The first day you walk in on a group of underperforming restaurants, the temptation is to fix something visible. Change the schedule. Cut a menu item. Move a general manager. The instinct is right that something is wrong. The instinct is wrong about what to do about it.
I have run this play a few times now. The last time was a $30M Michelin-recognized Bay Area group where three of the five locations were leaking money. The board wanted a plan in a week and results in a quarter. Eleven months later the three locations were back to healthy contribution and the group had recovered roughly $4.9M in operating profit. This is the sequence that got us there, and the parts I would do differently.
What "underperforming" actually looks like on the P&L
Before you can sequence a turnaround, you have to be honest about what you are looking at. A badly broken restaurant P&L does not fail in one place. It fails in a cluster. The pattern is almost always the same across cuisines, price points, and formats:
- Labor as a percent of sales runs 4 to 8 points above the healthy peer benchmark. Not because the wage rate is wrong, but because the schedule does not follow demand. Someone is being paid to stand still.
- Food cost runs 2 to 5 points high with waste and comps under-recorded. The cost of goods sold number on the P&L is understating the real leak because the count discipline in the walk-in has drifted.
- Repair, maintenance, and supplies line trends up quarter over quarter, usually a sign that small equipment failures are being solved by throwing petty cash at them rather than being logged and fixed root cause.
- The general manager is running the restaurant from behind the line, not from the P&L. They know the food. They do not know their cost per cover.
None of those symptoms are the problem. They are what the problem looks like from the outside. The actual problem, almost every time, is that the operating system has decayed. The unit is running on the memory of the last person who was good at it, not on a living operating rhythm.
Fig. 1 · Diagnose, stabilize, rebuild. In that order.
Days 1 to 30: Diagnose without disrupting
The first month is diagnostic and only diagnostic. You are gathering signal. Anything you change in the first 30 days corrupts the picture, and the picture is the most valuable thing you will get in this whole process.
Here is what actually goes on the calendar:
Walk every unit on a working shift
Not a quiet Tuesday afternoon. A Friday dinner. A Sunday brunch. The moments where the operating system is under load and you can see where it flexes and where it snaps. Bring a notebook, not a laptop. You are watching, not measuring yet. The measuring comes from the numbers.
Read 90 days of daily P&L, line by line
Every single day. Every single line. Not the monthly rollup. The monthly rollup will lie to you because it hides the volatility, and the volatility is the whole story. When labor cost swings from 28 percent to 41 percent across two days in the same week at the same location, that is your operating rhythm telling you it does not exist.
Map the actual decision rights each general manager holds
Ask this directly, in a one-on-one: "If a truck delivery arrives short, who decides whether to accept it?" "If a line cook does not show up, who decides whether to close the section or pull someone from prep?" "If a guest asks for a comp, up to what dollar amount can you say yes without calling anyone?"
You will find, almost every time, that the general manager has been given responsibility for the P&L but has been quietly stripped of the decision rights that would let them protect it. That gap is not their fault. It is the shape of the problem.
The general manager is not underperforming. The system around them is. Fix the system first. The person can prove themselves inside a fixed system in about six weeks.
Do not fire anyone in the first 30 days
The urge will be strong. Do not do it. Firing in month one teaches the rest of the region to hide problems, and the hidden problems are the ones you cannot fix.
Days 31 to 60: Stabilize the top three leaks
By day 30 you know where the money is going. Now you attack the three biggest leaks, in that order, with the general managers, not to them.
Fig. 2 · Typical gap versus healthy peer benchmark, percent of sales.
1. Labor variance: get the schedule to follow demand
Pull the last 12 weeks of hourly sales by day part and overlay the schedule. You will see the gap immediately. In an underperforming unit the schedule is usually built off the last schedule, not off the demand curve. The fix is not a labor cut. The fix is putting the same labor hours in the right hours. Done well this recovers 3 to 5 points of labor as a percent of sales in the first 60 days without any headcount change.
2. Food waste and comps: get the counts real
Do a full physical inventory in every walk-in and every dry storage the same week, across all locations. Compare to what the system says you have. The gap is your unrecorded waste and your unrecorded comps. Then build a five-minute end-of-shift closing ritual where the closing manager writes the day's waste and comps into the system before they lock up. Simple. It usually recovers another 2 to 3 points of food cost inside 60 days.
3. Repair and maintenance: log everything for 30 days
Every equipment issue gets logged the day it happens, with a photo, a description, and the temporary fix. At the end of 30 days you will see three or four equipment failures that keep repeating. Fix those root cause. This is small money in isolation and large money over a year, and it also fixes a real morale problem: line cooks stop wasting an hour of every shift on broken equipment.
Days 61 to 90: Rebuild the operating rhythm
Stabilization gets you the numbers. Rhythm makes them stick. The last 30 days of the turnaround window is where most operators lose the gains they just made, because they never install the thing that will hold them.
The operating rhythm has three parts, and all three have to be running by day 90.
The weekly P&L review, with the general manager
Every Monday, 45 minutes, general manager and area director, same time, same agenda. Last week's labor as a percent of sales, food cost, top three variances, one thing to fix this week. Not a status meeting. A working meeting. The general manager runs it. You listen.
The monthly regional operating review
Every general manager in one room once a month. Each one presents the last month's P&L and the next month's plan, in five minutes, with numbers. Peer accountability does more work here than manager accountability. General managers hate looking bad in front of other general managers more than they hate looking bad in front of you.
The dashboard the general manager actually owns
Not a report they receive. A live view they open every morning. Labor variance yesterday, food cost yesterday, top three items sold, comps and voids yesterday. If they cannot get to it in one click from their phone, they will not use it, and if they do not use it, none of this holds.
What holds a turnaround in place after the operator leaves is not the operator. It is the standing meeting, the shared dashboard, and the number the general manager can name from memory.
Beyond 90 days: what actually held
In the case I keep coming back to, the three underperforming locations went from a combined roughly $1.8M in annual operating losses to roughly $3.1M in annual contribution over 11 months. That swing is the $4.9M number. The stabilization phase did about 55 percent of it. The rhythm phase did the other 45 percent, and would have decayed within a quarter without it.
Fig. 3 · Combined swing from operating loss to contribution.
What I got wrong the first time
I want to be honest about the mistakes, because the playbook above is what I would do now. It is not what I did the first time. Three things I would do differently:
- I moved on the schedule too early. In one location I rebuilt the schedule in week two, before I understood the demand curve properly. The new schedule was better than the old one, but I lost two weeks of clean diagnostic signal because I could not tell what was labor variance and what was my own change.
- I underestimated how much the general managers wanted the rhythm. I framed the weekly P&L review as accountability. The general managers who stayed later told me they experienced it as relief. For the first time, someone was going to look at their numbers with them every week, and they were not going to have to defend themselves alone at quarter end.
- I did not build the dashboard until month four. The single highest impact tool in the whole turnaround, and I treated it as a phase two thing. If I do this again, the dashboard is a day 40 deliverable, not a day 120 one.
The point
A multi-unit turnaround is not a series of heroic decisions. It is a sequence, and the sequence is the whole thing. Diagnose first, without changing anything. Stabilize the three biggest leaks with the general managers holding the pen. Install the operating rhythm that will hold the gains after you are done. Do the rebuild in the open. Do not fire early.
Most turnaround stories get told as personality stories. The operator walked in, made a call, saved the day. That is almost never what actually happened. What actually happened is that someone patient built a picture, then attacked the three things the picture pointed at, in the order the picture said to, and then installed the standing habit that made those fixes permanent.
Cadence beats charisma. Every time.