District M’s new algorithm focuses on ‘non-potential clickers’

The new feature is designed to weed out audiences least likely to click through an ad and avoid wasteful digital placements.

Montreal-based ad tech company District M has completed work on a new proprietary AI algorithm designed for campaign optimization.

The algorithm was built using predictive models that identified “non-potential clickers” in a brand’s audience during a digital campaign. It also identifies ad placements on a website that are best able to deliver on a brand’s campaign objectives and that sit within its specific industry.

Ricardo Machado, the senior software engineer who led the project, explained that the main difference between this AI and others on the market is its audience focus. “Whereas certain optimization algorithms use data to target specific users, ours uses non-personal and anonymous data to determine those who have the lowest probability of clicking the ads,” he said.

The point of filtering out the uninterested rather than casting a broad net, according to District M, is to make audience targeting more specific while reducing wasteful digital ad placements.

According to Benoit Skinazi, SVP of sales at District M, during A/B testing, the algorithm saw a 17% increase in click-through rate, as well as a 60% increase in post-click conversions. The resulting savings on the client side were a 31% reduction in cost-per-click.

The AI is the first project to come out of District M’s new Helix lab, an in-house development space that was established to determine and capitalize on changing tech trends that are relevant to the marketing industry. District M will begin integrating the user identification AI into all of its campaigns by the end of August, and the inventory and placement targeting, still being tested, are expected to run by the end of September.