Advertisers know that artificial intelligence holds great promise, that it’s becoming pervasive as a way to maximize the results they can get from programmatic advertising.
What they don’t necessarily know is how to use AI to achieve those goals. Many don’t know even how to take the first steps in bringing AI into their advertising strategies.
These were just some of the pointers that speakers shared at a recent Advertising Advertising Week panel in New York which explored big data and AI and the ways to use them to improve brand marketing.
Some of the hesitation around AI comes from “fear,” said panelist Diaz Nesamoney, Founder, President & CEO of Jivox. To them, AI “is this black box” and it’s difficult to “understand what it’s actually doing.“
What it is doing at its core, he and the other panelists agreed, is increasing the ability to quickly and deftly handle and process the masses of data now available for advertisers.
“Years ago, we said we don’t have enough data. Now, we have too much,” Nesamoney said. “It’s really about a level of precision that AI can give you.”
Start with the questions
One big trick in getting started with AI is to figure out what you want it to do.
Start not with the technology in mind but focus instead on desired business outcomes, and work from there, the panelists agreed.
Kathleen Comer VP, Client Services at The Trade Desk, gave the example of a quick-serve restaurant that wanted to encourage their most profitable customer segments to come visit.
They hypothesized that messaging a core customer group, families, would best boost sales. After launching the campaign they found through matching the advertising data with offline attribution from a partner that they were scoring even better with college students who were coming into their establishments late at night.
“They started to build out cohorts of college-age audiences, and a bidding model against that,” Comer said. Soon their business shot up.
That data may have been available before AI and been processed by people crunching numbers who delivered their analyses weeks after the campaign had finished. AI let them adjust and optimize the campaign mid-stream — and reap the rewards.
Allow for surprises and pivots
“When you have the right rules set, the performance can hockey stick up,” Comer said.
Setting those rules might mean letting go of some preconceptions, or at least allowing for the fact that hypotheses are made to be challenged.
Brand managers for another advertiser, a CPG food company, thought their messaging would be most effective when they bought placements based on an audience-based targeting model. The brand managers knew which segments liked what product, and how to reach them.
They were surprised to learn that time-of-day messaging matched to the meal at hand worked even better and were able to greatly maximize the effectiveness of their spend by adjusting to what the data were telling them.
“People said, ‘I’ve run this brand for 15 years, I know what I’m doing,'” Nesamoney said. “But when they saw what the data could do, they said, ‘I’m going to do this [other thing] more.’“
Conduct a data audit
The panelists also concurred that AI’s most powerful capabilities are unearthed when brand marketers feed in first-party data.
Businesses with strong customer touchpoints such as retailers, restaurants, and car dealerships tend to have rich data on who their customers are, their buying habits and preferences, their locales, what incentives might motivate them, and more.
The hard work comes in getting ahold of it all — from CRM systems, websites, POS databases, frequent shopper cards, and other online and offline sources — normalizing it, and figuring out how to get it to interact properly when mashed together.
Xaxis CEO, North America, Matt Sweeney suggested conducting a data audit. He advised advertisers to work with data scientists and engineers to have them help use data to answer questions, then apply machine learning to better hone in on goals.
AppNexus Director of Data Science Abraham Greenstein agreed: “Get the pipes going. Get your data in, and understand how to get your data out. Think quantitatively about metrics.”
He and Sweeney advised looking well beyond the simple clickthrough to multiple steps that interrelate and together indicate how a customer is getting closer to taking actions that are favorable for the brand.
Stay on top of strategy
What artificial intelligence can’t do is the higher level tasks that people with real intelligence can.
AI augments people, elevating marketers to stay focused on the higher level work; the strategies that lead to the business outcomes they want their marketing to achieve. It handles repetitive and rote functions at massive scale and speed but doesn’t have empathy or judgment.
“Conversations around AI are not about pressing the easy button,” Sweeney added. “It takes a lot of smart people doing a lot of smart collaboration.”
It’s no surprise that every brand Nesamoney has spoken to wants to use AI if they aren’t doing so already.
They just have to take the first steps.