December, 2018 Blog

Ask the Expert: What is Machine Learning?

‘Ask the Expert’ is a blog series that breaks down the complicated tools, tech, and trends you’ve been hearing about in the trade pubs and around the office. We reach out to our in-house experts to ask the tough questions and turn them into bite-sized Q&As for your reading pleasure.

This month’s topic? Machine learning. We brought in Centro’s VP of customer success, Steve Monti, to give us the breakdown.

In terms of advertising technology, what is machine learning?

Machine learning is a type of artificial intelligence (AI) where a computer system learns from data. In advertising technology, this is often used to improve the performance of ad serving systems, mostly in programmatic advertising, to show ads to individuals in order to achieve a result. The result could be a click or a variety of other types of actions that the advertiser wants. Different elements of machine learning can be present in many common systems used within the advertising tech space, such as demand-side platforms, ad exchanges, ad servers and more. 

How does it work within ad buying systems?

Software with machine learning can optimize the performance of campaigns by gathering data to learn how key performance indicators (KPIs) were achieved in the past. The system then learns what element combinations are needed in a campaign to have the highest likelihood to generate the same result or KPI. The system will then structure a campaign to match those parameters to obtain the impressions that can lead to those same KPIs. Elements could include targeting data, device or location of a user, ad unit size, content of page, and more. Machine learning can build a model that it will continue to hone as a campaign is running.

Is it related to algorithmic optimization?

Somewhat, of course! Machine learning uses algorithms, but not all algorithms lead to machine learning. Algorithms have the ability to make finite, black and white decisions. Machine learning implies a logic of learning from those decisions over time.

Is machine learning strictly for optimizing to generate clicks?

Machine learning technology builds a model. In an ad-buying system, a model is made up of a mix of targeting combinations that calculate the probability of a result occurring. A click is a highly desired result, but a marketer can, theoretically, program their system to model towards other types of results as well.

Ok, I’m interested — what does Centro offer in terms of machine learning within its platform?

Inside Centro’s platform, Basis, users can optimize campaigns with algorithmic optimization or machine learning. These features can also be turned ‘off’ or ‘on.’ When ‘on,’ the system continually learns based on more than 30 targeting parameters and will set bids based on the probability of success.

Are you interested in other Centro resources that will help you understand machine learning? Reach out to info@centro.net.