Photo Credit: Cara Harman for Infatuation LA

Designing Operational Systems at Mozza2Go

2026
Service Design
Case Study
Operational Systems
ABOUT
Mozza 2 Go is a high-volume, fast-paced Italian restaurant featuring to-go versions of popular dishes served at Pizzeria Mozza and Osteria Mozza. It is co-owned by Chef Nancy Silverton and Joe Bastianich, whose flagship Osteria Mozza holds a Michelin star, and is located in Los Angeles, CA.

As a shift lead, I own onboarding, training, ordering, day-to-day floor and expo systems across shifts. I rebuilt the onboarding, training, and floor-operations systems while capturing a departing trainer's knowledge, sequencing the workflow around hard time constraints, and designing a verified order-handoff system. This new operational system cut new-hire ramp time roughly in half and reduced order errors from several per shift to a few per week.

Role: Shift Lead and Trainer, Mozza2Go

The Problem:

The operation was running on implicit knowledge and improvised processes, which showed up in four concrete ways:

  1. No real onboarding system

    The only "documentation" was a 129-page employee handbook so long that no one read it — and that wasn't even reliably handed to new hires (it wasn't given to me when I started). There was no defined training program: no sequence for what a new hire learns, no sense of how many days to competence, no shared standard. Onboarding happened by improvised shadowing, so how well someone learned depended entirely on who trained them and how slammed that shift was.

  2. Inconsistency between shifts

    Without shared checklists or standards, the same task got done differently depending on who was working. No opening checklist, no closing checklist, no side-work checklist meant steps got skipped and quality was unpredictable shift to shift.

  3. A disorganized physical domain

    The workspace itself was sometimes messy and unstructured, which slowed everyone down, made mistakes more likely, and made it harder for new people to find their footing.

  4. Unreliable order tracking and kitchen handoff

    Completed tickets were thrown away, leaving no record of what had actually been made, and communication between the kitchen and the expo station was informal and easy to lose. A single misplaced ticket or a misnumbered pizza box could make an entire order quietly disappear. Errors ran around one to three per shift.



How I Figured Out What to Build:
  • Captured knowledge before it was lost - The person training me was the person leaving. I used my own opening-training sessions as a capture opportunity — doing a full walkthrough with them and documenting every step in real time, in context, while the work was actually happening. Critical knowledge that lived only in a departing employee's head got preserved instead of walking out the door.
  • Sequenced the workflow by working backward from hard constraints - I identified what was time-sensitive (example, the front doors must open by 3:55pm) and worked backward from those fixed deadlines to order tasks first to last. The sequence was built around real operational dependencies, not arbitrary staff preference.
  • Built a validated specification, not an assumption - For prep, I logged every item (dressings, salads, parmesan, basil, croutons), the specific container each belongs in, required quantities, and expiration dates — then confirmed the expiration details directly with the chefs. The result was a single, validated source of truth affirmed by the people who actually knew.



The 129-page handbook still existed (now emailed for new hires to read on their own) — but a document that long was never going to help someone in the middle of a shift. The real gap wasn't missing information; it was missing *usable* information on the floor. So I built the layer the handbook couldn't be: task-based tools designed around when and how people actually work. These mainly came from consistently observing patterns where myself and others kept stumbling.



The Buildout


1. Built point-of-use job aids

I created opening, closing, and side-work checklists so the standard for each task lives in the moment it's performed. This gave every shift the same baseline regardless of who was on and helped staff not forget tasks.

2. Designed targeted tools for specific, observed failure points

When I noticed staff repeatedly forgot or got confused about ordering gelato, I built a gelato ordering sheet listing every flavor plus clear directions for emailing the distributor to place the order. The tool came directly from a recurring real-world pain point I observed and questioned the staff about; designing in response to behavior, not assumption.

3. Built a structured training program where none existed

I defined what a new hire learns and when — a clear sequence and timeline — so onboarding stopped depending on which person happened to be training and how busy the shift was. Every new hire now gets the same path to competence.

4. Reorganized the physical workspace around function

I restructured the kitchen so everything had a defined, logical home: all prep consolidated in one area, and all administrative / front-of-house materials (cash register, books, menus, office supplies, paper, binders) in another. Grouping the space by how work actually flows reduced friction, made things findable, and made the environment itself easier for new people to learn.

5. Designed for access and adoption

Building the tools was only half the job; I made sure everyone actually had access to them, so the system got used rather than sitting unread like the handbook. While reorganizing the cashier station, I stocked it with all checklists and quick reference material.

6. Built a dual-ticket order system with verified handoffs

To stop orders from slipping through the gap between kitchen and expo, I replaced the throw-away-the-ticket approach with parallel records: the kitchen keeps its ticket, the expo station keeps a matching one, so no single lost slip can erase an order. I paired it with a structured verbal protocol — kitchen and expo confirm what's in progress, how long until it's ready, and what's coming next, and each completion is called and confirmed out loud ("#54 complete" → "heard"). On completion, the kitchen calls the ticket number, expo matches it against its own ticket and confirms the order is complete before it goes into the warmer, then a second pack-and-check happens before handoff to the driver or customer. The result is layered verification with a read-back step, so errors get caught before they ever leave the building.


The Outcome



Training Time Cut In Half

Before the redesign, a new hire could take up to a month to fully grasp the operation. Now a new hire can be trained across the entire operation from menu knowledge to opening, closing, prep, customer-service standards, and ordering in roughly two weeks.

Greatly Reduced Order Errors

The dual-ticket system and verified handoff took mistakes from roughly one to three *per shift* down to one to three *per week* — with virtually no errors on busy weekend services.

Failure Prone Middle, Closed

Everything missed in between: forgotten mid-shift prep, order mistakes, no shared customer-service standard, and dropped recurring tasks like ordering gelato or flagging a manager when supplies ran low. The redesigned training sequence and job aids target exactly that middle, so the parts that used to slip are now covered systematically.

More Consistent Service Across Shifts

Shared checklists and a single validated prep specification meant the same standards held regardless of who was working, reducing the inconsistency and skipped steps that came from improvisation.

An Environment That Supports Work

Reorganizing the kitchen by function made tasks faster to perform and the whole operation easier for new people to learn.


Overall Results:

  • Identifying a systemic problem affecting real users and designing an intervention for it
  • Building standards and processes where none existed (systems thinking)
  • Implementing change in a live, high-stakes environment without disrupting operations
  • Iterating based on observed results and patterns