target women 25 to 45 interested in baking and desserts, within 5 km of each outlet. Results were steady but expensive. The broad version kept the 5 km radius and an 18 to 65 age range, switched on Advantage+ audience, and changed the ads instead: a birthday cake Reel for parents, an office treat static for workers, a festive hamper carousel for gift buyers, and a "forgot the anniversary?" ad for last-minute couples.
The radius did the essential targeting. The ads did the rest, each pulling in a different group the old interest settings had excluded, such as men ordering for partners and office managers ordering for teams.
Exclusions done right
- Exclude recent purchasers of one-time products, using a customer list or purchase audience, for example the last 30 to 90 days.
- Exclude employees and partners if your team is large enough to waste spend.
- Don't exclude broadly (for example everyone who engaged with your page), as you'll remove warm buyers.
- Review exclusions monthly, because lists go stale.
Using your own customer data
Your customer list is one of the most useful signals you can give Meta: as a suggestion for Advantage+ audience, a source for lookalikes, or a way to identify existing customers. Upload only data you're allowed to use, keep it to what's needed, and mention advertising tools in your privacy policy. India's Digital Personal Data Protection Act asks businesses to be clear about how personal data is used.
Signals matter more than settings
Under Andromeda, the strongest targeting comes from the signals you send, not the boxes you tick:
- Conversion events: accurate purchases and qualified leads teach the system who your buyers are. See conversion tracking.
- Catalogue: product data helps match products to people.
- Creative: the words and images in each ad.
- Customer lists: a head start on who your buyers resemble.
Targeting by industry
- Fashion and beauty: India-wide or delivery zones, broad, creative showing different body types, ages and occasions.
- Education: broad within the cities you serve, with separate concepts for students and parents.
- Home services: tight radius, broad within it, ads for urgent needs (repairs) and planned ones (renovation).
- B2B services: broad with a lookalike of best clients as a suggestion, and creative naming the industry and job role you serve.
Next in the series: Andromeda, GEM and Lattice explained. Or go back to the main Andromeda guide.
Frequently asked questions
Yes, as signals. Adding a lookalike of your best customers as a suggestion gives the system a head start, especially in new accounts. Using it as a hard limit usually restricts delivery.
Less than you'd think, if your conversion tracking is right and your creative is specific. The system optimises for people likely to take your chosen action, not for everyone in the country.
For one-time purchases, usually yes, unless you have a separate offer for them. For repeat-purchase products, include them or give them their own budget.
Only if the product is truly for one gender. Even then, gifting buyers can be the other gender. Let the creative attract the right buyers.
