Facebook introduced the lookalike audience in 2013 and since then various advertising platforms have created their own version. Lookalike modelling in digital marketing was a game changer as it allowed advertisers to target with precision whilst reaching many more people outside of their existing customer base.
An integral part of the success of a lookalike is the seed audience it is based on. Meta requires a minimum of 100 members but recommend making it as large as possible so that the algorithm has enough data to work with1. With this in mind, it might make sense to include as many customers, or top of the funnel conversions as possible to maximise the size of the audience. In our experience, this won’t always, and in fact rarely, outperforms seed audiences based on a more defined approach. A balance between volume of data and quality / relevance of members needs to be considered.
Of late, many businesses have reported a decline in the performance of lookalike audiences that rely on conversions through the Meta pixel. The iOS 14.5 update has reduced Meta’s ability to track user behaviour and therefore limited the size of any seed audience made from a website or app activity custom audience.
One way to combat the changes in iOS tracking is to make use of the ‘customer list’ custom audience. Rather than relying on the diminishing data gathered by Meta’s pixel, why not use the single source of truth – first party customer data. The two main benefits of using first party data in place of pixel data, is a larger, more extensive data source and the ability to use attributes to segment.
Firstly, utilising customer data enables all converting customers to be used in audiences and not just those that are tracked by Meta. All conversions that happen offline (in store, via the call centre, or by other means) can also be used in audience segmentation leading to larger, more robust seed audiences.
Volume, however, is only powerful when the audience is composed of valuable and relevant people. Segmenting customers into cohorts depending on their engagement with the brand is very powerful. Commonly used segments include all active customers, recent purchasers, lapsed customers etc, however, with AUDIENCES, further creativity is possible, and encouraged. Customers can be segmented across multiple areas, product type, value, seasonality, propensity modelling etc. This allows for more accurate targeting on each campaign.
A leading omnichannel retailer, with both an online and offline presence, tested this theory using AUDIENCES. They set up an A/B test comparing a cookie value-based lookalike and a value-based lookalike using first-party customer data. The lookalike based on first party data attracted higher value customers (+17% in avg rev/trans) as well as driving a higher volume of customers (-20% in CPA).
A step-by-step process of how you can achieve success using first-party lookalikes can be found at:
https://www.audiencesdata.com/use-case/target-prospects-that-look-like-your-most-valuable-customers
Along with other ways you can optimise your current marketing approach by using first-party data.
1 https://developers.facebook.com/docs/marketing-api/audiences/guides/lookalike-audiences/
SELECT EMAIL FROM RETAILER
WHERE (MARKETING_OPT_IN IS NOT NULL)
AND ((AOV >= 50) AND (LTV >= 150));