Three locations are bleeding. Same region, same brand, three different colors of red. The board wants a plan by Friday and results by the end of the quarter. Every instinct you have will tell you to run at all three at once. Do not do it.
Sequencing is the single hardest and most underrated skill in multi-unit turnaround work. The right sequence recovers dollars fast enough to buy you the runway to work the hard problems. The wrong sequence spreads your attention so thin that by week six you have three units with half-installed fixes, no visible wins, and a board that has quietly stopped believing in you.
This piece is about how to decide the order. Which unit gets your attention first, which cost line gets attacked first inside that unit, and how you stage the second and third units so the whole region ends up healthy at the same time instead of one location at a time. It is written from the middle of a $4.9M operating profit swing across three underperforming Bay Area locations, where the sequencing decisions in week one probably mattered more than any single operating move I made in the next eleven months.
The severity versus fixability matrix
The framework I keep coming back to is a simple two-by-two. On the horizontal axis, severity: how many dollars of operating profit is this unit costing you per month. On the vertical axis, fixability: how quickly can a defined, known operator move recover those dollars. Every failing unit lands in one of four quadrants.
Fig. 1 · Triage matrix. The top-right unit gets your first two weeks.
The unit that wins your first two weeks is the one in the upper-right quadrant. High severity, high fixability. It is the location where you can move the biggest dollars with the least resistance, and where you can produce a visible win fast enough to reset expectations with the board, the general managers, and yourself.
The unit in the lower-right quadrant, high severity but low fixability, is the second move, not the first. It probably has a deeper problem, a chef partnership issue, a lease question, a broken host relationship. You can and should work it, but you cannot lead with it, because if you spend your first two weeks in a low-fixability unit and produce no visible progress you have burned the political capital you needed for the harder question later.
The upper-left, low severity but high fixability, is a rhythm rollout target. You do not spend leadership time there in week one. You install the rhythm you built at the first unit and let the general manager run it.
The lower-left is the closure question. It is the unit where the numbers are small and the effort to fix is large. It might not be a save at all. But that decision cannot be made in week one, because you do not yet have the credibility to make a closure recommendation stick. Stage it for month three.
Why fixability beats severity in the first two weeks
The instinct is to attack the worst unit first. It feels right. It also almost always fails, and the reason it fails is not operational, it is political.
A turnaround operator has two currencies: dollars and belief. Dollars come from the P&L. Belief comes from visible progress. In the first two weeks you do not have much of either. Every day you spend on a unit that will not respond quickly is a day the board, the general managers, and your own team lose a little more belief. Belief runs out before dollars do. When belief runs out, general managers stop cooperating and boards start asking whether they should have hired someone else.
Momentum is a resource. Spend the first two weeks producing some. Then you have enough of it to attack the harder unit.
A visible win at the most fixable unit funds the rest of the engagement. It gives you a story to tell in the first board update. It gives the general manager at the second unit a peer to point to. It gives the general manager at the third unit a reason to trust that their turn will actually come. All of that is real, and none of it happens if you spend your first two weeks failing to move the needle at the worst unit.
Inside each unit: attack the largest gap to peer, not the largest cost
Once you pick the unit, you pick the cost line. The same trap sits inside a single unit as sits across the region: the biggest absolute number is not the biggest opportunity. Attack the line with the largest variance from a healthy peer benchmark.
An example. A unit with food cost at 32 percent and labor at 34 percent looks like a food cost problem. But if a healthy peer in the same concept runs food at 30 percent and labor at 28 percent, the labor line is six points above peer while the food line is only two above. The recoverable dollars, in that case, are twice as large in labor as they are in food. Attack labor first.
This sounds obvious. In practice, operators pattern-match on the biggest line item almost every time. Food cost is 32 percent, that is a big number, let us fix food cost. That instinct will cost you weeks. The right instinct is to look at every line, subtract the peer benchmark, and rank by dollar gap. The largest dollar gap is the first fight.
The two moves that almost always work in week one
Once you know the target line, you need a first move that produces a visible result inside two weeks. Two moves reliably do this:
- Rebuild the schedule against the last twelve weeks of hourly demand. If labor is the target, this recovers three to five points of labor variance in the first ten days without any headcount change and without any wage rate conversation. It is the highest-leverage single move in restaurant operations.
- Install a closing-shift comps and waste log. If food cost is the target, a five-minute end-of-shift ritual where the closing manager writes the day's waste and comps into the system before locking up recovers one to two points of food cost inside three weeks. It is not glamorous. It works.
Neither move requires anyone to be fired. Neither move requires new technology. Neither move requires the general manager's permission, only their participation. That combination is what makes them the right first moves.
Staging the second and third units
By the end of week two the first unit has a visible win and the general manager is running a new rhythm. You leave. You do not stop caring, you shift your attention. Weekly check-in with the general manager, one-page dashboard, standing Monday P&L review. That is enough. Your leadership time now moves to the second unit.
The transition off the first unit is a decision, not a fade. Book it. The last onsite day at unit one becomes the first onsite day at unit two, on the same calendar page. If you let the transition drift, the first unit will pull you back into every equipment ping and staffing shuffle for the next month, and the second unit will silently rank you as somebody else's problem. General managers read your calendar as accurately as they read a P&L.
The second unit is the hardest, most severe unit, the one you deliberately did not start with. Now you have two things you did not have in week one: a proof point and a working playbook. Both compress the timeline. What took two weeks at unit one usually takes ten days at unit two, because the general manager at unit two has already heard from the general manager at unit one that this is not a fake exercise.
Weeks five and six are the third unit. By this point you are largely running a template. The third unit gets the rhythm, the schedule rebuild, the closing ritual, and the weekly review, mostly transplanted from the first two. This is where operators over-invest, because it feels like the whole region is fixed. It is not. The rhythm is fragile at all three units through month three. Cadence is the only thing that keeps it in place.
The mistakes I have made in this sequence
I have gotten this wrong enough times to be specific about how it goes wrong. Three patterns.
Working all three at once
In my first multi-unit engagement I ran at all three units in the same week. Two team meetings, three schedule rebuilds, three closing rituals installed on the same Thursday. By day ten none of it had held, because none of it had my attention. I lost three weeks. I would rather lose two on one unit and get the win.
Starting with the worst unit because it felt honest
The pull to start with the worst unit is moral, not operational. It feels wrong to skip the location where the most damage is being done. It is worse to attack it and fail. In one engagement I spent my first month on a location that turned out to need a lease decision, not an operating decision. That month was mostly wasted and I never fully recovered the trust of the general manager, who could tell that I was reaching for tools that did not fit the problem.
Not telling the general managers the sequence
Early on I kept the sequence in my head. The general manager of the third unit found out from a peer that she was going to be worked in week five, not week one, and read it as being deprioritized. She spent two weeks defending her unit from me instead of preparing to work with me. Now I put the sequence on a whiteboard in the kickoff meeting. Third in line still stays engaged if they know their turn is coming and they know why.
Confusing a small fix at unit one with a proof point
The first unit does not need a small fix. It needs a fix that is large enough for every general manager in the region to hear about it inside a week. In one engagement I rebuilt a bar cost sheet at unit one and moved beverage cost from 24 to 21 percent. Real work, real dollars, and completely invisible to the rest of the region because nobody who did not run bar operations understood what changed. A schedule rebuild that shows up as calmer Friday nights and a shorter labor line is legible to everyone. Pick the fix that reads clearly to a line cook, not just to a controller.
The point
When three locations are bleeding at once, the temptation is to run in every direction. The discipline is to pick one, produce a visible result in two weeks, and use that momentum to earn the runway to work the harder units. Severity tells you which units matter. Fixability tells you which one to start with. Variance to peer tells you which cost line to attack inside each one. Sequence tells you when to move.
The whole region gets healthy in the same nine months either way. The question is whether it gets healthy through a sequence you controlled or through three parallel half-fixes that never quite installed. Pick the sequence. Say it out loud. Then work it.