Marketers use a look-alike audience in order to find and connect with new customers. But what does “audience modeling” or “look-alike modeling” really mean?
Let’s distinguish between two types of audiences:
- Observed audience: Customers who were directly “observed” taking a specific action, such as clicking an ad, filling out a form, or purchasing a product. Observed audiences are often used as the basis for a “seed set,” which can be used to model a larger, relevant audience.
- Look-alike audience (or “modeled audience”): A new, expanded audience of people with similar attributes, behaviors and demographics to the original “seed set” audience. Statistically, this audience is likelier than the average consumer to take a same desired action (such as click an advertisement or buy a product).
Look-alike audiences are great for targeting consumers whom have not considered you before, but are—hopefully—likely to resonate with your brand or offer. Many brands begin by modeling audiences based on their best customers (their most loyal or high-value customers, based on their own first-party data) to find similar people who are likely to spend more or connect with their brand.
However, if your goal is to hit shoppers who have intent in purchasing a product for a direct response campaign, then you may want to focus on observed audiences, who consists of people who have actually taken a desired action, such as clicked on a product page or purchased a specific item before.