How Restaurant IT Decision Makers Should Assess AI Technology Solutions for Their Marketing and Loyalty Programs

You may have even seen that some restaurants are exploring artificial intelligence (AI) technology that claims it can predict your order simply based on what you look like. These machines use cameras to analyze a customer’s appearance, assume traits like age or gender, and display menu items they believe that customers will be most likely to order.

Regardless of if that type of facial recognition intelligence is worrisome or not, the fact is, machine learning (ML) and AI continue to create buzz in the restaurant industry. But many technology vendors play fast and loose with those terms, crossing their fingers that potential customers will buy into the hype without actually pressing into what those words mean, how vendors are using them, and if they are even successful.

Let’s start with the basics. What are artificial intelligence and machine learning? AI leverages computers and machines to mimic the problem-solving and decision-making capabilities of the human mind^[1]. Machine learning is a branch of AI and computer science that focuses on the use of data and algorithms to imitate the way that humans learn, gradually improving its accuracy^[2].

When done correctly, AI and ML are powerful. But, successful execution requires tons of accurate data and even more diligent monitoring and maintenance of the algorithm’s quality over time to avoid programmatic bias and to ensure applicability and accuracy. And the reality is, many technology providers fail at one or all of these requirements.

A tremendous amount of data is needed to effectively train a AI/ML model.  Historically, the restaurant industry has struggled with data capture, but digital experiences and e-commerce are changing all that. COVID has accelerated customers’ adoption of digital channels; and with this digital shift, restaurants have the opportunity to capture more data than ever before. Data from Deloitte shows more than 57 percent of consumers now use a digital app to order restaurant food for off-premises dining and 64 percent, nearly two-thirds of consumers, prefer to order their food digitally.

So, can AI/ML help restaurants? The answer is: it depends. Avoid the sleight of hand companies use when claiming they offer AI or ML as a marketing ploy. Here are a few questions that should be asked when evaluating a technology vendor advertising AI or ML capabilities:

• Does it work?
  • What percent of the necessary outcomes can be described by the model (for example, when used to predict customer lifetime value, what percent of your customers’ lifetime value can be accurately predicted)?
  • What is the accuracy and precision of the model itself (how often does the model get within a reasonable range to be considered correct)?
  • How often does the model need to be retrained to maintain accuracy and what commitments can the company provide that they will maintain retraining?
  • How much data is needed to train the model? What data is being used to do so? How can we be sure that biases won’t be introduced?
• Can we monitor that it’s working?
  • How can we monitor performance and accuracy over time?
  • What information is available for monitoring model health?
• Can you provide a customer reference?
  • How many of your clients are using the model in a production environment and what business outcomes have they seen?
  • What customer proof points can you share about the specific outcomes that the clients have seen?

While some vendors claim they can predict customer preferences and future lifetime value based on an algorithmic approach, oftentimes, more simple approaches have higher efficacy. For example, on the Thanx platform, a one percent increase in conversion results in a $25 increase in average revenue per customer. In fact, restaurants that leverage the growth of their CRM from their loyalty program to deliver targeted, personalized campaigns to their guests see a 6-times increase in revenue per customer versus generic, “send to all” campaigns. That is real ROI without the gimmick.

And all without the risk of misgendering or guessing the wrong age of your customers, which frankly, sounds like a terrible customer experience as well as a potential PR nightmare.

^[1] IBM Cloud Learn Hub “Artificial Intelligence”

^[2] IBM Cloud Learn Hub “Machine Learning”

Emily Rugaber serves as VP of Marketing at Thanx, the leading loyalty, CRM and guest engagement platform for restaurants. She has spent her entire career in the tech industry working across a variety of industries, consulting with large companies including Target, Nestle, Virgin America and SAP on business intelligence projects aimed and mining data for actionable insights As the VP of marketing at Thanx, Emily leads with a deep knowledge and understanding of loyalty trends, innovations, and best practices for enterprise restaurant brands. Emily is the author of Thanx’s Loyalty Disrupt newsletter.