General Manager burnout is the workforce topic that every restaurant leadership conference has a session on. The framing is always the same: the role is demanding, the hours are long, the emotional load is heavy, the retention numbers are bad. All of that is true. None of it is the cause.
The cause is architectural. The GM is carrying every decision in the operation because the operation has no layer that can make decisions without her. She is not burning out because the work is hard. She is burning out because she is the single point of decision for an operation that generates decisions faster than any one human can process them.
What the GM Actually Does
Sit a GM through a Friday dinner service. Count the decisions. The line is behind on the eight-tops. The bartender is slammed. Table 14 flagged the server twice. The back of house is running low on the halibut. The dishwasher just called out. The hostess seated a four-top in the wrong section. The GM decides on each of these. Every shift. Every service.
Now ask which of those decisions required her specific judgment. Maybe three. The rest were policy-level decisions that a competent system could have made: allocate the halibut, resequence the line, reroute the four-top, backfill the dishwasher from the pre-committed standby pool. She made them because there was no system to make them. So she did.
The Workforce Framing Misses the Cause
The sessions on GM burnout will recommend better scheduling, better management pipelines, better retention bonuses, better training. These will help. They will not fix it. Every GM, trained better or paid more, will still be the single point of decision in an operation with no decision layer beneath her.
The retention problem is the decision-load problem. GMs leave because the role is structurally impossible at scale. It is not impossible because the work is hard. It is impossible because the role requires one human to be in every decision, continuously, across a 12-hour service where each decision is consequential and none waits.
What an Execution Layer Changes
An execution layer handles the decisions that do not require judgment. The halibut gets reallocated because the allocation logic ran, not because the GM decided. The four-top gets re-seated because the hostess received a corrected assignment before she walked the guests. The line resequences because the kitchen display reprioritized based on fire times. The GM is not notified about any of this. It happened. She was not the decision.
The decisions she is notified about — a specific guest issue that requires her personally, a judgment call on a staff situation, a VIP arrival that needs her attention — are the ones she was hired to make. Those decisions do not burn her out. The accumulated weight of 147 routine decisions per service burns her out. Those are the ones the execution layer absorbs.
Retention as a Downstream Metric
In operations that run an execution layer, the GM role changes. Her cognitive load drops. Her decision queue shortens. Her shift ends with her still having judgment to spend. She does not leave at the 18-month mark because the work is no longer grinding her down in the way it was.
This is what the workforce framing cannot see. Retention is a consequence. It is not the lever. The lever is the decision architecture. Fix the architecture, and retention moves. Leave the architecture in place, and every retention program is a patch on a structural leak.
Your GM is not burning out because she is weak. She is not burning out because the work is inherently brutal. She is burning out because the operation is built to terminate at her. That is fixable. It is not fixed by hiring a more resilient GM. It is fixed by building the layer beneath her.