Login is restricted to DCN Publisher Members. If you are a DCN Member and don't have an account, register here.

Digital Content Next


InContext / An inside look at the business of digital content

The pros and cons of using artificial intelligence in ad tech

January 31, 2018 | By Albert Wang, Product Marketing Manager—SpotX @SpotX

Artificial intelligence (AI) is being used more and more in ad tech to solve a variety of problems. Between the high profile acquisitions and its rise as the industry’s latest favorite buzzword, it’s clear that AI is an extremely powerful tool. However, it’s definitely not a silver bullet. Let’s take a look into a few AI pros and cons.


  • Workflow Efficiencies: One of the largest benefits of AI is how much time it can save on the user side. Without AI, proper campaign optimization takes a lot of time and is absolutely more art than science. Just consider how much data is available on each individual. Even with a target persona in mind, sifting through vendors and guessing at which attributes will perform best is a costly and time consuming exercise at best. Once that’s done, the ad trafficker then needs to toggle pacing, pricing, and potentially dozens of other variables. AI can automate much of that. At which point, the user just needs to pick a goal the AI can optimize toward and let it run giving directional guidance where necessary.
  • More Data Processing than Humanly Possible: Big data and AI go hand in hand regardless of the industry. JP Morgan even published a massive white paper on how they think those two trends will affect investing. When it comes down to it, programmatic trading isn’t all that different from programmatic advertising. It’s all about automated buying and selling to maximize value. AI can “see” and consider as many features as it’s been trained to, considering hundreds or even thousands of variables over the course of a campaign to determine significance. That’s just not something humans can do in any cost-efficient manner.
  • ROIWhat happens when you put workflow efficiencies and maximum data activation together? Cost savings. Lots of it. Assuming your AI strategy is working (and your mileage may vary), adopters of AI stand to reap massive benefits. Since AI requires less human capital to operate, adopters stand to gain from not having to hire as many heads and the heads they do hire aren’t focused on tweaking knobs and levers manually. Additionally, since AI learns as it goes, performance constantly improves over time as it begins to distinguish between what’s important and what’s irrelevant.


  • Black Box AlgorithmsUnless you’re building your own, it’s pretty difficult to know exactly how an AI algorithm works. Two primary reasons for this: 1) The features an algorithm considers are typically a company’s secret sauce, and asking a company to publicize everything that goes in is like asking KFC to share their 11 herbs and spices. 2) Even if there is a degree of visibility into what features are being considered for optimization, oftentimes the amount of data being processed is more than what a human can parsethrough (see Pro #2). Which begs the question…what’s the point of performance if you can’t explain it?
  • Not All AI is Made Equally: If AI is a brain made to learn for a specific purpose, who’s to say whether you’ve chosen the AI equivalent of Einstein or your bratty seven-year-old neighbor? Every partner’s going to represent themselves like they’re Watson, but realistically, that’s impossible. Some partners are better for specific industries, some are probably pure vaporware. Choosing the right partner isn’t easy, and if everyone’s offering an AI solution it’s difficult to say which is the best one for you without at least some degree of upfront investment and a decent amount of research.
  • ROI: Similar to how properly implemented AI can generate huge savings, it can also be a massive sunk cost. The initial barrier to entry – either investing in developing your own algorithms or paying a partner to use theirs – is going to be fairly substantial for most advertisers or publishers. There’s also no guarantee that it’ll work in every scenario. As much as partners would love for you to believe that their AI will make it rain gold bricks every Sunday, that’s just not true. When choosing a partner, don’t just think about their historic performance, but also whether they meet your needs in terms of transparency in both costs and reporting.

As far as AI pros and cons go, it’s hard to say whether AI is right for you. That said, AI is becoming an increasingly important part of a greater shift in the digital advertising ecosystem, and I’m personally interested in seeing how it adapts to other trends. Will AI specced for second price auctions succeed in first price environments? How about in a post-GDPR world? Will the new data restrictions affect performance and will new strategies arise as a result? Who knows, but I’m looking forward to finding out!

Liked this article?

Subscribe to the InContext newsletter to get insights like this delivered to your inbox every week.