The coming wave of AI in advertising will be defined by the autonomous execution of cohesive, multistep campaigns. For brands, this will mean relinquishing control and trusting the technology to do what they've traditionally relied on complex technologies and teams to handle -- but at a far greater pace and scale.
Before handing over the reins, it's helpful to understand how AI works. Here's a look at five overarching steps that go into converting human thought processes into algorithms, and algorithms into digital marketing programs that run autonomously, from start to finish.
1. Understand why marketers do what they do.
Creating AI for "self-driving" marketing technology is not so different from creating AI for a self-driving car. In the case of the car, it must know how close it is to other cars, how to make a turn and end up in the right position after the turn, when to hit the gas pedal, what the road conditions are like and so on -- all without the driver telling it what to do.
Like driving a car, many of the decisions that drive the day-to-day execution of marketing programs also happen largely on the subconscious level. Transforming these processes into algorithms requires understanding why each decision was made by acutely observing marketers as they execute them and then asking them to verbalize the reasoning behind the decisions they made:
"Why did you keep these words and ditch those?" "How did you decide bid size?" "Say you increase spend on a specific keyword by 20% ... why did you choose 20%?" "What's the best time to send stuff to that person?" "What about that other person?"
2. Teach technology how to understand abstract information, such as creative.
Data is unquestionably the domain of AI, but what happens when technology is asked to process and make decisions that are more creative in nature?
For a human, understanding why certain images and text make more sense as a first interaction with a consumer rather than as a secondary or final interaction is almost second nature. A machine, on the other hand, needs to be told (or programmed) with this knowledge in order to be able to judge images and text and determine where they should appear along the journey, without relying on a human.
3. Program it to consider all scenarios and outcomes before each and every move.
At any moment, there are several variables -- and combinations of variables -- that influence an exponential number of outcomes in a campaign. "If I do this, that will happen. But if I do that, this will happen."
Take deciding which headline to use with which creative on a specific channel. It could be that there are 10 creative options and six headline options. The technology must create a real-time model to predict which of several thousand possible headline/creative combinations will perform best in relation to all other combinations, considering variables such as known audience, past behaviors of similar audiences, the specific channel, geographic region, time of day and so forth. Once it's predicted every possible outcome, it must execute the headline/creative combo that's determined by the model to perform best.
4. Make individual building blocks work together as a holistic system.
A major issue in digital marketing is the fact that different aspects of the program -- Facebook, search, Twitter, display, email, SMS and so on -- are handled by different people and technologies.
Each is privy to different insights and uses them to calibrate their respective efforts, but it's impossible to manually gather insights from one channel and apply them across all relevant channels at the rate and efficiency of a machine.
For AI to do this, such systems will require an understanding of the interplay of all the moving pieces that go into a cohesive, holistic program -- plus the ability to sequence them to create a whole that's greater than its individual parts.
5. Introduce checks and balances so the AI doesn't go rogue.
Making sure AI doesn't go rogue is a huge concern for companies, so it's necessary to introduce built-in rules that prevent it from making decisions that are at cross purposes with the people or organization it's serving.
This is especially critical when it comes to budget-related decisions. Imagine, for instance, that the machine predicts you should triple your regular ad spend. In this case, checks and balances would kick in to give the team the opportunity to understand the market conditions and potential outcomes before agreeing to let the machine act on its recommendations.
Finally, all of this must happen autonomously, with little to no input from marketers, and at a far greater scale than is possible by even the largest teams.