Andon does not require a cord, a light, or a flag in a software system. Underneath the slogan, it requires a bargain proven hundreds of times: if you surface a problem, help comes, you are thanked for surfacing it, and your livelihood does not get worse for telling the truth. Without that bargain, a “speak up” program is just hardware. Without it, people will watch a problem get walled in and say nothing — and they will be acting rationally.
An acquired shop cannot offer Toyota’s bargain. Can it offer something smaller, still real?
Two flags in a quarter
Picture a typical shape. A 30-person restoration shop, six months after being acquired by a larger platform. The work is familiar. The ownership is not. The platform worries about quality, and installs a program that sounds right.
Any employee can flag a job “red” in the system. A red flag triggers a supervisor review within four hours. The rollout uses language almost straight from Toyota: surface problems early. No blame. Empowerment.
In the first quarter, exactly two red flags are filed. Both are filed by the same lead. Both are filed on other crews’ work.
Meanwhile callbacks are running at 9%. In exit interviews — where people say the unsayable because they are already leaving — the same sentence keeps surfacing: “I saw the moisture behind that baseboard. I wasn’t going to be the one who stopped the job.”
The tech saw it and knew what it meant. The reporting system existed. What did not exist was a reason to believe using it would end well.
The old culture had a rule nobody needed to write down: the person who finds the problem owns the delay. Six months of new ownership had not repealed it. Why would it have? In month two, everyone watched the platform cut four positions. People do their own math after that: keep your head down, finish your part, let the problem become somebody else’s callback.
This is a mismatch problem, in John Boyd’s sense. The program’s words were oriented to a world where surfacing a problem is safe, while the shop lived in a world where it was not. The program assumed the orientation it was supposed to create.
What the cord actually is
Jidoka, the root of andon, did not start as a motivational idea. It started with Sakichi Toyoda’s loom, which stopped itself when a thread broke, so a person could deal with the cause instead of weaving defective cloth at speed. That principle — stop, surface, fix the system — became a pillar of Toyota’s production system.
Andon is the human version. The boards and lights that spread through plants like Kamigo from 1966 made a problem visible to everyone. But the common retelling gets the mechanics wrong. A worker pulling the cord, or pressing the call button, does not automatically halt the line in an act of confession. The pull summons the team leader to help. Only if the problem cannot be resolved within the cycle does the line stop. The first movement is help arriving, not the worker standing alone next to a red light.
The prescribed response matters as much. The leader’s question is diagnostic: what is the system telling us? Not who messed up. The worker is thanked for pulling. That is not politeness. It is training. A problem is a signal about the system — unclear standard work, wrong material, a sequence that invites error — not a flaw to pin on an individual. Ohno’s point stands behind it: where there is no standard there can be no kaizen, and a standard only improves if workers can safely say where it fails.
None of that works on the first pull. It works after hundreds of reactions have taught the same lesson: pull, and help comes; pull, and you are not punished. Remove that response history and the cord is decoration. People learn from what happens after a pull, not from the rollout deck.
Underneath the response history sits the part most copies leave out. Toyota’s candor rested on an implicit bargain — long-term employment security that made surfacing a problem rational. If you will not be discarded for the line stopping, you can afford to stop it. If your security depends on never being associated with a stop, you cannot. The bargain is the economic ground the practice stands on.
An acquired shop does not have that ground. The platform may sell the branch, restructure it, or replace its leadership in five years. Everyone knows it, because they have already watched an acquisition happen once.
The expensive thing, done in public
What does the smaller, honest version look like? One manager spending money in public.
On a drying job in that same shop, on day three, a tech found microbial growth behind a cabinet run — the exact failure mode the branch manager had been briefed on, and easy to quietly work around if you wanted the job to stay on schedule.
The manager stopped the job. He paid the crew for a standby day instead of sending them home short. He called the homeowner himself and explained the delay as the system working: we found something that should not be sealed in, and we are dealing with it now instead of handing you a bigger problem later. In the Friday huddle, he credited the tech by name for the stop, not just for the find. The find was skill. The stop was the behavior the shop said it wanted.
He did not give a speech about culture. He answered a flag the way Toyota’s team leader answers a pull: he came with help, absorbed the cost at his level, and treated the growth as a signal — about scope and inspection points on days one and two — rather than as that tech’s problem. That is genchi genbutsu with a price tag: go and see, then own what you saw. It is also a snowmobile move, in Boyd’s terms — Toyota’s response discipline fitted to a shop with no lifetime bargain to stand on. The crew had heard the program described. Now they had watched it cost someone above them something.
It worked, within limits. Red flags tripled the next quarter. Two stops that followed prevented five-figure rebuild rework — work that would have been torn out and redone at the shop’s expense, the familiar tax on earlier silence. The 9% callback rate did not become zero. What changed was narrower: people started believing a flag might bring help instead of trouble, because they had seen it happen.
The cost, without softening it
That stop cost real money: a crew paid for a standby day on a job that was already stopped. It cost real time: a manager on the phone with a homeowner, explaining a delay he could have let a subordinate deliver. And it created a real promise that has to be kept every single time after. Not most times. Every time. One punished stop — one tech who flags a job and gets a reputation for being slow, one flag that gets an eye-roll and no supervisor within the promised window, one pull that is merely unrewarded and quietly forgotten — re-teaches the old lesson instantly. The re-teaching is more powerful than the original teaching, because now it carries proof that the new program was a performance.
That is the ongoing cost. It is not a program you install. It is a debt the manager personally underwrites, flag by flag. The moment keeping the promise gets expensive in a new way — a bigger job, an angrier homeowner, a bad month — the temptation will be to make one exception. The crew will be watching for that exception, and will calibrate to it faster than to the good example.
And here is the remainder to admit out loud. A platform that may sell or restructure the branch in five years cannot honestly offer Toyota’s side of the bargain. It cannot promise lifetime security, and a manager who implies it is lying in a way his people will eventually be billed for. What this shop could offer was local and personal: this manager, in this branch, will respond this way, and will spend his own credibility and the branch’s money to prove it. That version works exactly as long as that manager is there. If he leaves, the bargain leaves with him unless his successor re-earns it pull by pull.
That is not a reason to skip it. Rework avoided and problems surfaced on day three are real gains. But name it correctly. It is not Toyota’s system transplanted. It is a manager standing in for an institution his company has not built, paying the difference out of a finite account.
The practice
- Audit the response, not the flag count. Trace the last five times anyone raised a problem: who came, how fast, what did it cost the person who raised it? That record is your actual andon system.
- Make the next stop expensive at your level, free at theirs. Make the homeowner call, protect the crew’s hours, credit the stop by name. If surfacing costs the finder anything, you do not have a speak-up program yet.
- Say the limit of your bargain out loud. Promise only the response you can deliver — help within a set time, no penalty for a good-faith stop, credit for surfacing — and keep it without exception.
From the Boyd & Toyota series at Restoration Intel — what two old disciplines, tempo and standard work, look like inside a restoration shop.