two marble staircases
PHOTO: Christopher Sessums

The prospects of what artificial intelligence (AI) can do for marketing and other business initiatives is well-documented. McKinsey & Company, for instance, reported in a study that AI improved on traditional analytics techniques in 69 percent of potential use cases. 

The real challenge is how to actually implement AI in your marketing processes and where you begin. Marketers are asking that question over and over, according to William Ammerman, executive VP of Engaged Media. “That's their first question: what do I do now?” Ammerman told CMSWire when asked common questions that marketers ask around AI. “What they're looking for is practical, pragmatic advice on what to do right now.”

The first steps? Knowing if your marketing processes are even a candidate for AI to begin with and identifying a targeted business use case. This is not to mention testing your original hypothesis and knowing when to abandon the project if it’s not on target, according to Cal Al-Dhubaib, managing partner of Pandata

We caught up with Ammerman and Al-Dhubaib about things marketers should know about making the investment into AI for marketing processes.

Gaining a ‘Deep Grounding’ in Al Algorithms

The first thing marketers can do is gain a deep grounding in how to use data to identify audiences, Ammerman said. “And that’s been a real challenge for marketers,” he said, “but increasingly we can use algorithms to define audiences automatically.” Sure, marketers don’t need to be the data scientists, but they will have a seat at the table so any education is a bonus.

But can marketers even begin to understand the nuances that come with AI, algorithms and data science? Although they’ve been great with managing brands through the era of digital transformation, some CMOs today have a fear they are not worthy when it comes to data science and AI knowledge, Ammerman said. “Where in the process have marketers had a chance to get formal education in algorithms and algorithmic thinking?” Ammerman said. “How do they start thinking strategically about the kinds of data they want to capture as a corporation? These are really critical conversations for CMOs to be having right now.” 

Also in the learning department, Al-Dhubaib said marketers need to be intimate with data literacy and understand there's a data tagging process involved where categories of data need to be defined and consistent. “They need to understand where data lives, and how quality is maintained within those relevant data sources,” Al-Dhubaib said. “Because if you do not trust the data, you'll never trust the outputs of data.”

Related Article: Cutting Through AI Marketing Hype: It's About Machine Learning

Start Small, Targeted

Marketers should know that with AI, you should be starting small. This isn’t an organization-changing, 180-degree marketing flip. “What you're trying to do is you're trying to solve very specific, very tangible business challenges in a data-driven way that ultimately informs the decisions that have an impact on your bottom line,”  Al-Dhubaib said. “There is no AI fairy-dust you’ll sprinkle and make it all run better. It's got to be targeted.”

Starting small is important because you can prioritize an easy return on investment. “I think the mistake is we try to get AI to do everything," Al-Dhubaib said. "That is a surefire way to lose trust within your organization and set yourself up for failure." Gain confidence in your efforts on a small scale, and push for bigger initiatives later, he added.

Ask Yourself: How Can AI Help?

Determine what can be specifically improved with AI by asking yourself this question, according to Al-Dhubaib: “If I could identify (blank), I could (blank).” He used this example: “If I could identify when a customer becomes a higher risk for churn, I could alert our retention team to intervene.” 

“It’s really important to start by understanding that decision that you're trying to influence with your artificial intelligence solution: who is touched by that decision, who are the stakeholders, and whose voice needs to be heard as part of this?” Al-Dhubaib said. “Because when you do that, you're going to start understanding the different assumptions or nuances that you need to take into consideration when you're evaluating tools.” 

Related Article: 3 Misconceptions About AI in Marketing

Don’t Let AI Be All About the Data Scientists

Many AI projects fail because organizations forget about actual stakeholders behind the marketing programs, according to Al-Dhubaib. “I can't stress this enough the stakeholder doesn't just pop in and out of the data science team; they need to be on the data science team,” he said. Dhubaib said subject matter experts in the trenches can help data science team members deal with nuances and anomalies AI detects in the data.

Set up Your Testing Checkpoints

Before you get up and running with deployment, you should have checkpoints in place to see if your original goals are realistic, Al-Dhubaib said. “I think we're at a point now where most people can accept the fact that artificial intelligence is experimental,” he added. “You're gonna have to go through some trial and error. … There needs to be a distinct test and proof of concept before you do anything. Start measuring it to see if it’s giving you the return that you expected it to give you before investing further.”

Related Article: Marketers: Don't Fit Your Strategy to AI, Fit AI to Your Strategy

Questions to Ask

To summarize, Al-Dhubaib suggests marketers ask some very important questions before starting with AI in a small marketing segment:

  • What are the decisions you're going to influence? Who's going to make the decisions? 
  • Can you quantify the potential return in advance? Is it worth the investment? How are you going to test the solution? What's your stake in the ground? 
  • Who will the solution affect? What regulatory risks are posed?

Ethics questions are “really, really important because oftentimes there are unintended consequences to using Artificial Intelligence.”