Lookalike Audience.

The starting point is a source audience, such as a list of existing customers, visitors to a specific page or people who submitted a form. The ad platform analyses what these people have in common and looks for users with similar characteristics in its own user base. Which characteristics it uses is not disclosed. For the advertiser the new audience remains anonymous; only its approximate size is visible.

Lookalike audiences are best known from Meta Ads, i.e. Facebook and Instagram. According to Meta's help centre, you choose a source and a country and set the size of the lookalike as a share of 1 to 10 per cent of that country's population; the smaller the share, the closer the resemblance. Other platforms have changed the approach: according to its help centre, LinkedIn discontinued lookalike audiences on 29 February 2024 and now points to predictive audiences and audience expansion. Google Ads retired similar audiences in 2023 in favour of optimised targeting.

The quality of a lookalike depends on its source. A list of your best customers or of people who actually enquired is more meaningful than all website visitors. Uploading customer lists means processing personal data: platforms match email addresses or phone numbers in hashed form, but the obligations of the Federal Act on Data Protection (FADP) still apply, such as disclosure in your privacy notice. Closely related is Customer Match in Google Ads, which targets or excludes your own customer lists directly.

A lookalike audience should be distinguished from retargeting. Retargeting addresses the same people who have already visited the website; a lookalike finds new people who resemble them. As platforms increasingly decide for themselves who sees an ad, manual audience selection is losing weight. Good signals matter all the more: clean first-party data and conversion data, for example via the Conversions API, from which the platform learns who actually becomes a customer.