I have run multi-unit turnarounds for about ten years now. Three of them at scale. Others as advisory or interim engagements. The playbook I use today looks nothing like the one I used on the first one, and the reason is that I have made every mistake in this category at least once. Some of them twice.
What follows is the honest inventory. Five specific decisions I would not make again. Each one comes with a real cost, in dollars where I can put a number on it, in time and trust where I cannot. This is not the theory version. This is the version I would tell a peer over coffee if they were about to walk into a broken $30M restaurant group and ask me what to watch out for.
The reason I am writing this now, rather than five years from now with a rounder narrative, is that the mistakes are still specific in my head. Once they get smoothed out into wisdom they stop being useful. The cost is what makes them stick.
Mistake 1: Moving on the schedule in week one
The first serious turnaround I ran was three underperforming locations inside a Bay Area group. I walked in on a Tuesday, saw the labor line running eight points over benchmark, and by that Friday I had rebuilt the schedule for two of the units. It felt like the right thing to do. Labor was the biggest single leak on the P&L. Fixing it was what they had brought me in for.
What I did not understand at the time is that a schedule change in week one destroys the diagnostic signal on labor for at least three weeks after. You cannot tell what was the underlying variance and what was your own new schedule shaking out. I lost two full weeks of clean data on the two units I touched, and I ended up rebuilding the schedules a second time in month two once I actually understood the demand curve.
What it cost
Roughly $80K across the two units in overcorrection. I cut hours too aggressively out of the wrong shifts. Two sections went short on a Friday dinner and one catering delivery went out an hour late. The catering client did not renew that quarter. I could have caught all of it if I had held my hands off the schedule for three more weeks.
What I do now
Nothing on the schedule until day 22. Nothing. I pull the historical schedule and the historical demand data, I sit with the general manager and walk through it, but I make no live changes. The first schedule change goes in during week four, after the demand curve is clear and the general manager has signed off on the shape.
Mistake 2: Firing the general manager on day 40
The second turnaround included a general manager who, by every visible measure, looked like the problem. Numbers were bad. Team morale was bad. He was defensive in the diagnostic meetings. I gave him a 30-day window, saw no movement, and moved him out on day 40. It felt decisive at the time.
Here is what I did not do. I did not fix the system around him first. I never installed a real weekly P&L review at his unit. I never gave him a live dashboard. I never made his decision rights explicit. I gave him 30 days to fix an operation with tools he did not have and information he could not see, and when he failed I concluded he was the problem.
The general manager who followed him was better on paper. She spent six weeks onboarding, six weeks stabilizing, and the unit went sideways for a full quarter. In hindsight the first general manager, given a real operating rhythm, would have delivered inside 90 days.
What it cost
About $220K across the quarter. Recruiting fees. Onboarding time from the area director. Two guest complaints that turned into refunds. And the second-order cost that never shows up on the P&L: the rest of the region watched me fire someone on day 40, and they spent the next two months hiding problems from me. Every general manager became careful. That carefulness is the opposite of what a turnaround needs.
What I do now
No general manager decisions before day 90. If someone is actively harming the operation, they get moved into a different seat. But the fire-versus-keep decision waits until the fixed system has been in place for six weeks and the person has been given a real shot at running it.
Mistake 3: Cutting a menu item without the chef
Third turnaround. There was a menu item selling steadily but running a food cost above 40 percent. On paper it was a candidate to cut. I cut it. What I did not know is that the chef had built the item as a signature dish for a specific catering client, and half the item's cost was in a plating garnish that we could have swapped for a third of the price. The chef would have told me that if I had asked.
Instead I sent an update to the menu system on a Thursday. The chef found out on Friday when he saw the printed menus. He was, correctly, furious. Not because I cut the item. Because I did it without him.
What it cost
The chef gave notice inside three weeks. He had been at that group for six years. His replacement took four months to hire and another two months to onboard properly. The menu drifted during that gap. Guest scores fell about eleven points on the post-visit survey. Fully loaded I put the cost around $140K, and the trust cost with the rest of the kitchen was worse than the dollar cost.
What I do now
Any decision that touches the menu, the plating, or the recipe gets walked through with the chef first, in person, with the P&L open on the table. Not a memo. Not a slack message. Not a data point in a deck. A conversation, with the numbers visible, before the change ships. Every time.
Fig. 1 · The five mistakes, in dollars of avoidable cost.
Mistake 4: Not communicating with line staff early enough
In the third turnaround I waited until week five to hold an all-hands with the line staff at each of the three locations. My thinking was that I did not want to promise things I could not deliver. If I told the dishwashers and the line cooks that a turnaround was happening and then things did not visibly change in a month, I would look worse than if I had said nothing.
That reasoning was wrong on every dimension. Line staff already knew something was happening. The regional director was on site every day. New spreadsheets were floating around. The general managers were tense. What silence created was not calm. It was rumor. And the rumor was always worse than the truth.
What it cost
Line-staff turnover ran 14 percent across the three units in the first eight weeks. Two of the resignations were people I would have kept for years. I had to backfill 11 hourly positions in a tight labor market, and roughly half of those hires washed out inside 60 days and had to be replaced again. Fully loaded I put the cost around $185K in recruiting, training, and productivity loss.
What I do now
Every turnaround gets an all-hands per location in the first 10 days. I do not use the word turnaround because it scares people. I say some version of "the group is changing how it runs, here is what will look different this month, here is what will not change, and here is when you will hear from me again." That last part is the important one. The cadence of communication matters more than the content.
Mistake 5: Skipping the exit ritual
This is the biggest one. Two of my first three engagements ended without a real handoff. I finished the stabilization work, delivered the numbers, walked the board through the results, and left. Both engagements decayed within a year. One of them lost about 60 percent of the gains inside eight months.
What decayed first was the operating rhythm. The weekly P&L review stopped happening within three weeks of my exit. The monthly regional operating review became a status meeting with slide decks instead of a working session. The dashboard I had built stopped being updated because nobody owned the data pipeline. Six months in, the standing habits that had held the turnaround in place were gone, and the P&L started to slide back to where it started.
The last month of a turnaround is the most fragile month. It is the month everyone treats as a formality. It is also the month that decides whether the gains hold or unwind.
What it cost
Across two engagements, roughly $440K in decayed contribution over the following year. Plus the reputational cost, which is real. When gains unwind after an outside operator leaves, the story that gets told is not "the handoff was weak." The story that gets told is "the turnaround was fake." That is not fair, but that is how it lands.
What I do now
The last 30 days of every engagement is treated as the most important 30 days, not the least. Written handoff documents for every operating cadence. A trained successor who shadows me for four weeks and then runs the cadence for two weeks while I sit in the back. Dashboards migrated to the internal team's ownership. A 60-day and 90-day check-in built into the exit agreement. I will not run a turnaround now without this ritual in the contract.
What ties them together
Look at the five mistakes and there is a pattern. In every one, I moved without first building the thing that would let the operation absorb the change. I moved on the schedule before I understood the demand. I moved on the general manager before I had built the system that would let them succeed. I moved on the menu before I had built the trust with the chef. I moved on operations before I had communicated to the line. I exited before I had installed the successor.
The pattern is not that I moved too fast. The pattern is that I moved before the operation was ready to receive the move. Speed is not the enemy of a turnaround. Impatience is. And what looks like patience from the outside is usually just sequencing done properly.
Every operator I know who runs turnarounds well has their own version of this list. Different mistakes, same shape. The specific dollar figures matter less than the discipline of writing them down and looking at them before every new engagement. I keep my list printed and tacked to the wall of whatever temporary office I set up on day one. It is the first thing I read every morning of a rescue.
What I wish someone had told me at the start
Two things. First: a turnaround is not five parallel workstreams. It is one sequenced program with a start, a middle, and an end. The urge to run everything at once is what breaks most first-time turnaround operators. Pick one thing. Do it right. Then pick the next thing.
Second: your job is not to be the smartest person in the room. Your job is to install the rhythm that makes the room smarter every week without you. If you leave and the team gets slower, you did the work wrong. If you leave and the team keeps compounding, you did it right.
These mistakes cost about $1.07M in the aggregate across three engagements. They also produced the playbook I use now, which is the reason the last turnaround delivered $4.9M in operating profit recovery across three units in 11 months. The mistakes were the tuition. The playbook is what got built with the diploma.
Every mistake in this article is one I would trade for a shorter learning curve. Every one is also one I do not need to repeat. That is what a real playbook is: the list of things you already paid for.