New Media Trial Puts Machine Learning to the Test

Download the full report here

The process in which marketers identify and reach the ideal demographic for their product has remained fairly static over time. Historically, primary research is conducted, followed by a media buy that is deployed against that demographic. However, the latest advancements in technology, specifically machine learning, are opening up the possibilities for optimization beyond what humans can achieve on their own.

In our latest media trial, we put machine learning to the test to uncover whether human-driven media buys would benefit from this additional optimization.

Key findings include:

• Machines working with humans beat humans alone in virtually every category measured, including brand interest (machines +8.6%, humans +1.9%) and purchase consideration (machines +5.7%, humans +0.8%).
• Machine learning was also more efficient in doing so with only 3.08 ad exposures per consumers vs. 4.13 for humans.
• Humans couldn’t simply mimic the machine’s learning. Even when controlling for key audiences (e.g. those in-market for the product), the machine performed better by seemingly optimizing towards those most receptive to the ad.

Download the full report here.