← Selected work
Case study · Hospitality tools

Toast

One menu, every surface in the restaurant

Restaurant menus are messy, deeply nested, and change constantly. A single item has to appear correctly everywhere it is ordered, made, printed, and picked up: on the terminal, the handheld, the kitchen screen, the kiosk, the receipt, the app, and the delivery partner. I led the ground-up redesign of how restaurants build and manage their menus, and a design sprint that rethought the pickup moment itself.

Toast is the all-in-one platform that runs point of sale, online ordering, and operations for tens of thousands of restaurants.

Role
Lead Product Designer
Scope
Product design · Information architecture · Field research · Cross-surface and service design
Company
Toast
Year
2019
The refined Toast menu builder: color-coded items with a selected dish's bread, size, and allergy modifiers edited inline
Overview

A restaurant menu is deceptively complex. A single item sets its own temperature, the sides it comes with, and common substitutions like no onions, a veggie patty, or a gluten-free bun. Its price can shift too: happy-hour rates by time of day, different pricing per location across a chain, and separate prices by size. And menus nest all of it: menus hold groups, groups hold items, items carry modifiers.

I led design on a menu builder that makes that structure legible and editable in one place, then keeps every surface in sync from a single source of truth: the POS terminal and Toast Go handheld, the kitchen display (KDS) and printed receipts, the kiosk, online ordering on the web, and the Toast Takeout app. It also maps out to third-party delivery partners like Uber Eats and DoorDash.

What I led
  • End-to-end UX for menu management
  • In-restaurant field research with operators
  • Cross-surface preview before publishing
  • Information architecture for nested menus
  • A pickup experience design sprint that shipped and became a pending patent
  • Concept work for faster menu setup
4
levels of structure, from menu to modifier, in one view
1
source of truth for the whole menu, everywhere it is sold
0
re-keying between systems to publish a change
01The before

One cheeseburger, eight different pages

In the original tool, menu management was a dashboard of separate destinations: Menus, Open Items, Item Tags, Price Levels, Pre Modifiers, Advanced Properties, Price Editor, Items Database. Adding one cheeseburger, with its temperature, sides, and substitutions, meant knowing which page did what and re-entering the same thing across several of them.

Screen recording of the old Toast menu management: a dashboard of many separate links (Menus, Open Items, Item Tags, Price Levels, Pre Modifiers, Advanced Properties, Price Editor, Items Database) that a single cheeseburger edit had to be spread across
The previous experience · menu work scattered across a dashboard of separate pages
02Field research

Researched on-site, alongside PMs and engineers

I ran on-site restaurant research with my product managers and engineers, watching real customers collect orders in a working restaurant. Bringing the whole team into the field meant the people building the product saw the same friction I did, so the findings landed as shared priorities instead of a handoff.

The team interviewing a restaurant operator with clipboards in a working cafe
On-site restaurant research · with PMs and engineers
03Concept exploration

We tested a four-column flow and a tree view

We explored multiple concepts for navigating a deeply nested menu. One kept menus, groups, items, and modifiers side by side in a four-column flow. Another paired a left-nav menu tree with the item, modifier-group, and modifier sections stacked below. Testing both with operators showed where each helped, and the final design took the clearest parts of each.

Toast menu builder with four linked columns: Menus, Groups, Items, and Modifiers
Concept: four-column flow · menus, groups, items, and modifiers side by side
Single menu group view showing items, modifier groups, and modifiers with an edit menu open
Concept: tree view · a left-nav menu tree with sections stacked below
04The model

Menus hold groups, groups hold items, items carry modifiers

Required layer Optional layer Contains Shared referenceModifier points to one shared definition, counted across every menu and location
01
Menu
Required
02
Menu group
Required
03
Subgroup
Optional
04
Menu item
Required
05
Modifier group
Optional
06
Modifier
Required
07
Nested modifier group
Required
FOOD
1
Salads
Entrees
Family Size Salads
Family Caesar
Side Caesar
Grilled Chicken
3
Dressing Choice
Shared · used in 3 places
2
Sides
5
Caesar
Shared reference
Italian
Shared reference
Side Salad
Shared reference
4
Dressing Choice
Pick one with Side Salad · same shared group
Rice
Shared reference
  1. 1Menu groups and subgroups organize items into the structure guests browse.
  2. 2A modifier group can attach to a group, a subgroup, or a single item.
  3. 3Define a modifier group once, then attach it wherever it applies.
  4. 4A modifier can carry its own nested modifier group.
  5. 5A modifier points to one shared definition, so Caesar dressing is counted as one thing across every menu and location.
The menu data model · one structure shared by the back office, the POS, online ordering, and the kitchen
05One change, every device

One item, built once, right everywhere

Follow a single order through the restaurant. A cheeseburger, cooked medium, no onions, with sweet potato fries, is built once in the menu builder. From there it has to show up correctly on the server's terminal, on the handheld at the table, on the kitchen display where the line cooks it, on the printed ticket, on the kiosk, in online ordering and the Toast Takeout app, and at Uber Eats and DoorDash. Every one of those surfaces has different users, different screen sizes, and different moments of pressure, from a host at the door to a cook mid-rush. The menu builder's job was to make one change reach all of them, correctly, at once.

Cheese­burger
Cheeseburger
TEMPERATURE
Rare
Medium
Well
POS terminal
Server
ENTREES
Cheeseburger
Temperature · Substitutions · Sides
Handheld
At the table
#21400:42
CHEESEBURGER
MEDIUM
NO ONIONS
SWEET POTATO FRIES
Kitchen display
Line cook
1 Cheeseburger
Medium
No onions
Sweet potato fries
Receipt printer
Printed ticket
Menu builderPublished
New menu item
Cheeseburger
Temperature
RareMediumWell
Substitutions
No onionsVeggie pattyGluten-free bun
Sides
FriesSweet potato fries
Cheeseburger
SUBSTITUTIONS
No onions
Veggie patty
Gluten-free bun
ADD TO ORDER
Kiosk
Guest self-order
Cheeseburger
Temperature · Substitutions · Sides
Web online ordering
Laptop
MENU
Cheeseburger
Temperature · Substitutions · Sides
Takeout app
Phone
Uber Eats
Cheeseburger
Temperature · Substitutions · Sides
DoorDash
Cheeseburger
Temperature · Substitutions · Sides
Delivery partners
Uber Eats, DoorDash
One source of truth · a new item, built once, published to every surface it touches
06Preview before publish

See the change on the terminal before staff do

Publishing a menu used to be an act of faith. We added an in-product preview that renders the edited menu exactly as it appears on the POS, in the kitchen, in online ordering, and in Toast Takeout, so an operator can catch a misplaced item or a wrong price before it ever reaches a server mid-shift.

Preview mode rendering the menu on a simulated Android POS terminal in test mode
POS preview in test mode · the operator's menu as the terminal sees it
07Pickup

Finding your order without exposing who you are

Pickup was breaking down at busy restaurants. As online ordering through Toast and the Toast Takeout app grew, guests crowded the dining room waiting for their food, then scanned shelves of identical bags to find it, pulling staff away from the line to help. To make orders findable, most restaurants taped the receipt to the bag, which put each guest's full name and mobile number on display for anyone walking past.

Part of the crowd was a timing problem. Most restaurants quoted a single default prep time for every order, the same for a lemonade as for a catering order, so guests showed up early and lingered.

As lead designer, I ran a multidisciplinary design sprint with engineering and product to fix the whole pickup moment: when guests arrive, how they find their order, and what the shelf reveals about them.

We improved pickup notifications so guests could time their arrival to their order instead of to a guess, which thinned out the crowd waiting inside.

For the shelf, we gave every order an emoji, printed large at the top of the receipt, and took the phone number off. A whale is easier to spot on a crowded shelf than a name, easy to call out across a noisy counter, and means nothing to anyone it does not belong to. The idea ran through the whole service: the order flow, the notification, the receipt printer, the bag on the shelf, and the handoff. It shipped to Toast restaurants and became a pending patent, with me as a named inventor.

ORDER PLACED
Cheeseburger
Pickup
01
Order placed
Online or in the app
 
ALERT
Your order is ready for pickup
02
Notification
Arrive when it is ready
🐋
1 Cheeseburger
(555) •••-••••
03
Receipt prints
Emoji up top, no phone number
🐬
🐋
🦉
04
Bag on the shelf
Spotted at a glance
🐋Whale?
🐋
05
Handoff
Called out by emoji
Before
ONLINE ORDER
ALEX MORGAN
(617) 555-0142
1 Cheeseburger
Medium
No onions
The receipt taped to the bag: a full name and mobile number, readable by anyone walking past.
Before · the guest's name and number on the shelf
After
🐋
Alex M.
1 Cheeseburger
Medium
No onions
An emoji printed large at the top, readable from across the counter. The phone number is gone.
After · an emoji to spot, no phone number

"A whale is easier to spot on a crowded shelf than a name."

Design sprint prototype: takeout bags on a pickup counter, each with a receipt showing a large emoji (a dolphin, an owl, a pufferfish), the guest's first name and initial, and the order items
Prototype · emoji receipts tested on the pickup shelf during the design sprint
A takeout receipt from Saloniki Greek in Boston, dated 09/24/19, with a large whale printed at the top, resting on a lamb meatball plate
Shipped · an online pickup order at Saloniki, Boston, 2019, identified by its whale
08Concept

Point a camera at the printed menu

The slowest part of onboarding a restaurant is typing its menu in for the first time. I explored a concept, Toast Lens, that lets an operator scan a printed menu and turn it into a structured draft, so the builder starts from something real instead of a blank screen. It pressure-tested how far we could shrink the gap between a paper menu and a live, sellable one.

Leadership · How it shipped

One model, owned across every team

The hard part of menu management is not any one screen. It is agreeing on a single data model that the back office, the POS, online ordering, and the kitchen all trust. As the most senior individual contributor on the design team before the company's IPO, I worked closely with engineering and PM to hold that shared structure steady while each surface rendered it in the way its users needed.

Grounding decisions in field research kept those debates anchored to how restaurants actually run, not how the org chart was drawn.

Executive feedback
S
Steve, Co-founder (and original designer)

Jess just shared the Menu UX work, and it's great to see all the effort that's gone in. It seems like we've defined the problem quite well, and it's great to see the different wireframes and the customer testing helping us find the optimal designs.

Impact · Where it landed

One menu, trusted on every surface

The menu builder shipped as the single source of truth behind every Toast surface: the POS and handhelds, the kitchen display, the kiosk, online ordering, and Toast Takeout, mapping out to delivery partners. A restaurant builds its menu once and trusts it everywhere it is sold, with no re-keying between systems. The pickup work shipped alongside it, giving guests a clearer arrival time and a private way to find their order. It all shipped as Toast scaled toward its 2021 IPO, running point of sale and operations for tens of thousands of restaurants.

Next case studyGoogle Photos→ Contact ↗︎