Targeting is how you tell a platform who should see your ad. The options fall into a few broad categories, and one of them consistently outperforms the others.
| Type | Based on | Typical use |
|---|---|---|
| Demographic | Age, location, language, sometimes job or education | Basic filtering — usually a floor, not a strategy |
| Interest | Topics and pages a person engages with | Reaching people likely to care about your subject |
| Behavioural | Actions taken, such as recent purchases or device use | Narrowing to demonstrated behaviour |
| Custom | Your own data — site visitors, customer list, video viewers | Retargeting; usually the strongest |
| Lookalike | People resembling one of your custom audiences | Scaling once you know who converts |
Retargeting means showing ads to people who already interacted with you — visited your site, watched a video, opened an email, or added something to a basket without buying.
It reliably outperforms advertising to strangers, for an obvious reason: these people already demonstrated interest. They are not being introduced to you; they are being reminded. Cost per result is usually substantially lower.
A small piece of code on your site — commonly called a pixel or tag — records that a browser visited a particular page. The advertising platform can then include that browser in an audience you target later.
Two practical notes. First, install it before you need it: audiences build over time, so a pixel added the day you start a campaign has no history to work with. Second, this is personal data and is subject to privacy law in most jurisdictions — consent requirements, a privacy policy that describes it, and honouring browser and platform privacy settings are obligations, not optional extras.
Retargeting has become less precise as browsers and operating systems restrict cross-site tracking. It still works, but audience sizes are smaller than they once were and matching is less reliable. Any advice written a few years ago overstates how much of your traffic you can expect to reach.
A lookalike asks the platform to find people who resemble an audience you supply — typically your customers or your best site visitors. It is the main way to scale beyond people who already know you, and it works considerably better than guessing at interests.
Its quality depends entirely on the source. A lookalike built from a small or mixed list produces a vague audience; one built from a clean list of actual paying customers produces a much better one.
The instinct with targeting options is to stack them — a specific age, in one city, interested in three things, who recently did a fourth thing. Each addition shrinks the audience, and past a point the platform has too few people to optimise across, which makes results worse and costs higher.
On modern platforms, the systems are generally better at finding responsive people than manual interest stacking is. Give reasonable constraints and enough room to work.