Artificial intelligence is already having a huge impact on industry, helping researchers crunch numbers, making it easier to snag tickets to Broadway shows, and doing countless other jobs in between. It’s also set to revolutionize the way some of the biggest tech companies generate ad revenue.
Ads served up on websites are typically meant to be relevant to the person that’s seeing them — an evolution of the idea that TV ads should be placed among shows that fit the product’s target demographic. AI can use deep learning techniques to analyze large amounts of user data and predict which ads an individual is most likely to click on.
Companies including Microsoft, Google, and Chinese ecommerce giant Alibaba have released research into the benefits of deep learning to online advertising. One common thread is that incremental improvements can offer up large amounts of revenue for advertisers working at this scale.
“Even a 0.1 percent accuracy improvement in our production would yield hundreds of millions of dollars in additional earnings,” read a report from Microsoft’s Bing group, published in April 2017. It claims to have recorded an improvement of 0.9 percent on one advertising metric using AI.
Of course, there’s still a way to go before these companies perfect the art. The team behind the Microsoft study intend to explore different kinds of learning algorithms going forward. A similar report from Google indicated that better optimization is a priority for any follow-up research.
We’re still early in the process of seeing what AI can really do to benefit advertising revenue — but while this technology is certainly going to benefit the companies selling ad space, it’s also set to make things better for the advertisers and their audiences.
Online ads can be annoying, which has prompted the rise of services like AdBlock. However, most internet users would agree that ads are only really unbearable when they’re not relevant.
At the end of the day, all parties stand to benefit when ads are displayed to a receptive audience. There’s not much point in trying to market a product to someone who has no interest in it whatsoever. The tough part is ensuring that advertisements are served up to the right person.
Deep learning can turn reams of data into something actionable: of thousands of users who viewed an ad, they can hone in on the categories of people who responded positively. Admittedly, it’s not an exact science, but companies like Microsoft and Google have access to sufficient amounts of data such that their algorithms can make fairly accurate assumptions.
This isn’t the most radical use of AI to market a product (we’re not seeing an artificial mind serve as creative director on a television commercial, after all). However, the basic idea is based on the same principles; it makes sense to target as wide a swathe of potential customers as possible, but people who are predisposed to be interested in the product should probably take precedence.
Online ads can reach a huge amount of people, and the information that companies have about the potential audience for those ads makes it easier to ensure that their marketing efforts hit the mark.