Criteo launches advertising prediction engine based on actual purchase behavior

Criteo (NASDAQ: CRTO), a global technology company that specializes in digital performance advertising, yesterday announced the roll-out of significant technical enhancements to its proprietary prediction and recommendation engine, The Criteo Engine, after more than three years in the making. The new technology enables Criteo to decide whether to buy each ad impression based on the likelihood that a user will both click on an ad and then purchase on the advertiser’s site. This advancement creates the opportunity for clients to expand their reach to a greater number of potential buyers all within their previous cost-per-sales target. As a result, the enhanced engine strengthens the overall performance of the Criteo platform with a 38 percent increase in sales at exactly the same cost-of-sale. This 38 percent lift has been measured across billions of impressions, and millions of users and clicks.

“The enhanced Criteo Engine is a major breakthrough. Criteo has always been known for its performance and CPC (cost-per-click) pricing. We are very excited to be able to announce this significant increase in our performance. The Criteo Engine is now processing huge volumes of data with improved models, and as a result we are able to predict events that happen only once out of every 10,000 times – like when an ad impression generates a click and a sale,” said in a statement Jonathan Wolf, chief product officer at Criteo. “This is where advertising needs to be and we are very excited to continue to prove the value that can be generated from focusing on pure post-click performance.”

Over the past three years, Criteo has committed substantial time and investment to build out its infrastructure and improve the engine’s speed and accuracy through dedicated research, development and testing.  The newly reengineered Criteo Engine has the ability to process up to 15 million predictions per second and respond within 20 milliseconds. Criteo now stores more than 20 Terabytes of data daily, including actual purchase behavior. Many weeks of data need to be analyzed every few hours in order for the new prediction model to work, which required open source technologies including Apache Hadoop.

The next generation of The Criteo Engine also introduces several new capabilities for Criteo clients, like simplified segmentation designed to streamline the workflow for planning and management, making it easier for advertisers to focus on their marketing objectives.

These major technological enhancements to The Criteo Engine will be automatically deployed globally to existing clients immediately at no extra cost.