Audience targeting is standard procedure for marketing teams, and lookalike modelling is an increasingly used, although not widely understood, option in that procedure.

 

Lookalike models are an important tool for marketing teams looking to reach out beyond the narrow scope of their CRM. They help media planners like ourselves to reach prospects that look like existing customers, and by doing so help achieve a more efficient plan with less wastage.

 

And their importance is only going to grow in the future as the industry increasingly turns to programmatic media buying. Having a lookalike model is very useful for all media buying, but it is in the programmatic world that it really shines, providing the necessary solid (data driven) base from which automation can progress.

 

All of which makes this a good time to take a moment to unpack the process a little and better understand it.

 

Lookalike modelling finds audiences that the advertiser would otherwise struggle to identify, helping us to create reach by building extensive possible audiences from existing/ideal customers. The characteristics and behaviour of the ideal/current audience is extrapolated and mapped out, creating a yardstick set of characteristics and behaviour. These standards, closely replicated, can be used by data providers to reach new prospects that look and behave like the ‘best customers’ of an organisation and who, as a result, are more likely to turn into ‘best customers’ themselves when reached with relevant advertising.

 

The tighter the model is, the more chance of reaching the ‘best customer’. Conversely, the model can be loosened to grow the list of (less than perfect) prospects.

 

The process is digital, as the large amounts of data needed has only become viable with the advent of online companies selling data for the targeting of advertising. With the digitisation of TVs, radios and other household devices/screens however, this type of data retrieval and use will spread beyond the traditional online user.

 

The whole point of lookalike modelling, and audience targeting in general, is to provide better value for an organisation, but the quality of the data the advertiser uses to build the model is vital to delivering that value. Bad quality data not only wastes considerable time and money creating the model, but the (likely automated) media buys won’t be as efficient as they should be. Transparency is important, both in terms of how the data is accrued and how it is used. In an ideal world, the impressions created by lookalike modelling would be sifted out from those created via traditional source data, so the media planner can accurately gauge the success of the model.

 

By Oliver Brown