← Back to blog

Sequencing Fixes When Everything Is Broken

A turnaround is not a to-do list. It is a dependency graph. The operator who wins is the one who fixes in the order that lets the next fix become fixable, not the one who moves fastest.

Kiyong (Eric) Lee·Operations Executive·August 2026·9 min read

The short version

Sequence beats speed. Moving fast in the wrong order costs 30 to 60 days of rework you cannot afford.
Cash first if the doors are at risk. Morale second, through visible small wins. Labor third. Food cost fourth. Operating rhythm last.
The biggest cost line is almost never the right first fix. The most upstream fix is.
Some fixes can run in parallel. Labor and food cost cannot. Both require the kitchen to focus and asking for two at once gets you zero.
The dependency map is more portable than the playbook. Every turnaround is different. The graph is the same.

Every broken restaurant looks the same from a distance and different up close. Labor is high. Food cost is high. Morale is bad. Cash is tight. Guest complaints are up. Turnover is up. Comps are up. The general manager is exhausted. The area director is exhausted. The board is out of patience. Every line on the P&L is a fire and every fire is real.

The instinct is to attack the biggest fire first. That instinct is wrong. In a turnaround, the biggest fire is almost never the most upstream fire, and if you attack the wrong one, you will burn 60 days moving nothing and lose the room in the process. The single most important skill in the turnaround job is not speed. It is sequencing. Knowing which fix has to come first so that everything else becomes fixable.

I have run this play at scale. The most recent was a $30M Michelin-recognized Bay Area group where three underperforming locations came back to roughly $3.1M in combined annual contribution over eleven months, a swing of roughly $4.9M. The sequencing decisions we made in weeks two through eight determined 70 percent of that outcome. Everything downstream of the sequence was execution.

The dependency map

Fixes have dependencies. That sentence is obvious in software and heresy in operations, and it costs a lot of operators a lot of time. In a broken restaurant, the fixes are not independent line items. They are nodes in a graph, and the graph tells you the order.

Fig. 1 · Fixes as a graph, not a list.

Node 1: Cash, if the doors are at risk

Cash is only the first fix if it is a first-order problem. If the operation has 90 days of runway or better, cash is not on the critical path in month one and you skip to morale. If payroll is at risk inside 60 days, cash jumps to the front of everything else and stays there until the near-term cliff is cleared.

Cash stabilization in a restaurant is unglamorous and it works. Renegotiate the two largest vendor payment terms. Pause every discretionary spend line on the P&L. Delay any capex that is not a safety or health issue. Confirm the line of credit is available and not being called. This is 5 to 10 days of focused work and it buys the runway that lets every other fix have time to work.

The mistake here is treating cash as a permanent workstream. It is not. Once the doors are 90 days safe, cash goes back to being a monitoring line item and you move to the next node.

Node 2: Morale, through visible small wins

This is the fix that operators skip and the one that decides whether anything downstream sticks. Morale in a broken operation is usually not what people think it is. It is not a pizza party problem. It is a "nothing works and nobody cares" problem, and it compounds into every operational fix you will try to make over the next 90 days.

The way you fix morale in week two is by fixing three visible small things that have been broken for months. These are not the big financial moves. They are the ones that buy operating credibility, and every downstream fix will spend that credibility.

What actually counts as a visible small win

The prep cooler that has been running warm for six weeks. Fix it in 72 hours. Everyone in the kitchen will notice.
The shift meal program that quietly stopped. Restart it. Line staff will tell every friend at every other restaurant in the city.
The schedule that gets posted 48 hours before the week starts. Move it to 10 days out. Every server will notice inside two weeks.
The bathroom faucet that has been broken since spring. Get a plumber onsite. It is not about the faucet.

Morale is not a soft metric. It is the multiplier on every operational fix you will try to make. A demoralized team executes at 40 percent of a stable team. Fix morale first and everything downstream doubles in speed.

Morale can run in parallel with cash stabilization because they touch different systems and different people. Do them together. Give yourself two weeks to complete this phase before you touch labor.

Node 3: Labor, because it moves fastest

Labor is the third node and the first big financial move. It comes third for two reasons. First, it is the fastest cost line to move, so the win compounds quickly and buys more credibility. Second, it unlocks the food cost fix that comes after it, because a properly staffed line has time to execute the count discipline that food cost requires.

The labor fix is almost never a headcount cut. In every underperforming restaurant I have walked, the wage rate is roughly right and the headcount is roughly right. What is wrong is that the schedule does not follow the demand curve. The same total hours are in the wrong hours.

How to move labor without cutting anyone

Pull 12 weeks of hourly sales by day part. Build the actual demand curve for each unit.
Overlay the current schedule. The gap will be obvious. Overstaffed weekday afternoons, understaffed Friday and Saturday evenings.
Rebuild the schedule against the demand curve with the general manager doing the pen work. You are teaching, not doing.
Set the labor target as a rolling four-week average, not a weekly target. This lets the general manager smooth through week-to-week noise.

Done well, this recovers 3 to 5 points of labor as a percent of sales inside 60 days with no headcount change. That is the win that funds the next 90 days of work.

Node 4: Food cost, once the team is ready to hold it

Food cost is the fourth node, and this is where operator instinct fights the framework hardest. Food cost is usually the biggest gap versus benchmark. Operator instinct says attack the biggest gap first. The dependency map says wait.

Food cost is a discipline fix, not a policy fix. It requires the closing manager to actually enter comps and voids into the system before locking up. It requires prep cooks to portion accurately when the line is slammed. It requires line cooks to log waste rather than throw product in the bin quietly. Every one of these behaviors requires a stable team with time to execute them, and that is only possible after labor is aligned and morale is intact.

The food cost sequence, once you are ready

Full physical inventory in every walk-in the same week across all locations. Compare to system count. The gap is your baseline hidden leak.
Install a five-minute end-of-shift closing ritual where the closing manager logs waste and comps before locking up.
Rebuild the prep sheets against actual sales velocity, not against last week's prep sheet.
Weigh three high-cost proteins portioned by line cooks for two weeks. Standardize the target weight and coach the deviation, not the person.

This is 4 to 6 weeks of consistent work and it recovers 2 to 3 points of food cost. Do it before morale is stable and it will not hold. Do it after and it becomes durable.

Node 5: Operating rhythm, once there are real numbers to run on

The operating rhythm is the fifth and final node, and this is the one that separates a turnaround that holds from a turnaround that decays inside a quarter after you leave. The rhythm needs real numbers to run on, and the first four nodes are what produce those numbers.

Install the rhythm before the numbers stabilize and you get a weekly meeting reviewing garbage data. General managers stop taking it seriously inside three weeks. Install it after the first four fixes and the same meeting becomes the tool that carries the operation for the next two years.

The three pieces that have to be running by month three

The weekly P&L review, general manager plus area director, 45 minutes, same day and time, standing agenda. Last week's labor, food cost, top three variances, one fix for this week.
The monthly regional operating review with every general manager in one room. Each one presents their last month and their next month in five minutes. Peer accountability does more work than manager accountability.
The dashboard the general manager opens every morning. Not a report they receive. A live view. Labor variance yesterday, food cost yesterday, top three items, comps and voids. One click from a phone.

What can run in parallel and what cannot

Not all nodes are strictly sequential. Some fixes can run in parallel because they touch different systems. Others cannot, because they compete for the same operational attention. This is where the map gets more useful than the linear list.

Cash and morale can run in parallel. Vendor renegotiation is a leadership workstream. Fixing the broken faucet is an operations workstream. Different people, different meetings.
Labor and morale can run in parallel once cash is stable. The schedule rebuild is a general manager conversation. The small-wins fixes are a facilities and culture conversation.
Labor and food cost cannot run in parallel in the same 30-day window. Both require the kitchen and the line to focus. Asking a team to change two operational habits simultaneously produces zero changes and a lot of frustration.
The dashboard build can run in parallel with any operational fix. It is a data and tools workstream, not an operational habit change. Start it in week three and it will be ready when the numbers stabilize.

Where I got the sequence wrong

The framework above is what I would do now. It is not what I did the first time.

Early in my career I attacked food cost in month one because it was the biggest gap versus benchmark. Sixty days later food cost had barely moved, morale was worse than when I started because the kitchen felt targeted, and labor was still wrong. I had spent two months on the wrong node. The lesson was expensive and I have not repeated it.

I also under-invested in morale on my second turnaround. I told myself the visible small wins were a distraction from the real financial work. They were not. When we finally fixed the walk-in cooler that had been broken for four months, the head line cook cornered me and said, "That is the first thing that has changed in this building in a year." Everything downstream of that conversation moved faster.

The point

A turnaround is not a to-do list. It is a dependency graph. The operator who wins is the one who fixes in the order that lets the next fix become fixable, not the one who moves fastest on the biggest visible fire.

Cash first if the doors are at risk. Morale second through visible small wins. Labor third because it moves fastest and unlocks the rest. Food cost fourth because it requires a stable team to hold. Operating rhythm last because it needs real numbers to run on. Some fixes can share a lane. Labor and food cost cannot.

The map is more portable than the playbook. Every turnaround is different. The graph is the same. Learn the graph and the sequence becomes obvious even when everything is on fire at once.

Everything on fire and unsure what to fix first?

I take on a small number of multi-unit turnaround engagements each year, mostly restaurant and hospitality groups in the $10M to $60M revenue range. If you are staring at a broken P&L and trying to figure out where to start, I am happy to trade notes.

Get in touch

Questions people ask

How do you decide what to fix first in a restaurant turnaround?

Fix in the order the dependency map demands, not in the order of biggest cost line. Cash comes first if the doors are at risk. Morale comes second because every downstream fix requires the team to execute it. Labor comes third because it moves fastest and unlocks the others. Food cost comes fourth. Operating rhythm comes last, once there are real numbers to run on.

Why does sequence matter more than speed in a turnaround?

Because fixes have dependencies. A food cost fix ordered before a morale fix will not stick because the kitchen has no reason to execute it. A rhythm fix installed before the numbers stabilize will run on garbage data and get abandoned inside a month. Moving fast in the wrong order costs you 30 to 60 days of rework. Moving in the right order, even at moderate speed, holds.

Should you fix labor or food cost first?

Labor first, in almost every case. Labor is the fastest line to move because the schedule can be rebuilt in a week. Food cost requires count discipline, prep discipline, and portioning discipline, all of which are people problems that require a stable, non-panicking team to execute. Fix labor first, use the labor win to build credibility, then attack food cost.

What if cash is a problem?

Then cash jumps to the front of every other fix. If payroll is at risk in the next 60 days, you renegotiate the two largest vendor terms, pause discretionary spend, and clear the near-term cliff before you touch anything else. Nothing else matters if the doors close. Once cash is 90 days out, the standard sequence resumes.

How do you handle morale during a turnaround?

Fix three visible small things in the first two weeks. Broken equipment that has been broken for months. A shift meal program that stopped. A schedule that gets posted 48 hours late every week. These are not the big financial moves. They are the ones that buy you operating credibility, and every downstream fix will spend that credibility. No credibility, no execution.

Can you run multiple fixes in parallel?

Some, not all. Cash and morale can run in parallel because they touch different systems. Labor and food cost cannot run in parallel in the same 30 day window because both require the kitchen and the line to focus, and asking for two operational changes at once produces zero. The dashboard build can run in parallel with any operational fix because it is a separate workstream.

What is the biggest sequencing mistake operators make in turnarounds?

Attacking the biggest number first instead of the most upstream one. Food cost is often the biggest gap versus benchmark, so operators go straight at it. It does not move because the underlying morale and labor conditions do not support it. Two months later they are frustrated and behind schedule, when the answer was to fix labor first and let it pull food cost along behind it.