© 2019 by Elaine Schertz

Chef Help

Real-Time Restaurant Inventory Management

An inventory system consisting of speech and touch input methods combined with a tablet-based feedback application that enhances the ability of chefs and kitchen mangers to track inventory usage.

 

Team Members: Sarah Brooks, Cooper Colglazier, Hanyu Gong, Elaine Schertz and Yan Tan

 

My Contributions: User Research, Speech and Touch Input Prototypes, User Evaluation, Project Manager

Duration: 3 Months

Research

Semi-Structured Interviews | Observation Sessions | Survey | Competitive Analysis

Research Questions

 

Our original aim was to investigate the potential pain points that chefs and kitchen managers encounter when ordering food for the restaurant. However, after our initial research, we realized that the larger problem was not with the systems that restaurants used to ordered supplies, but with users knowing when and how much to order.

 

We found that over 40% of restaurants use only pen and paper to track their inventory. This process of manually tracking the inventory takes restaurants more than 5 hours per week and does not capture any information about inventory shrinkage (the difference between what is purchased and sold by the restaurant).  The profit margins in the restaurant industry are tight, and food costs can run as high as 38%; this means that there was a missed opportunity for the restaurants to analyze their inventory data in a way that would allow them to eliminate food waste. Thus, our research questions evolved to focus on the inventory process.  ​

Current Inventory Methods

What is the process for determining the current inventory status?

2

What are the barriers to using digital tracking methods?

3

What inventory information is most valuable to the users?

User Needs

 

Through our research, we found that any useful restaurant inventory system must allow users to:​​

1

Know when an ingredient is below the minimal operating value

2

Discover and eliminate sources of inventory shrinkage​

3

Track inventory without interrupting the kitchen workflow

Based on our interviews with various users and stakeholders, we developed the following personas to help guide our design.

 
Design Requirements

 

Our research further helped us to identify a few key design requirements:

  • Easy to learn, requiring minimal training due to high staff turnover rates

  • Efficient to use, minimizing interruption to the primary workflow of kitchen employees

  • Sanitary for use in kitchen environments

  • Affordable for smaller, family-owned restaurants

 

Design​

 
Ideation

 

We gathered as a team and brainstormed ~20 different solutions to address our users’ needs. From these ideas we created 4 different design alternatives with varying degrees of functionality and complexity. After getting feedback from industry experts and comparing the strengths and weaknesses of each of the alternatives, we ultimately decided to prototype two real-time inventory tracking methods and one feedback application for the tablet system. Users preferred this alternative because it was comprehensive, simple and affordable.

See below for the early concept sketches, wireframes and storyboards of the tablet system alternative.

Tablet system feedback application wireframe

Tablet system feedback application wireframe

Tablet system touch-input mockup

Tablet system concept storyboard

Prototypes

 

A teammate and I used Sketch and Invision to create an interactive version of our design alternative. Both systems were developed enough to allow users to complete specified benchmark tasks. I developed the speech input method as a simple Alexa Skill for initial testing.

Since one of the largest contributors of inventory shrinkage in the restaurant industry is theft, our idea for the touch input tablet was that users would first enter a unique employee passcode. This would identify the user to the system, increasing employee accountability and forcing any lies to be made explicit. Users would then make selections from the recipe or ingredient pages to indicate what they were removing from the inventory; they would then press enter to open the inventory room.

Inventory Input Code

Selected Recipes Screen

Inventory Recipe Screen

Inventory Note Screen

Inventory Recipe Selection Screen

Inventory Thank You Screen

 

For the speech-based input method, users would simply say, "Alexa tell chef help I am taking..." and then specify the item and quantity that they were removing. Alexa was programmed with two different responses, one a simple affirmative and one which repeats the item name and quantity back to the user, in order to test how users felt about the speed versus accuracy trade-off. Watch below to see an early prototype of the Alexa skill.

Selected screens from the feedback application, which allows kitchen manager to import paper inventory sheets, track the inventory in real-time and monitor employees are shown below.

Inventory feedback initial setup

Inventory overview, editing and activity monitoring

Purchase list screens

Evaluation​

Preference Testing | Usability Testing (Think Aloud and Benchmark Tasks) | SUS

Evaluation Findings​

 

In order to evaluate the two input methods, we had users complete the following tasks:

Task 1: Enter the inventory code

Task 2: Remove four batches of pancakes from the inventory

Task 2a: Make a note that there were portioning problems with the pancakes

Task 3: Remove six batches of fried okra from the inventory

Task 4: Remove two dozen eggs from the inventory

Task 5: Remove nine pounds of bacon from the inventory

Task 5a: Make a note that the bacon was burnt

These tasks were selected to test whether the interfaces allowed users to quickly and accurately update the inventory in real-time as they were taking the items out of storage. While using the tablet input, we had the users think-aloud to help us identify problems with the design. For the Alexa skill, we simply had users discuss how they felt after completing the tasks. Additionally, we asked a few open-ended questions about each method and had users compare the two experiences.

There was no clear winner between the two input methods, with users seeing different advantages for each system depending on the restaurant's individual situation. For the speech-based system, users unanimously preferred that Alexa respond with the item name and quantity even though this response took more time. For the tablet-input, the indication of two separate recipe and ingredient pages needed to be more obvious.  An overview of the outcomes is shown below.

 

Users were asked to complete several tasks using the feedback system, a few of which are shown below.

Task 1: Determine which ingredients are below the par

Task 2: Add “chicken breast” into the purchasing list

Task 3: Assuming you have received 60 lbs chicken breasts, mark the purchase completed and check if the inventory has been updated

Users were asked to think-aloud as they completed the tasks. After completing the tasks, users were asked a few open-ended questions about their experience.

Overall, most users thought that the feedback system provided useful information although some had concerns about how successfully the entire system could be incorporated into the kitchen environment. The feedback system received a SUS score of 72.5 with many improvements possible, a few of which are discussed below. 

Design Recommendations

  • Audio system should always respond with both item and quantity being removed from the inventory, even if this increases feedback time

  • The ability to swipe between the recipe and ingredient pages on the tablet input system should be made more obvious with an arrow and page indicator

  • In feedback application, inventory items should be editable by clicking on the item card directly

  • Feedback purchase list functionality should be more explicit, perhaps by changing the page to 'Orders' and sub-dividing the page into Purchase List, Pending Orders and Complete Orders sections

Links

Home | Self-Service Gas Pump | Tech Fair | Resume