Audience Targeting With Lookalike Modelling

Lookalike modelling is a fantastic way to use current knowledge of your key audience segments to expand campaign reach and maximise performance. It is a technique that helps reach new users online who are likely to be interested in your product or service based on current customer loyalty. With an integrated marketing approach insights gained from lookalike reports can feed into other digital channels, as well as guide offline activity and PR decisions. This blog outlines the basic technicalities of lookalike modelling that will enable you to make the most of your programmatic display campaigns, and offers suggestions for improving your audience prospects and marketing efforts as a whole.

The Technicalities

Audience Composition Report

DoubleClick Bid Manager hosts an array of reports to help with campaign optimisation and finding new audiences. The audience composition report currently sits at the top of my priority list for when I want to expand my prospecting opportunities, gain exciting insights, and expand current targeting capabilities. The fundamental use of this report is to help predict which new audience segments may perform better, based on data. It compares the behaviour of audiences that have been to your website (first-party data) with third-party data provider audience lists. Given the hundreds of thousands of audience segments available via the DoubleClick platform, the opportunities are huge! Please note, due to privacy considerations this report may not display data for targeting combinations that match fewer than 3,000 unique cookies.

For example, a composition report for an online retailer selling high end watches could show a strong match to the following audiences:

Metric - Match ratio

This is the metric that you should be filtering for within the audience composition report. It’s a correlation that describes the similarity between your first party audiences to the users on a third-party audience list. Numbers greater than 100 represent a positive correlation, where higher figures indicate a greater similarity. Keep an open mind, as some audience lists may surprise you! For example, the image above may not have been the stereotypical audience for those people in-market to buy an expensive watch but, based on the data, they would be worth a trial.

The Opportunities

Visitors vs. Converters

Once you’ve got to grips with the audience composition report, you can start filtering by the type of first party data list. A key strategy here is splitting out lookalike reports for site visitors and comparing that to past purchasers, or dividing your audience segments based on micro converters. Are there differences between them that can be integrated through audience targeting? Can we identify avenues offline that should be pursued based on our understanding of purchase behaviour – for example, does offering a £50 Waitrose voucher influence abandoned baskets to come back and convert (because our audience composition report tells us they like to shop at Waitrose)? Can we update our landing page with USPs that are likely to persuade visitors to sign up? Or should we be supporting a particular charity, because this is what is important to customers?

Similar audiences

This is a relatively new feature in DoubleClick Bid Manager that allows you to target ‘similar audiences’ automatically in the interface. There is a tick box that activates a powerful lookalike modelling algorithm for reaching individuals who share interest profiles with your existing customers. It is a great strategy for reaching additional intelligent targeting at scale and extends to mobile identifiers too!

The similar audience feature is the first step in enhancing your audience targeting opportunities, but should not be used in isolation. It is important that you are regularly pulling the audience composition report too in order to learn insights, expand your audience knowledge and adapt the rest of your campaign based on the results.

At Periscopix we have a committed audience-centric approach to campaign management and are regularly using lookalike reports to push our display campaigns further to reach the most relevant customer at key touch points in their path to purchase. If you are interested in finding out more, please contact us.

To view the blog as written by Jessica Maunder on Periscopix's website, please click here.