Imagining the possibilities of marketing and artificial intelligence

The headlines say ad tech is dying or even proclaim that it’s dead.

Consolidation is inevitable, no doubt, but I suggest that ad tech is evolving. Anyone who says that ad tech is dying is simply looking in the wrong place.

Sure, buying and selling ads in squares and rectangles is becoming commoditized, and users are boycotting them. And why not? It’s 20-year-old format; you can argue that it’s lived past its lifecycle.

But it’s important to look past the crowded landscape and see what’s ahead — there are lots of new and exciting solutions in advertising technology. Artificial intelligence is the area where I see incredible possibilities. Here’s why.

As digital advertising matures, artificial intelligence (AI) will become ubiquitous. However, there are gaps in the evolution of AI and advertising solutions that will require a whole new set of service providers creating solutions to fill specific voids.

From my point of view, ad tech isn’t dying, it’s simply morphing to meet the needs of today’s marketer.

I recently discussed how marketers are currently leveraging artificial intelligence (AI) with Ritesh Soni, SVP of data science, analytics and engineering at SapientRazorfish.

I asked him to name the biggest challenges involved in doing his job and handling an AI practice and here’s how he answered:

“The hardest thing that people are trying to wrap their heads around is being able to imagine ways in which AI can be applied and customized for their business. And Hollywood hasn’t helped.”

“However,” he went on to say, “once you speak to people on the ground like marketers and demystify what AI can do, it opens up a whole new dimension around understanding consumers and increasing relevancy in messaging.”

Here are three key areas where marketers are leveraging AI today.

Behavior-driven segmentation

It’s no secret that personalized messages drive higher conversion rates than those aimed at a broader audience. When MailChimp measured stats “across all segmented campaigns,” segmented campaigns performed much better than their non-segmented counterparts.

Email opens were 14.31 percent higher than with non-segmented campaigns, unique opens were 10.64 percent higher, and clicks 100.95 percent higher. Meanwhile, bounces, abuse reports and unsubscribes were all lower compared with non-segmented campaigns.

Essentially, this is personalizing marketing through analytics sitting on top of a campaign. It allows for top-down segments to evolve and be enriched with bottom-up data-informed signals, making creative execution clear and more relevant.

While behavioral-based targeting and expected returns has been a widely researched and published domain for well over a decade, it has surprisingly low penetration in its adoption within marketing operations in Fortune 100 companies.

Propensity modeling

Propensity, or predictive, modeling helps marketers identify the propensity that their customer will do something. This could be buying a related product or service. It could also be identifying if the consumer who purchased something in one product category will buy another item from another category.

We can add a level of scoring to actions which increases the data sets (why you need the AI) but allows for much richer actions.

Specifically, “[W]e can then determine the next best action to point our consumers towards. These models quickly become the core of any AI program in marketing departments,” Soni said.

Insights from massive scale signal data

Massive amounts of signal data — the more the better. Think location beacons, social media messages and audio and listening data. AI and machine learning enable companies to derive actionable insights from large data sets.

Another overarching theme to consider is the pace of algorithm development. “In today’s marketplace, you are getting something new every six months, and the new tech is rendering the old bits obsolete,” Soni said. “The pace is exciting.”

Just recently, Google open-sourced a new Tensor2Tensor library that significantly accelerates applications in deep learning, lowering the barrier to entry for a wide range of applications in the marketing space as well.

Where are we in the life cycle of AI and advertising? In my opinion, IBM/Watson/Weather Company was the first commercially recognized usage of AI in advertising.

I asked Soni what he thought about that statement, and he said, “The Weather signals were a good start, but the real gap is helping organizations understand how these practices will get applied to their businesses. People are collecting a large amount of data, but they’re still lacking the breadth and depth of talent to activate against it. Articles and presentations like these go a long way into educating the market.”

AI is just the beginning of the ad tech evolution, particularly when we consider the transformation of the living room — how folks use devices while they watch TV, the data that jumps across devices, and of course, the machines which will have to manage and parse all that data. But that’s a conversation for another article.