When the brand standard, the TPA standard, and your crew’s best method are three different sequences, none of the three is standard work on its own. Not in Toyota’s sense.
Standard work there is not the thickest manual. It is the current best known method for the work as it actually is — sequence, timing, in-process stock — owned and revised by the people closest to the work, and posted at the work.
By that test, a franchise shop can have a local standard that works at the truck and still not own it at the system level. What a franchisee can do is narrower than what a Toyota worker could do: build the better sequence, prove it on real jobs, run it where she has authority, and petition where she does not. The petition may fail, and running the better sequence without approval carries risk.
Three audiences, one loss
She is an owner-operator. Nine employees. She bought the territory six years ago, runs estimates and the schedule herself, and still rides to larger losses when the crew is thin.
Sixty percent of her revenue is program work, through two TPAs under agreements the brand signed above her shop. That work carries its own physical facts, whether the manual writes them down or not: photo documentation at set intervals, estimates in the carrier’s format, response-time clocks that start when the assignment hits the portal, not when a phone rings.
Her brand manual prescribed one sequence: start wide, establish the structure, work inward in a fixed order. Clean in a training room.
The TPA scorecard rewarded a different one. It graded source-of-loss shots first, early, on the portal clock. A wide shot on time did not save a late source shot.
Her crew had a third, built job by job: source first, because that is where the argument with an adjuster starts; moisture readings in the order the meters were already in their hands; wides on the way out, once equipment was placed. Faster — and documented nowhere except in habit.
So a tech photographed the same room three ways: brand order for a brand audit, scorecard order for upload, crew order to keep his own notes straight for the estimate. She timed it: an estimated 40 minutes per job on duplicate documentation — re-walking rooms, moving a containment flap to get a shot he already had at a different point in the sequence.
And they still took deductions for “late initial photos” — not because the photos were late to the loss, but because the brand sequence put wide shots first, and by the time the tech reached the source shots the scorecard graded first, the portal clock had marked them late.
Go and see the sequence, not the manual
Taiichi Ohno’s line matters here: “Where there is no standard, there can be no kaizen.” No standard, no improvement — nothing stable enough to improve against. The corollary matters more for a franchisee: standards were to be set by the workers doing the job, not forced from above.
That authorship is the mechanism. In Toyota’s bargain, people closest to the work could revise the standard because Toyota granted that authority. The standard lived at the workstation: takt time, available time divided by demand, recalculated roughly every ten days; the exact sequence; the standard in-process stock. A better method became the new baseline.
None of that authority exists in a franchise agreement. Hold that thought. It comes back as cost.
First, she did the part that transfers completely: genchi genbutsu — go and see for yourself. Not the manual, not the dashboard. The job. She picked one completed loss that had drawn a late-photo deduction and mapped its documentation three ways on the shop whiteboard: brand sequence, scorecard sequence, and what her lead tech actually did — recorded by standing with him on the next job, steps as they happened.
Side by side, the mismatch Boyd would have called concept against reality was plain. The manual described a concept: a structure documented from the outside in. The scorecard described the reality she was paid inside: a claim file that lives or dies on whether source, cause, and first readings are captured before the clock runs. Her crew had already oriented to that reality — orientation, in Boyd’s terms, the decisive act of seeing the situation you are actually in. Their problem was not speed. It was running three orientations at once, in series, on every job.
So she built a snowmobile — Boyd’s word for a synthesis combining parts proven in different places into something that works in the terrain in front of you. One merged sequence: TPA-graded source shots first, inside the portal clock; brand-required shots embedded where they cost nothing extra; the crew’s moisture-map order preserved, because it matched how work moved through a wet room. One walk. One set of photos, ordered once.
Five jobs, timed
She posted the merged sequence at the work — one page, inside the truck’s documentation kit — and ran it on five jobs, timing documentation the same way she had timed the three-way version.
It passed both readings that mattered locally. Uploads cleared the scorecard with no late-photo deduction. The shot set contained every brand-required image, just no longer in brand order. And time fell by 25 minutes a job. Not the whole 40 — the merged sequence still carries steps that exist only for an audience outside the room — but 25 minutes back per job, on a nine-person shop running program volume, is capacity she can feel in a week.
Inside the crew, she could grant the other mechanism a system needs: an andon habit, jidoka in practice. Any tech who hit a loss where the merged sequence did not fit could stop, shoot what the loss required, and flag the exception the same day, no blame. That bargain was hers to make — it governed only her people. Without it, a posted sequence hardens into the forced standard Ohno warned against, and techs go back to quietly keeping second notes.
She acted as shusa — keeper of the merged method — and the whiteboard stayed up as a small obeya: the three sequences, five timed jobs, and exceptions visible instead of filed away.
The petition, and what it costs
Then she did the un-Toyota thing she could do, and confronted the Toyota thing she could not. She petitioned: submitted the merged sequence up the brand channel as a proposed revision, times attached — three-way baseline, five test jobs, deduction record before and after. Not a complaint. A worked revision, with receipts.
What she could not do was revise the standard. That is the honest strain in the lens, and this story will not soften it. A Toyota worker revising standard work exercised authority the system had deliberately granted him. A franchisee submitting a proposal is asking an authority she does not hold to change its mind. The brand owns the book for reasons that are not foolish — consistency across territories, promises to carriers at national level, liability when documentation is challenged. Her sequence can be right on every job in her territory and still be, in the system’s terms, an unapproved deviation.
The costs followed. The petition: months of committee pace, follow-up calls, re-sending the timing sheet to people who had not been in the room, no guarantee at the end. The standing risk cost more: until the brand answered, her better local standard was also a potential brand-compliance finding at the next audit, by an auditor reading the manual as written rather than the scorecard as graded. She ran it anyway, openly, whiteboard up and petition on file — her judgment was that a documented, submitted, time-proven sequence beat photographing every loss three times and eating deductions for work done on time. That judgment is hers to carry if an auditor disagrees.
Boyd stops transferring here, too. His frame was built for adversarial contests. A brand and a TPA are not opponents — they are the structure she bought into, and treating a petition as a campaign would wreck the relationships her territory runs on. What transfers is orientation, naming a concept-reality mismatch, and the snowmobile instinct to build the working synthesis instead of dying pure. The adversarial half stays on the shelf.
Locally, the result held: duplication dropped, deductions stopped, one sequence posted in the truck. At system level, the lens strained where she expected. Toyota’s standard work assumes reviser and authority sit inside the same bargain. Hers do not.
The practice
- Map one job three ways before you argue about any of them. Take a job that drew a deduction, lay brand sequence, scorecard sequence, and what your tech actually did side by side on one board — from watching the work, not memory — and count the duplicated steps. That board is your evidence.
- Build one merged sequence and time it on five jobs. Graded shots inside their clock first, other required shots embedded where they cost nothing, crew field order preserved where it matches the work. Post the one page at the work and keep the numbers. A revision proposal without times is a preference, and will be treated as one.
- Grant the stop locally, petition formally upward. Give your crew andon authority — break sequence when a loss demands it, flag it same day, no blame. Then submit the merged sequence through the brand channel with the timing sheet attached, eyes open about audit risk until you get an answer in writing.
From the Boyd & Toyota series at Restoration Intel — what two old disciplines, tempo and standard work, look like inside a restoration shop.