Facebook ads targeting has quietly flipped over the last few years, and most of the advice floating around hasn’t caught up. The old game was stacking interests and demographics until you’d hand-built a narrow audience. The new game is mostly the opposite: give Meta’s algorithm a clear goal and a broad-enough pool, then let it find your buyers. Get that shift wrong and you’ll strangle your own campaigns with over-targeting. Get it right and the same budget reaches dramatically better people.
This is the practitioner’s guide to Facebook ads targeting in 2027 — the three audience types that matter, when to go broad versus narrow, and the setup I use on most accounts.
The targeting landscape in one box
- Core audiences — targeting by demographics, interests and behaviors you define
- Custom audiences — people who already engaged with you (site visitors, lists, video viewers)
- Lookalike audiences — new people who resemble your best custom audiences
- The trend is broader targeting + better creative, letting the algorithm do the finding
- Your pixel and conversion data are what make all of it work
Why broad beats narrow now
For years the instinct was to narrow: layer “interested in yoga” AND “parents” AND “income over $100k” until you’d carved out a tidy little audience. That made sense when Meta’s delivery was dumber and you had to do the finding yourself. It doesn’t anymore. The delivery system now reads your conversion signal and finds the people inside your audience most likely to act — and the wider the pool you give it, the more room it has to optimize.
Over-narrow targeting fights the algorithm. It forces delivery into a tiny box, raises your costs because you’re competing for the same sliver of inventory, and starves the system of the volume it needs to learn. The modern move is to hand Meta a broad audience, a clean conversion event, and strong creative, then let it sort. Your creative does the targeting that interest stacks used to do — a yoga ad naturally attracts yoga people even in a broad audience. This is the single biggest mindset change for anyone coming back to the platform, and it’s a theme we go deep on in the 2027 Facebook ads strategy.
Core audiences: the ones you build
Core audiences are the ones you define from scratch using Meta’s targeting options — location, age, gender, languages, detailed targeting (interests and behaviors), and connections. They’re your starting point when you have no data yet, and they still matter for controlling the obvious boundaries.
What to actually set
Keep it light. Set the boundaries that genuinely matter — location you serve, age range that can buy, maybe one broad interest to give a cold account a nudge — and then stop. Resist the urge to stack five interests. A good rule: if removing a targeting layer wouldn’t let an obviously wrong person in, the layer isn’t earning its place. Location and age usually earn it. The fourth interest usually doesn’t.
Over-targeting vs. letting it breathe
✓ Give the algorithm room
- Broad audience, one clear conversion event
- Creative that speaks to the buyer directly
- Location and age as the only hard limits
- Enough budget to exit the learning phase
✕ Strangle delivery
- Five stacked interests “to be safe”
- Tiny audience competing on thin inventory
- Narrow box with no room to optimize
- Too little budget to ever learn
Custom audiences: your warm pool
Custom audiences are built from people who already have a relationship with you, and they’re the most valuable targeting you own. You can build them from website visitors (via the pixel), customer and email lists you upload, people who engaged with your Facebook or Instagram content, video viewers, and more. These people know you, which makes them cheaper to convert and perfect for retargeting.
The engine behind all of this is your pixel and the conversion data it collects. Install it properly, fire the right events (view content, add to cart, purchase, lead), and you’re feeding Meta the signals it needs to both optimize prospecting and build these warm pools. Without it, you’re targeting blind. Custom audiences are also the raw material for retargeting campaigns, which I treat as their own discipline — the segments, windows and messaging get detailed in retargeting audience segments.
Lookalike audiences: scaling what works
Lookalikes are where custom audiences become a growth lever. You hand Meta a source audience — ideally your purchasers or your highest-value customers — and it finds new people who resemble them. A 1% lookalike is the tightest match to your source; widening to 3%, 5% or 10% trades precision for reach.
The quality of a lookalike is entirely about the quality of its source. A lookalike of “all website visitors” is mediocre because the source is noisy. A lookalike of “people who purchased in the last 180 days” or “top 25% of customers by spend” is powerful because the source is pure signal. Feed it your best people and it goes looking for more of them. Feed it everyone and it finds anyone.
Your targeting is only as smart as the data you feed it. A clean purchase-based custom audience is worth more than any interest stack you can build by hand.
The setup I use on most accounts
Here’s how I actually structure targeting when I’m building or rebuilding an account. It’s a funnel, not a single audience.
- Prospecting: go broadCold campaigns use broad or a single-interest core audience, plus a strong lookalike of purchasers. Let creative and the algorithm find buyers.
- Nurture: lookalikes + engagersScale with lookalikes of your best customers and campaigns to people who engaged but didn’t buy.
- Retargeting: warm custom audiencesClose with site visitors, cart abandoners and video viewers — the cheapest conversions in the account.
Each layer gets creative suited to its temperature. Cold audiences need a hook and a reason to care; warm audiences need a reminder and an offer. Same product, different message, because the person’s relationship with you is different at each stage.
Privacy changes and what still works
You can’t talk about Facebook targeting in 2027 without the privacy context. Signal loss from iOS changes and cookie restrictions shrank the data flowing back to Meta, which is exactly why the broad-plus-creative approach won: narrow targeting depended on rich signal that’s now thinner. The counter-moves are first-party data (your customer lists, server-side tracking through the Conversions API) and giving the algorithm volume to work with. Build on data you own, feed Meta clean conversion events, and your targeting holds up even as third-party signal degrades.
Common targeting mistakes
The patterns I see over and over: stacking too many interests and wondering why costs are high; building lookalikes off weak sources like all-traffic instead of purchasers; never installing proper conversion tracking, so the algorithm is optimizing toward nothing useful; and changing audiences every few days before delivery has had a chance to learn. All four come from the same place — trying to out-think the algorithm instead of feeding it. Feed it well and get out of its way.
Targeting is one piece of a bigger Meta picture that includes objectives, budgets and creative. For how it all fits together, start at the Meta ads hub. And if you’re weighing where paid social budget should go beyond Meta, TikTok ads for beginners is a useful companion.
Frequently asked questions
What are the main Facebook ads targeting options?
Three types: core audiences you build from demographics, interests and behaviors; custom audiences of people who already engaged with you; and lookalike audiences of new people who resemble your best customers. Custom and lookalike audiences depend on your pixel and conversion data.
Should I use broad or narrow targeting on Facebook?
Broad, in most cases. Meta’s delivery now finds your buyers inside a wide audience better than hand-built interest stacks can, as long as you give it a clear conversion event, strong creative and enough budget to learn. Over-narrow targeting raises costs and starves the algorithm.
How do lookalike audiences work?
You give Meta a source custom audience and it finds new people who resemble them. A 1% lookalike is the closest match; 3% to 10% trades precision for reach. The quality depends entirely on the source — build lookalikes from purchasers, not from all website traffic.
Do I still need the Meta Pixel with privacy changes?
More than ever. The pixel and the Conversions API feed Meta the first-party conversion signals that targeting and optimization now rely on, especially as third-party cookies and tracking degrade. Without it you can’t optimize toward buyers or build custom and lookalike audiences.
Targeting on Meta rewards the people who feed the algorithm clean data and trust it to find buyers. Build your custom audiences, lookalike off your best customers, and let broad creative do the rest. If you want to build these audiences alongside practitioners running real ad accounts — and see which setups actually scale — come work through it with us at SEO Spring Training in Chandler this April. Grab your pass here.
