Next year, it will be 30 years since Joe Pine wrote the book Mass Customization, in which he coined the term personalization. So much has happened in the restaurant industry over the last 30 years that it’s almost impossible to remember back to a time when companies didn’t understand what personalization was. Pine’s premise, that customers would no longer value mass produced goods and services, has borne out. Today, for breakfast, lunch, and dinner, people want their food “their way.” From Subway to Papa Murphy’s, a wide range of quick-service restaurant brands have been built around personal customization. The value that customization brings to restaurant goers is profound, the financial benefit to quick-service brands is gigantic. 

From Customization to Something Smart

But what exactly do we mean by “customization?” Pine defined it as creating a specific offering for an individual customer. His original definition for “personalization,” was to take a standard offering and change it for an individual customer. “Personalization” today often means to do things for the customer that signal to them that the company knows them and knows their preferences. 

Because of data, analytics, AI, and automation, companies can go much farther than ever before in personalizing experiences for customers. “Hyper personalization” refers to using these resources to customize every step of the customer’s journey based on past preferences and behavior. Because personalization brought tremendous value to both company and customer, many today argue that hyper personalization will produce manifold benefits to both. It will reduce time, costs, and inventory while at the same time increase value, quality, and trust. 

But hold on. There is an assumption that is being made by the proponents of both personalization and hyper personalization. We call it the past activity trap. Once companies start down the personalization track, they often stop asking the customer what they want next. There’s a whole body of mathematics that argues that past behavior is the best predictor of future preferences, and most personalization technology is based on algorithms that prioritize past activity over any other data sets. 


  • Definition: A type of customization where the company takes a standard offering and changes it for the individual
  • Example: A soft drink with the customer’s name printed on it. Sending an offer based on known preferences of the customer. 
  • Data Requirements: Segmentation and Past Preferences

Hyper Personalization

  • Definition: Using data, analytics, AI, and automation to create advanced personalization
  • Example: Being able to anticipate what food order the customer is likely to want next. 
  • Data Requirements: Past Food Preferences, Behavior Data, Channel Data



  • Definition: Allowing the customer control over the customization experience. 
  • Example: Giving customers access to their data so that they can plan food choices. 
  • Data Requirements: Past Data on Food Choices, Patterns, Location



  • Definition: Using data, analytics, AI, and automation to address the situational needs of customers 
  • Example: Anticipating customer food needs based on their circumstantial needs
  • Data Requirements: Individualization Data plus Contextual Data: Location, Time, Weather, Social, etc.


The Past Activity Trap

The past activity trap almost always includes two fatal flaws: the customer has little or no control over the data experience and the recommendations generated by the tools do not include situational analysis. A customer goes into their favorite coffee chain for a donut and hot chocolate every morning. Five days a week. Then they get married, and their spouse decides they will have breakfast together. The brand has all the past data about this customer and notices that the customer no longer comes in in the morning. So, a personalized outreach campaign begins to bring that customer back that includes rewards for donuts and hot chocolate and messaging about starting the morning off right. The company becomes hyper focused on the store the customer went to (the channel), the behavior (daily purchase), and the food items (preferences). 

How successful will the campaign be? The risks are high that the restaurant chain will alienate the customer. They clearly don’t have the data they need to adjust to the customer’s new situation. 

Imagine, however, if the brand had a tool in their app that allowed the customer to track their own activity and adjust their own preferences. Said customer can now gain control of the experience and direct the experience to their new needs. The company gives the customer data control and in return, the customer points the company in the proper direction; which could be donuts for lunch, right? On the sly. 

Context Empowers Customers and Companies

Let’s take this logic a step further. The coffee chain shares not just the purchase data with the customer but also allows for contextual data to be shared or included. Things like time of day, weather, family members, and so forth. Now both the customer and the company have a fuller picture of the need that the food is being chosen for. The customer is empowered, and the company understands the context. Now they can anticipate even better what the customer will want the next time the situation arises. 

Our fear for quick-service restaurant companies is that they will buy into the hype of hyper personalization without thinking through the design of the data experience. People want control over their customized experiences. It is arrogant to assume that the company knows exactly what the customer wants the next time. It is also naïve. 

Instead of focusing AI on hyper personalization (based in Bayesian analytics of past performance) companies should give the customer the controls and allow the AI to act as co-pilot of the experience. That’s individualization. 

Instead of building data models that focus solely on past preferences, behavior, and channel choices of individuals, companies should focus their dashboards on common situations that people generally find themselves in. Then capture data that will help them understand when a customer is in that situation. They will be far more successful at predicting what customers will want. 

Situational Analytics is Simpler

And here’s the kicker, based on our years of research into personalization, we can say that a focus on common situations that people generally find themselves is easier to execute against than trying to anticipate customer needs based on past activity. The data is generally simple to collect, simple to analyze, and simple to understand. And the customer appreciates it more.

There is nothing quite so frustrating as having a super smart tool provide the wrong recommendation. There’s also nothing quite so comforting as knowing that a company understands your situation. 

Dave Norton, Ph.D. is the Founder and Principal of Stone Mantel, a research-led consultancy at the forefront of customer and employee experience strategy. With the support of lead experience strategists like Mary Putman and Aransas Savas, they guide, research, and build frameworks to help companies like Marriott, Jason’s Deli, Coca-Cola, and O’Charley’s deliver on Time Well Spent for customers and employees. Learn more at

Consumer Trends, Fast Food, Outside Insights, Story