Meta Ads targeting has three layers: who you manually describe (core targeting), who already knows you (custom audiences), and who Meta finds for you based on people who already convert (lookalikes). Most successful campaigns use all three at different stages.
The classic approach — picking interests, behaviours, demographics, and locations to define an audience. It still works as a starting boundary, especially for location and age range, but it's less precise than it used to be on its own. Meta's delivery algorithm now does most of the fine-tuning within whatever boundary you set, so treat interest targeting as "who's eligible," not "exactly who will convert."
For local businesses this matters more than almost any other setting. A radius around a physical location, or a specific city/area, keeps budget from being wasted on people who could never realistically become customers — a salon in Gulberg doesn't need impressions shown to someone in Faisalabad.
Targets people based on actions Meta has observed — device usage, purchase behaviour patterns, travel habits. Useful for niche cases (e.g. frequent online shoppers) but generally a smaller lever than interest and location for most small business campaigns.
Built from a business's own data — website visitors (via pixel), a customer phone/email list, or people who've engaged with the Facebook or Instagram page. This is the foundation for retargeting and the source material for lookalikes.
Meta takes an existing customer list or pixel data and finds new people who statistically resemble them. This is usually the strongest-performing cold audience available, but it needs enough source data first — a lookalike built from 20 customers won't be nearly as sharp as one built from 500.
People who already interacted with the business but didn't convert — site visitors, video viewers, past ad engagers. Because they're already familiar, retargeting audiences almost always produce the lowest cost per result of any audience type. Any active campaign should have a retargeting layer running alongside cold targeting.
Core targeting (age, gender, location, interests, behaviours), custom audiences (people who already interacted with the business), and lookalike audiences (new people resembling existing customers).
Yes, but less precise than it used to be — Meta's own delivery algorithm now does much of the heavy lifting. Interest targeting works best as a starting boundary, not a fine-tuned selector.
A lookalike audience is a new group of people Meta identifies as similar to an existing customer list — built from pixel data, a customer list upload, or page engagers. It's usually the highest-performing cold audience once enough source data exists.
Any time there's an existing pool of people who visited the site, watched a video, or engaged with a previous ad but didn't convert. Retargeting audiences consistently produce the lowest cost per result because the audience is already warm.
Start broad with core targeting and a clear location radius, layer in retargeting once there's traffic to draw from, and move to lookalikes once there's enough conversion data (typically 50-100 conversions) for Meta to build an accurate model.
Meta Ads management starting from PKR 25,000/month. Live in 48 hours. No contracts.