Back to Basics: Lookalike Modeling Explained

As business owners, marketers, and sales teams, we’re always looking for new audiences to target. Based on how quickly things change, it’s likely your most valuable customers now are different from the ones you had 12 months ago. That is why looking at today’s data is more important than ever.

What is Lookalike modeling?

By collecting data, and we mean lots of data, you can begin to find new people who behave like your best and most profitable customers.

For example, let’s say you run an eCommerce store and you’ve identified that your best audience consists of people whose average purchase is over $150 and who make a purchase at least twice a month. Lookalike modeling will allow you to find more people like that.

How does it work?

The first step is placing a tracking pixel on your order confirmation page. This will allow you to track the behaviors of and collect insights on your converted audience. Let’s call this our “seed audience.”

Once enough data is collected – we recommend at least 1,000 unique hits from the past 7 days – upload it into a platform like a data management platform. Once the data is ingested, it can be analyzed to identify the behaviors that are most common among people who purchase from you.

Machine learning and AI tools do the heavy lifting through pre-configured algorithms that work quickly to identify consumer preferences and then find other consumers who share those preferences.

Once the custom attributes are identified, they can be used to engage with a larger audience that have a high probability to convert.

Best Practices:

The success of this tactic relies on the quality of the data provided by your seed audience. If you feed it bad data, then you will likely receive a bad model. Thus, the goal is to seed the model with the best users possible.

In display advertising, the best method is to utilize the users that trigger the conversion pixel as the seed audience because they’ve already demonstrated that they are interested in your product. You want to build a similar audience that will also like you product.

Final thoughts

Before you start building these models, get to know your audience well. Dig into your top customers and see what they have in common. Don’t make assumptions – you have the data to guide you. Leverage social media, conduct surveys, and most importantly- look at recent purchase history.

When done right, lookalike modeling is an audience tactic that will allow you to improve campaign performance and extend the reach of your online advertising campaigns.