Data Gran launches AI-powered tool for optimizing Facebook, Instagram (and soon, Google) campaigns
A new campaign testing and optimization tool for Facebook and Instagram advertisers launches today after a two-year development and beta testing period. Support for Google campaigns is in the works.
AdOptimizer from Data Gran uses artificial intelligence and machine learning to determine and fund the top-performing mix of creatives and targeting parameters among the more than 2,000 available in Facebook and Instagram.
Advertisers upload images and audiences and choose the channel and device targeting they want for their campaigns in AdOptimizer. The system then creates 50 ad variations defined by the targeting parameters and runs those ads for one day.
“Then we fetch all that data — who clicked, what they clicked, on what devices, what different parameters are working — and take that to our artificial intelligence algorithms to automatically predict which creative will work best and with which specific audiences for the rest of the campaign,” Carlos Mendez, founder and CEO of Data Gran, explained in a phone interview. That combination is often not what the advertiser had assumed.
After two days, the tool diverts the budget to the campaign variation driving the best performance.
The idea is to use artificial intelligence to dramatically shorten testing periods and get better outcomes at lower costs and churn rates.
“We built Data Gran’s toolkit with intention and caution, to ensure a product that addressed the pain points of digital advertising that I observed over my 15 years at advertising agencies,” said Mendez in a statement. “We designed the interface for ease of use, and the technology to take on the burdens of execution, identifying the best-performing campaign and allocating funds accordingly.”
Data Gran counts Miller, Coors and Bic among the beta testers it says have reduced the cost per customer by 20 percent or more. Mendez says Subway Mexico is running a large campaign for which the system initially created more than 5,000 ads in 10 different objectives between training phases.
Advertisers can access reporting and drill in to see performance by demographic, location and other parameters.
“Agencies are still doing all of this by hand,” says Mendez in talking about the interest the company has seen from agencies. “Also agencies want to produce real-time ads based on data. Now they can get that data and see what type of images and emotions are working best to use in their other creative.”
Data Gran is part of Facebook’s Marketing API Accelerator Program and a Google Partner. AdOptimizer pricing is a flat fee of 13 percent of media spend, though Mendez says that scales down for very large advertisers.
The company says support for Google search and display campaigns will be available soon.