On a multi-party large loss, a physical war room with one accountable keeper fixes what calls, dashboards, and email threads cannot: it gives everybody the same picture at the same time, with the reasoning still attached, and a named person who owns the whole job when a decision has to be made between meetings. The expertise is usually already on the job. What is missing is orientation — one shared place where the job can be seen whole, updated by hand, in person, often enough that no shift has to guess what the last shift meant.
Take a typical shape. A 60-person commercial and technical shop gets the call after a pipe failure floods two floors of a community hospital — an imaging suite, a pharmacy, and 22 patient rooms. The work runs around the clock under infection-control protocols. The facilities director controls access and shutdowns. An industrial hygienist controls clearance. A carrier large-loss unit controls scope and spend. Inside the shop, mitigation, contents, and rebuild are three divisions with three leads. Every party holds a true piece of the job. Nobody holds the job.
For the first days, coordination is the usual system: endless email threads and a call twice a day. It produces the usual failure, just faster and with higher stakes.
Decisions wait for people who are not on the call. A question about opening a corridor for equipment is parked until the facilities director returns. Drying equipment is staged for rooms facilities has not cleared to enter. The hygienist’s clearance criteria live in a careful PDF nobody on nights has opened — at 2 a.m. a PDF in an inbox is not information in the room where the work happens. And day shift re-litigates night shift’s containment decisions, not because they were wrong, but because the why never transferred. The decision traveled. The reasoning stayed in somebody’s memory and went home to sleep.
More communication does not fix this. The calls and threads get longer; the picture gets worse, because each party updates a private version of the job from fragments. In John Boyd’s terms, orientation is the Schwerpunkt — the main effort that shapes what you observe, decide, and do. If five parties have five orientations, you do not have one slow job. You have five jobs colliding in the same wet building.
A different phase needs a different machine
Toyota built a version of this for that phase.
The obeya — literally, the big room — came out of the Prius program and the G21 effort behind it. In the exploration phase, when no proven design and no standard work exist, Toyota put all the project information in one room, posted in one place where everyone could see the whole at once. One shared truth, on the walls, not distributed across departments.
Paired with the room was a person: the shusa — one keeper with total accountability for the whole program, not a department’s slice of it, carrying signature and approval power over the product as a whole. The functional specialists do not report to the shusa; their own lines stay intact. What the shusa owns is the whole — the trade-offs between parts, the sequence, the point where a local optimum stops serving the program.
The phase distinction matters. Toyota does not use shusa and obeya to run a mature line. Once work is repeatable, different machinery takes over: standard work stating the current best way, owned and improved by the people doing the work; takt time, available time divided by demand; the andon signal that lets anyone surface a problem without blame in exchange for a real response. That machinery is superb — for work that has settled.
A hospital flood in its first week has not settled. Every room is a slightly different problem, clearance depends on readings that change, and the hospital’s needs change by the hour. This is exploration, even if the individual tasks are familiar. Running it with repeatable-phase tools is how you get a beautiful dashboard describing a job nobody recognizes. The mismatch, in Boyd’s word, is between the system you are using to see and the thing you are actually looking at — the signal that your orientation no longer fits reality.
The room in the hospital
The fix started with asking the hospital for a room nobody wanted: an empty training room near the affected floors. Not an office for the project lead. The job’s room.
Wall one became the floor plan — both floors, large, with drying status shown room by room, updated in marker, by hand, at each shift change, in person. The person coming off shift stood at the wall with the person coming on and changed what had changed. That walk is genchi genbutsu in ordinary clothes: go and see the actual place and the actual thing, rather than managing from a summary.
Wall two became decisions pending. Each item carried two things email threads rarely carry clearly: the name of the one person who could decide it, and the hour that person would next be in the room. Not “waiting on facilities.” A name and an hour.
The shop’s project lead took the shusa role — and the limits matter as much as the authority. He did not out-rank the hygienist on clearance or the facilities director on his own building. Those calls stayed theirs. What he owned was the whole picture: no decision sat unnamed, no division optimized its own work by breaking another’s sequence, nobody left a shift without walking the walls. Specialists cycled through. The keeper stayed with the room.
One change did more than any software could have. The hygienist rewrote her clearance criteria onto wall one herself, in her own hand, in plain sequence, where the night shift worked. Same criteria as the PDF. Different reality. At 2 a.m. nobody had to open or interpret a document. They looked up.
What actually changed
Decision latency stopped being measured in calls and started being measured in wall walks. Tempo here is relative, not absolute — Boyd’s point was never just to go fast, it was to decide inside the other side’s ability to keep up. On this job the other side was not an adversary; it was confusion itself. When the decider’s hour is written on wall two, a question that used to wait half a day gets answered while the room it affects is still in the state the question was about.
Night-shift re-litigation ended for a plain reason: the reasoning was on the wall, not in memory. Day shift could see why a containment choice was made — what reading, what constraint, what hospital need forced it. They could disagree in the present, with new facts, instead of re-fighting the past with none. That is implicit guidance working as it should: shared orientation needs fewer explicit instructions to act coherently.
Taiichi Ohno’s dictum hangs over this — without a standard, there can be no improvement. The room did not pretend the job had one. It made the current state visible enough that, as patterns settled, standard work could be written from what the walls had taught — the current best way, owned by the people doing it. Exploration first; standard where it honestly fits. Boyd called that kind of assembling the snowmobile: skis from one system, tread from another, engine from a third — a vehicle none of the donors contained. Together, those unexotic parts became a machine the fragments could not produce alone.
The honest cost
The main cost is a person. The senior lead who keeps the room true is consumed by it, full-time. He runs no crew and produces no billable task work. His output is that the room does not lie. Let that slip for one shift and the room becomes wallpaper: present, colorful, and ignored — worse than no room, because people still believe it. A shop proposing this has to price a senior operator as infrastructure, out loud.
The second cost is the hospital’s: physical space in a building that has just lost two floors of it, and a paper wall — marker, tape, handwriting — inside a digital institution.
The third cost arrives at the end. The room dies the day the job ends unless somebody photographs and archives it, wall by wall, while it is still true. The learning in it is as perishable as marker ink, and most shops will skip the archiving because the building is dry and everyone is tired. That is exactly when the next large loss starts again from fragments.
The practice
- Name one keeper for the whole job. Assign one lead whose only accountability is the whole picture — decisions named, sequences held, walls walked — with authority over trade-offs between divisions, and no crew to pull them back into a slice.
- Build the room before the third coordination call. Floor plan and status on one wall; pending decisions, each with a named decider and the hour they will be in the room, on a second. Update both by hand, in person, at every shift change.
- Move the criteria that matter onto the wall. If a clearance rule lives in a PDF the night shift has to hunt for, have the person who owns it write it on the wall in their own hand.
- Photograph the walls before you tear them down. Archive the room as part of demobilization, so the next job starts from what this one learned instead of from fragments again.
From the Boyd & Toyota series at Restoration Intel — what two old disciplines, tempo and standard work, look like inside a restoration shop.