ASSISTANT OPERATIONS MANAGER OPERATING SYSTEM
THE ARCHITECTURE
You own one location. The results are yours. The DM sets the
targets, provides coaching, and holds the gate. You run the
kitchen.
This is not a checklist. It is the operating system for how
you run a ghost kitchen through shift leads and a team, with
a DM who checks in weekly and visits periodically.
FOUR MODES:
┌──────────────────────────────────────────────────────────────┐
│ │
│ 1. ON-SITE OPERATIONS What you do when you are at the │
│ kitchen. Three operations. 2-4 hrs. │
│ │
│ 2. OFF-SITE MANAGEMENT What you do when you are not there. │
│ Data, scheduling, ordering, admin. │
│ │
│ 3. WEEKLY MEETING 15-20 min with the DM. Prepare for │
│ (with DM) it. The meeting is your calibration │
│ point. Not a report-out. │
│ │
│ 4. SHIFT LEAD How you develop your SLs so the │
│ DEVELOPMENT kitchen runs without you present. │
│ │
└──────────────────────────────────────────────────────────────┘
YOUR LOCATION CONTEXT:
┌──────────────────────────────────────────────────────────────┐
│ Reports to: Lydell Tyler, District Manager │
│ Shift Leads: 2-3 │
│ Team Members: 12-14 │
│ Hours: 10 AM - 11 PM (close by midnight/11:30) │
│ │
│ PRIMARY METRICS (your targets) │
│ ┌───────────────┬────────────┬───────────────────────┐ │
│ │ Error Rate │ <= 0.75% │ 30-day rolling avg │ │
│ │ Star Ratings │ >= 4.8 │ 30-day rolling avg │ │
│ │ Prep Time │ <= 9 min │ 30-day rolling avg │ │
│ │ Food Safety │ >= 90% │ monthly score │ │
│ │ Labor % │ ~17.5% │ of projected sales │ │
│ └───────────────┴────────────┴───────────────────────┘ │
│ │
│ ADMINISTRATIVE METRICS (your hygiene) │
│ ┌───────────────┬────────────┬───────────────────────┐ │
│ │ Schedule │ 100% │ published Thu night, │ │
│ │ │ │ 2 weeks out │ │
│ │ Task list │ >= 90% │ on-time, mgmt days │ │
│ │ Admin items │ >= 90% │ within 24hrs mgmt day │ │
│ │ Punch audit │ 100% │ within 24hrs next │ │
│ │ │ │ mgmt day │ │
│ │ Accountability │ 100% │ all incidents logged │ │
│ │ Escalations │ <= 1/week │ preventable non-urgent│ │
│ └───────────────┴────────────┴───────────────────────┘ │
│ │
│ DATA SOURCES │
│ 7shifts schedule, hours, punch data, labor budget │
│ Platform/POS real-time errors, ratings, prep time │
│ MarketMan inventory, orders, variance │
│ Google Sheet par levels, ADU, order columns │
│ Telegram bots inventory, prep, accountability, par, audit │
│ Dashboard training guide: home, analytics, scorecard │
│ │
└──────────────────────────────────────────────────────────────┘
MODE 1: ON-SITE OPERATIONS
When you are at the kitchen, you are not there to "be present."
You are there to run three operations that cannot be done from
data. If you finish in 3 hours, leave. If you need 6, stay.
The operations completed are the metric, not the hours.
WHEN TO BE ON-SITE:
- Post-weekend (Monday or first day after weekend). Weekend
drift surfaces here. Prep sets the week. Close compliance
from Friday/Saturday/Sunday is visible.
- Midweek peak (Wednesday or Thursday peak window). See the
constraint live. Data tells you WHAT is slow. The floor
tells you WHY.
- Third day (if results demand it). Not a punishment. Floor
reps build the reads you need to manage from data later.
┌──────────────────────────────────────────────────────────────┐
│ │
│ OPERATION 1: MEASURE (30 min, first thing) │
│ │
│ Walk every station before peak. Not a social visit. │
│ A measurement pass. │
│ │
│ CHECK: │
│ □ Each station set up to spec (mise en place, labels, temp) │
│ □ Prep list posted and tracking (done vs not done) │
│ □ Walk-in/fridge temps logged (food safety, non-negotiable) │
│ □ Previous shift close compliance (was close done right?) │
│ □ Par levels vs actual (any items close to 86?) │
│ □ Date labels current? FIFO followed? │
│ □ Handwashing station stocked? Gloves available? │
│ │
│ This is not a checklist to file. It is the raw data that │
│ feeds your COMPARE and CORRECT operations during peak. │
│ Without this walk, you are reacting, not leading. │
│ │
│ OUTPUT: Mental map of what is on track and what is not. │
│ If something is off, note it. Do not fix it yet. See if │
│ the shift lead catches it. Their detection IS the system. │
│ │
├──────────────────────────────────────────────────────────────┤
│ │
│ OPERATION 2: COMPARE + CORRECT (during peak, 1-2 hrs) │
│ │
│ Be on the floor during peak. Not working a station. │
│ Walking. Watching. Your job during peak is to see the │
│ constraint in real time and decide whether to act. │
│ │
│ WHAT TO WATCH: │
│ │
│ FLOW: │
│ - Where are orders stacking? (the bottleneck) │
│ - Are orders moving through without backtracking? │
│ - Is expo keeping up or falling behind? │
│ - Is anyone waiting on anyone else? (dependency = slack) │
│ │
│ PEOPLE: │
│ - Is the SL walking the line or stuck at a station? │
│ - Is the team communicating? (callouts, ticket calls) │
│ - Anyone struggling? (slow, frustrated, making errors) │
│ - Anyone idle while others are buried? (rebalance signal) │
│ │
│ QUALITY: │
│ - Are builds matching spec? (portion, presentation) │
│ - Are errors caught before they leave? (expo check) │
│ - If errors are leaving, which station? Which person? │
│ - Is it one person or systemic? (training vs process) │
│ │
│ WHEN TO ACT: │
│ │
│ SL not seeing the bottleneck (5+ minutes): │
│ Redirect the SL, not the team member. │
│ "Hey [SL], what's happening at [station]?" │
│ Let the SL fix it. If you fix it directly, the SL stops │
│ looking for it next time. │
│ │
│ Error pattern emerging (3rd same-type error): │
│ Pull the person off the line for 60 seconds. │
│ "Show me how you build [item]." Watch. Correct once. │
│ Put them back. Check 20 minutes later. │
│ │
│ Team member struggling (slow, lost, frustrated): │
│ Do NOT take over their station. │
│ Ask SL: "What do you need [person] to focus on?" │
│ SL coaches. You verify later. │
│ │
│ WHEN NOT TO ACT: │
│ - SL is handling it (even if slower than you would). │
│ - Bottleneck resolving itself within 10 minutes. │
│ - Single error (one mistake is noise, not pattern). │
│ │
│ THE RULE: Your job during peak is to see the constraint │
│ and coach the SL to fix it. Not to fix it yourself. │
│ Every time you jump in, the SL learns to wait for you. │
│ │
├──────────────────────────────────────────────────────────────┤
│ │
│ OPERATION 3: COACH ONE THING (15 min, after peak) │
│ │
│ After peak settles, pick ONE person or ONE process from │
│ what you observed. Have a 5-minute conversation with the │
│ shift lead about it. │
│ │
│ FORMAT: │
│ "During peak I noticed [X]. What did you see?" │
│ Listen. Then: "What would you do differently next time?" │
│ │
│ NOT: "Here is what you need to do." │
│ The SL's answer builds THEIR reads. Your lecture builds │
│ nothing. │
│ │
│ ONE THING. Not three. Not a debrief of the whole shift. │
│ One observation. One question. One answer. Done. │
│ │
│ This is how you cascade the DM's system one layer down. │
│ The DM coaches you in the weekly meeting. You coach the │
│ SL on the floor. Same method. Same structure. The system │
│ replicates itself through layers. │
│ │
└──────────────────────────────────────────────────────────────┘
ON-SITE TIME TOTAL: 2-4 hours per visit.
Measure (30 min) + Peak overlap (1-2 hrs) + Coach (15 min)
+ buffer/transition = 2-4 hours.
Not a full shift. Not 8 hours. You are there for operations
that require eyes, then back off-site for analysis and admin.
Exception: During training (pre-graduation), on-site time is
higher because you are building the reads that let you manage
from data later. See Mode 5B in the DM OS for the training
phase protocol.
MODE 2: OFF-SITE MANAGEMENT
When you are not at the kitchen, you lead through data and
systems. This is where the administrative metrics live.
DAILY (15-20 min, every day):
┌──────────────────────────────────────────────────────────────┐
│ │
│ DATA REVIEW │
│ │
│ Pull yesterday's numbers: │
│ - Error count + error rate (platform dashboard) │
│ - Star ratings (platform dashboard) │
│ - Prep time (platform dashboard) │
│ - Any 86 events? (Telegram bots / MarketMan) │
│ │
│ COMPARE to your targets: │
│ Error rate <= 0.75%. Star rating >= 4.8. Prep time <= 9min. │
│ │
│ IF ALL CLEAN: done. Move to admin. │
│ IF FLAG: note it. What metric? What day? What time window? │
│ DO NOT text the team about it immediately. Bring it to │
│ your next on-site visit or the DM meeting. Exceptions: │
│ - Star rating below 4.5 (reputation damage active) │
│ - 86 event (supply chain, time-sensitive) │
│ - Food safety violation in data (temp logs) │
│ │
│ ADMIN TASKS (rotate by day): │
│ MON: Review weekend punches. Flag late arrivals, NCNS. │
│ TUE: Labor % check. On track for 17.5%? OT building? │
│ WED: Inventory variance scan. Anything above 5%/8%? │
│ THU: Schedule for next week. Published? Gaps covered? │
│ FRI: Accountability log review. Anything undocumented? │
│ Prepare for DM weekly meeting (Q1-Q5 answers ready). │
│ │
└──────────────────────────────────────────────────────────────┘
WEEKLY (admin blocks):
┌──────────────────────────────────────────────────────────────┐
│ │
│ SCHEDULING │
│ Build and publish by Thursday night, 2 weeks out. │
│ Built against: forecasted demand, role coverage, labor │
│ guardrails (17.5%), known availability issues. │
│ Manage time-off, swaps, call-outs. No gaps on the schedule. │
│ │
│ ORDERING │
│ Place all orders (Restaurant Depot, Webstaurant, Costco, │
│ Amazon, etc.). Par levels match demand. No waste, no 86. │
│ Investigate variance above threshold. Surface pricing or │
│ quality issues to the DM. │
│ │
│ ACCOUNTABILITY │
│ 100% of incidents documented: attendance, conduct, food │
│ safety, service, performance. If it happened and it is not │
│ logged, it did not happen. │
│ │
│ PUNCH AUDIT │
│ Review all punches within 24 hours of your next scheduled │
│ management day. Edit errors, flag discrepancies, catch │
│ early clock-ins or missed clock-outs. │
│ │
│ FOLLOW-UP │
│ Check if the one thing you coached on-site stuck. │
│ Did the SL implement? Did the metric move? │
│ If yes: reinforce at next visit ("That change is showing │
│ in the numbers"). │
│ If no: re-coach or investigate why it did not land. │
│ │
└──────────────────────────────────────────────────────────────┘
MODE 3: WEEKLY MEETING WITH DM
15-20 minutes. Same day, same time, every week.
This is your calibration point. Come prepared.
BEFORE THE MEETING (5 min prep):
┌──────────────────────────────────────────────────────────────┐
│ │
│ Have answers ready for the DM's questions: │
│ │
│ Q1: "What did you notice this week?" │
│ Review your data. What changed? What held? │
│ Scan: errors, prep time, ratings, food safety, │
│ attendance, conduct. │
│ If you have nothing for Q1, your measurement routine │
│ is broken. Fix that before the meeting. │
│ │
│ Q2: "What's the biggest problem?" │
│ The one thing that moved a metric the most. │
│ Data-driven, not feeling-driven. Not "it was a │
│ tough week." What NUMBER moved and why? │
│ │
│ Q3: "What's the constraint?" │
│ The upstream cause. Not the symptom. │
│ Test yourself: "If I fixed that tomorrow, would │
│ the metric move?" If yes, it is the constraint. │
│ If no, keep looking upstream. │
│ │
│ Q4: "What are you going to try?" │
│ Have a specific action. Not "work on it." │
│ What, when, how long. │
│ │
│ Q5: "When are you doing it?" │
│ Day, time, duration. "This week" is not an answer. │
│ │
└──────────────────────────────────────────────────────────────┘
DURING THE MEETING:
┌──────────────────────────────────────────────────────────────┐
│ │
│ The DM will run predict-then-compare situations. │
│ They present a scenario. You commit to a response before │
│ hearing what happened or what the DM would do. │
│ │
│ This is not a test. It is how your pattern library grows. │
│ The gap between your answer and the actual outcome is │
│ where the learning happens. The wider the gap, the more │
│ you learned. Do not try to guess the "right" answer. │
│ Commit to YOUR read. The gap teaches. │
│ │
│ The DM will also give identity reflections. Specific │
│ actions you took that they noticed. Accept them. They │
│ are evidence of your standard, not flattery. │
│ │
└──────────────────────────────────────────────────────────────┘
AFTER THE MEETING:
- Write down Q4 and Q5 (what you are trying, when).
- Execute during the week. This is what next week's
meeting opens with: "Did it work?"
MODE 4: SHIFT LEAD DEVELOPMENT
Your SLs are how the kitchen runs when you are not there.
If the kitchen falls apart on your off-days, your SLs are
not developed. That is your problem, not theirs.
THE PRINCIPLE:
You coach the SL the same way the DM coaches you.
- Observe before correcting.
- Ask before telling.
- One thing per interaction.
- Let them make mistakes that are not safety-critical.
- The mistake IS the rep.
ON-SITE COACHING (during Operation 3):
After every on-site visit, you coach the SL on one thing
from what you observed during peak.
"During peak I noticed [X]. What did you see?"
Listen. "What would you do differently?"
That is the interaction. Do not pile on. One observation.
One question. One answer.
WHAT GOOD SLS DO:
┌──────────────────────────────────────────────────────────────┐
│ │
│ - Walk the line during peak (not stationed, not in office) │
│ - See the bottleneck within 5 minutes and reposition │
│ - Catch errors at expo before they leave │
│ - Brief the incoming SL at shift change (handoff) │
│ - Complete the task list without being reminded │
│ - Handle team conflicts without escalating to you │
│ - Report decisions, not ask permission │
│ │
│ If your SL is doing all of these, the kitchen runs │
│ without you. That is the goal. Your job is to build │
│ SLs who make you unnecessary on the floor. │
│ │
└──────────────────────────────────────────────────────────────┘
WHAT TO DEVELOP:
┌─────────────────┬──────────────────────────────────────────┐
│ SL DOES │ YOUR RESPONSE │
├─────────────────┼──────────────────────────────────────────┤
│ Stays stationed │ "I need you walking the line during │
│ during peak │ peak. Who can cover your station?" │
│ │ Solve the coverage so they can lead. │
├─────────────────┼──────────────────────────────────────────┤
│ Misses the │ Next visit, watch. If they miss it │
│ bottleneck │ again: "Where's the backup right now?" │
│ │ Train the read, not the fix. │
├─────────────────┼──────────────────────────────────────────┤
│ Errors leaving │ "What's your expo check process?" │
│ expo │ If no process: build one together. │
│ │ If process exists but not followed: │
│ │ "What's getting in the way?" │
├─────────────────┼──────────────────────────────────────────┤
│ No handoff at │ "Walk me through what you tell the │
│ shift change │ next SL." If blank: "Here is what the │
│ │ handoff covers: [open items, staffing │
│ │ issues, prep status, any incidents]." │
│ │ One template. Practice twice. Check. │
├─────────────────┼──────────────────────────────────────────┤
│ Escalates │ "What would you do if I wasn't here?" │
│ everything │ Let them answer. If reasonable: "Do │
│ │ that. Tell me after." Build the habit │
│ │ of deciding, not asking. │
├─────────────────┼──────────────────────────────────────────┤
│ Handles a │ Note it. Tell them: "You handled [X] │
│ conflict well │ without escalating. That's shift lead │
│ │ work." Identity reflection, same as the │
│ │ DM does for you. │
└─────────────────┴──────────────────────────────────────────┘
THE TEST:
Your off-days are the test. If the kitchen runs clean when
you are not there, your SLs are developed. If it does not,
that is your next coaching target. Not "the team let me
down." Your system did not hold.
THE SINGLE PRINCIPLE
You are building a kitchen that runs without you on the floor.
Your on-site time builds the reads. Your off-site time builds the systems. Your SL development builds the layer that holds it when you are absent.
The DM measures whether you built it. The metrics prove it.
When the kitchen runs clean on your off-days and the DM visits without you present and finds nothing to flag, you have graduated from running a kitchen to running a system.