Advertising technology firm The Trade Desk Inc. is releasing a new set of machine learning-based tools that marketers and advertising buyers can use to plan and schedule campaigns to a greater degree of precision than the company said has ever been possible before.

Headed by the founder of AdECN Inc., which was the world’s first online ad trading exchange, The Trade Desk taps into a data set of nearly 9 million queries every second to match buyers with available inventory. Its self-service, cloud-based platform enables ad buyers to create, manage and optimize campaigns across multiple formats and devices.

The new products – which are collectively referred to as Next Wave — include a forecasting engine called Koa, a data-driven media planning tool for audience profiling and a new user interface that identifies tailored insights and offers recommendation for real-time decision-making.

“We’ve built a UI to let people look at 10 million impressions and pick the best five or six of them,” said Dave Pickles, cofounder and chief technology officer.

Programmatic and exchanges are increasingly becoming the engine that drives the $204 billion digital advertising industry. They match available advertising space on websites, mobile devices and multimedia to bids by ad buyers, making millions of decisions at millisecond speeds.

The Trade Desk said its approach to predictive analytics intentionally avoids the “black box” metaphor that is common in ad tech. Koa is intended to enable buyers to envision an entire campaign before spending any money, based upon key performance indicators supplied by the buyer, such as clicks, engagement and cost per acquisition. Customers can work with The Trade Desk’s data, their own or any combination.

The company, which went public in 2016 and reported revenue of nearly $86 million in its most recent quarter, distinguishes itself from competitors in part by its big-data-driven technology. It said its trio of new products provide deep insights into audience characteristics along with campaign modeling capabilities. Machine-learning algorithms create statistics-based predictions that can be applied to trading plans. Algorithms use historical data to precisely value impressions without overspending, eliminating the need for traders to make those adjustments.

“Koa was built to make your decisions easier and faster, but you always have complete control and full transparency,” the company said in a press release. “If you select Koa to automate optimization changes, you will have full visibility on these changes in your activity log.”

Availability is planned for this summer.

Image: The Trade Desk