Lotame introduces Precision Audiences to boost its data accuracy
The DMP is now employing machine learning to find more accurate sets of gender/age-based segments of users.
Data providers are increasingly looking to differentiate themselves with assertions of a higher level of accuracy.
Recently, for instance, predictive analytics firm EverString launched a new platform that it said offered twice the standard rate of accuracy for its business data. And now, data management platform Lotame has launched Precision Audiences, which it says results in higher accuracy levels than industry benchmarks.
In Precision Audiences, SVP and GM Jason Downie told me, a seed audience segment of known user profiles — grouped by gender and age range, such as female, 18-34 — is used to create a model for a machine learning engine, which is then trained on a separate segment of profiles.
In other words, the engine looks at attributes from the known seed segment and attempts to find the same gender and age range from the training set.
Finally, the results are compared to an industry benchmark which has been tested against a sample of logged-in users on Facebook, where their age and gender are known. The benchmark is a percentage of ad impressions that reach the targeted audience.
Downie said that, in this first release of Precision Audiences, segments will be based on age and gender. In future releases, he added, segments will be generated based on other attributes, such as online behavior. Previously, he said, Lotame’s segments were created from observed web behavior and from demographic data, but not via machine learning and a validation of accuracy.
Lotame says Precision Audiences has delivered results that are double the industry benchmark for an unnamed major consumer packaged goods (CPG) brand in the US, 30 percent above the benchmark for a CPG brand in Germany and 25 percent above the benchmark for a UK CPG brand.
There are a few competing third-party data providers that generate segments of comparable accuracy for the US and UK, Downie said, but added that no others can do so for as many as 30 countries, as Lotame can.
Opinions expressed in this article are those of the guest author and not necessarily MarTech Today. Staff authors are listed here.