Two of our leading subject-matter experts Christine Livingston, AI Center of Excellence Director and Chief AI Strategist, and Jim Hertzfeld, Digital Chief Strategist and principal of Perficient Digital’s Strategy & Innovation team, recently shared their expertise with eMarketer to weigh in on five best practices marketers and advertisers can follow for AI implementation and machine learning in 
their operations. AI is already disrupting core functions within the marketing and advertising world with things like ad targeting, media buying, content creation, and propensity modeling. But knowing where to begin can be difficult. Here are the five best practices to follow with insights from our Chief AI Strategist and Digital Chief Strategist.
1: Define the Problem First
When making a case to implement AI into your operation you should first a) consider if your problem can be solved with more traditional computing and analytics methods and if it can’t then b) create a clear business use case that defines the problem. Once you’ve accomplished those tasks it’s smart to start small. This strategy ensures that you’re moving in the right direction and helps manage expectations.
“One of the most common things we see people doing is trying to take on too much in their first implementation or their first phase,” said our Chief AI Strategist. “They’re looking at the holistic problem and trying to address it beginning to end. But we encourage them to break it down into digestible pieces.”
2: Pick the Right Tools for the Job
Figuring out how to implement AI or machine learning can be tricky, but it’s not impossible. While there are a myriad of options (free, build-it-yourself, open-source, SaaS, turnkey, etc.) there are also consultants who can help guide you and craft an AI strategy that is unique to your business case.
“We help identify when it makes sense to use AI, if we should be training our own model or working with another provider, and where the data lives,” Livingston told eMarketer.
3: Convene the Experts
Another critical component for successful AI implementation is to get buy-in from the people within your organization. Education and involvement early on can help avoid implementation pitfalls that are rooted in employee fear, skills mismatches, and inadequate training. Our Chief AI Strategist agreed that managing change is absolutely necessary.
“A lot of times, the same people who are afraid of automation are the people you need to train your systems and model,” Livingston told eMarketer. “So we bring in our change management group early and educate these people on the reality that there are very, very few roles that have been eliminated through [AI].”
4: Bring Together the Right Data
Formulating a data strategy that brings not only the right kind of data, but also the right amount of it is important to the success of the implementation. Without data you can’t obtain insights or predictions. And collecting data is often where projects get stuck. So a best practice is to use machine learning models on what you already have and move forward from there.
Jim Hertzfeld agreed. “There’s a phrase, ‘think big, start small, act fast,’ and there’s a good start in using the data you already have. The value curve is very steep at the beginning, so you don’t have to do as much to make a big impact early.”
5: Future-Proof the System
AI systems aren’t “set it and forget it” projects. While AI can assist and augment human work, it still requires fine-tuning, updates, monitoring, and measurement to gauge success by humans to ensure the system is operating as intended.
Learn More About AI
Learn more about AI from our Chief Strategists and the other industry experts by viewing the full report. You can also schedule a meeting to speak to our Chief Strategists to discuss things like AI readiness, use cases, road mapping, strategy, and more.
