Problem

Bypass POS, which became a part of Clover, had plans to expand its business from the sports and entertainment venues market to fast-casual restaurants. However, fast casual restaurants usually handle bigger menus and have more flexibility than POS systems in stadiums and arenas. This required a completely different approach while utilizing as many existing services and back-ends as possible.

It's important to note that the turnover rates of employees in fast-casual restaurants are very high, with an average of 150%. This means a restaurant employing 20 people can expect to see 30 workers come and go within a year. As a result, employee training becomes a frequent and time-consuming task. Therefore, the design process focused on the need for easy learnability and speed of service.

Research & Discovery

Initial Research Questions

• What are the most critical workflows we need to support to serve the fast-casual market?

• What are possible accelerators to improve speed and accuracy?

Methodology

• We conducted ethnography research at two food chains, with three locations per chain. We spent more than 12 hours per location observing operations in the main order-taking section, drive-thru, and prep stations from store opening to closing on given days.

• We conducted remote and in-person interviews (10+ interviews) with order takers, restaurant managers, and regional managers.

What did we learn?

• To improve the user's ability to take orders quickly, the user should enter the order in the sequence that customers generally speak. Common patterns are ADD, NO, EXTRA, ON THE SIDE.

• Order Takers rely heavily on muscle memory and color recognition.

• Errors generate a lot of frustration in customers and order takers. Recovery from errors is one of the most important interactions.

• Allowing for substitutions and customizations is a crucial part of the process.

Concept exploration and iterations

We started the project by creating a blueprint of the main interface. The goal was to map the sections and features we needed to cover and then explore the primary user flows and interaction patterns. Ideally, we wanted to reduce the number of interaction patterns and transitions.

Hypotheses

• By color coding the menu sections, the order taker could recognize the position of items faster and improve muscle memory.

• By providing default options for combos and menu items, we allow the user to move faster.

• By designing the interface to reflect the natural flow of speaking and ordering to the greatest extent possible, we would decrease the time needed to take an order. This is crucial for item modifications (e.g., add tomato, no onions, extra mayo) and preferences.

Validation

For the first set of concepts, we spent 3 weeks working on main interactions and four weeks testing users via www.usertesting.com and creating new iterations of the same concept. Testing gave us a very good understanding of problems that we needed to address before deploying a prototype on-site. We conducted approximately 20 unmoderated usability sessions before testing with real customers using a prototype.

Final Delivery - Prototype walkthrough

Outcomes

We successfully demonstrated the concept of conversation ordering to the senior executive team. Our prototypes and testing showcased the potential of this new approach to streamline the process of taking orders while reducing training needs. This helped us secure the necessary budget and stakeholder buy-in to construct a fully functional prototype for two prominent food chains based in Texas.