How AI-Driven Menu Clustering Can Drive Restaurant Sales

For restaurant and convenience store marketers, selecting campaigns is challenging. They must consider the cost of messaging guests as well as the discount or reward to be included. They also must weigh the relevance of the message and the final monetary goal.  

An offer that’s too broad can break the campaign budget, but an offer that is too narrow is in peril of low engagement.  

Consequently, marketers need to pair their intuition with something more concrete when trying to zero in on the campaign sweet spot. Artificial intelligence (AI) is a concrete tool that enables marketers to leverage all of the data at their disposal to achieve the desired results: 

1. Reduce the cost of the offer and drive a bigger ROI.  
2. Enable more personal, relevant, and targeted one-to-one offers for every guest. 
3. Easily tweak and replicate these offers to save time. 

With AI, marketers can use the mountains of data collected through their loyalty program to run campaigns that target guests based on their unique behavior, ensuring relevancy, cost-savings, and ultimately a huge return on their initial investment. At Paytronix, we refer to this as “AI to IA”—the practice of using artificial intelligence to drive individual action. 

This does not mean that the most successful marketing departments are run by computers. Instead, AI is an asset that helps marketers identify underlying patterns and make reasonable predictions about guest behavior. Although a brand could choose to let the AI determine every piece of the campaign, from guest segment to offer delivered, it’s more common to see marketers use AI to improve their own decision making.  

A human touch is often still beneficial. AI can optimize for any variable, but the mind of the marketer should guide which variable that ought to be. 

For instance, an algorithm built to segment alcohol purchasers might identify the optimal group as those who have purchased alcohol four times in the past month, but that group may be too small to truly move the needle. The marketer could instead opt for a different segment of guests who purchased alcohol twice in the past month, because it targets a significantly larger group.  

Ultimately, the best results come from humans and AI working together.  

Clustering 

Clustering, or k-means clustering, involves one of the most popular machine learning algorithms and provides a great example of driving AI to IA. In simple terms, the AI identifies points in a dataset and then creates clusters of the nearest ones while keeping each cluster as small as possible. In other words, it’s a method of grouping data points together to identify similarities that aren’t immediately apparent.  

But what does that mean in practice?  

One application of clustering enables marketers to choose an item or category they’d like to grow—say, a new burger on the menu or a high-margin offering like alcohol—and find what purchases motivate the guests who are buying it. Those guests can then be clustered into groups and sent offers that are relevant to their specific buying habits in hopes of enticing them to purchase the designated item.  

Let’s take a look at a use case.  

Creating Clusters for a Burger Joint 

“Menu clustering” exemplifies how this process works. It’s the practice of using AI to segment guests into groups based on their prior purchasing behavior. There are four main use-cases of menu clustering:  

1. Push members into desired behavior. 
2. Encourage members to continue existing behavior. 
3. Position new products correctly. 
4. Pair the right reward with the right guest. 

In this case, the brand found four key segments: guests who buy meals, guest who buy fries, guest who buy milkshakes, and guest who buy vegetable-based items. Then, the brand brainstormed offers that would drive each segment’s guests to a desired behavior.  

The brand was introducing a new, limited-time-offer burger. Armed with the segmentation information delivered by the AI, it launched four unique campaigns to provide guests with an offer that made sense for them. 

Using the guest segments, the burger brand’s marketing team crafted unique offers that catered specifically to the guests in each segment and achieved each menu-clustering objective:  

1. Push members into desired behavior. Guests in the meals segment always add fries, but only sometimes add a drink to complete the meal. Since these guests are being rewarded with double points, they’re more likely to add the drink to their order.  
2. Encourage members to continue existing behavior. Milkshake purchasers are high-value guests because they’re adding a dessert item for a bigger basket size. The brand encouraged these guests to continue buying shakes by offering them a reward on the third purchase.  
3. Position new products correctly. Guests who were identified only for regularly purchasing fries probably exhibit a fair amount of variance in the rest of their order, making them more likely to try new things. By enticing them with a free side, the brand encouraged these guests to try their new burger.  
4. Pair the right reward with the right guest. A message promoting the new beef burger would be off-putting to vegans and vegetarians. Because the brand knew the eating habits of its members, it was able to offer this group a relevant reward—a free veggie burger.  

Now, for the ultimate question: Were these campaigns successful?  

Yes! The campaigns, which only ran in-store, generated a whopping 61-fold ROI. 

The Bottom Line

Clustering enables brand marketers to understand their guests better and deliver customized, one-to-one communications and promotions. AI-to-IA campaigns can save money because they cast a smaller net to a more motivated set of guests, providing offers of greater relevance to them. In turn, that relevance drives guests to act with more urgency and visit more frequently.  

AI-to-IA campaigns also uncover behavioral patterns that may not be apparent to the human eye. This empowers marketers to communicate with guests in new and exciting ways.

Data analytics expert Lee Barnes leads the Data Insights team at Paytronix, a leading provider of reward program solutions whose guest engagement platform helps more than 300 restaurant and retail chains manage and grow more than $18 billion in guest spend. Barnes is a self-confessed data geek that can often be found digging into the data with his team to optimize guest engagement with more than 165 million loyal guests—through mobile, social and today’s most innovative digital marketing tools. His mathematics degree and his MBA from Harvard Business School gives him the unusual ability to both execute complex analyses and translate the results into ideas that business leaders can use.