A branch proves an integration concept wrong the same way a field crew proves anything else wrong: with times and what actually happened, written down while it is happening. Not with a better argument. With a log that puts the assumption on one side and the observed fact on the other, until the gap is too plain to explain away.
That is what saved one regional branch during its first big weather week under new ownership. The dashboard was green all week. The branch was losing hours it could not get back.
A good branch, on paper
The branch was doing about $9 million a year when a national platform acquired it, fourteen months before the event. It was not a turnaround. It had local commercial relationships, a dispatcher who knew which building engineers wanted a call before a truck rolled, and eleven field staff who knew the market street by street.
Integration, by the platform’s timeline, was on track. CRM migration was set for the third quarter. A unified chart of accounts would follow in the fourth. Brand conversion would come the next year. Each milestone had an owner, a date, and a status color. None of that was false.
The trouble was what the plan assumed a branch is.
In the integration concept, a branch is a reporting unit on a quarterly clock. Work arrives, work gets recorded, work rolls up. Intake, in that picture, is an administrative function — a queue to be standardized and staffed centrally. Centralizing it even produced a clean synergy case: fewer local dispatch roles, one system, one way of logging a loss.
In the field, a branch is a response business on an hourly clock. Intake is not paperwork before the work. It is the first act of mitigation.
Those two concepts can live side by side in a quiet month. A September hail-and-water event ended that.
The week the queue went quiet
The storm put roughly 60 new losses into the market in a short window — roofs, broken lines, water where water should not be, all at once.
Under the new system, assignments went through a centralized intake queue staffed two time zones away, not the local dispatcher.
On paper, the queue was working. Losses were entered. Cases were opened. Nothing looked stuck.
In the building, the clock told a different story. The centralized queue averaged about five hours to assign a loss. The old local dispatcher had averaged about twenty minutes — not because she was faster at typing, but because she knew the crews, the distances, the access problems, and which jobs could not wait until morning.
Five hours is an inconvenience for a report. For wet drywall, it is a change in scope. Water keeps moving while a case waits. A containment and extraction job becomes a demolition job. The delay shows up days later in the building, in the estimate, and in a client’s memory of who showed up when it mattered.
Twice that week, commercial clients stopped calling the system and called the branch manager’s cell directly. Water was in their buildings, and nobody had been assigned yet.
Through all of it, the dashboard stayed green. It was measuring migration milestones, and they were fine. It had no field for time from loss reported to crew assigned, and no field for a client bypassing the system. A dashboard can only go red on what it was built to watch.
When the concept and the building part company
John Boyd had a plain name for this. In Destruction and Creation, he argued that the concepts we use to make sense of the world will eventually stop matching what refined observation shows us — not because people are careless, but because the world changes and the concept does not change at the same speed.
The harder part is what happens next. A mismatched concept cannot reliably be repaired from inside itself. Keep looking only through the dashboard, and every new fact gets sorted into the dashboard’s categories, and the dashboard keeps confirming itself.
Boyd’s answer was to shatter the old pattern deliberately and build a new one from what observation actually shows. His snowmobile example makes the move plain: parts from different machines and settings, recombined to fit the terrain in front of you rather than the terrain the original designs assumed.
What decides whether that rebuilding goes well is orientation — how an organization reads a situation, fed by heritage and culture, prior experience, new information, and the back-and-forth of analysis and synthesis that turns information into judgment. Lock orientation to the old concept, and people make fast, confident, well-documented decisions that are wrong for the situation they are in.
Tempo is relative to the situation, never the calendar. Quarterly suits a chart of accounts. It does not suit assigning a crew to a building taking on water. The platform was slow relative to what the loss required — the only measure the client and the building use.
The branch manager did not need Boyd’s vocabulary to see the problem. She needed a way to make the platform see it without turning the conversation into branch versus corporate.
A table, not a complaint
She stopped arguing with the dashboard. Arguing kept the discussion inside the old concept: integration is on track, here are the dates, please be patient. She had heard that answer, and it was true on its own terms.
Instead she built a mismatch log — a table kept during the work, assumption in one column, observed fact with a time attached in the next. It did not editorialize or assign motive.
Assumption: intake is an administrative function that can be centralized without changing the work. Observed: intake is the first mitigation decision; a five-hour delay turned extraction into demolition where water sat unattended.
Assumption: a standardized queue treats all losses efficiently. Observed: the queue did not know local access or relationships, or which commercial sites would bypass it and call a cell phone — and two did.
Assumption: integration progress can be judged on a quarterly clock. Observed: during a declared event, the branch lives on an hourly clock, and a green week can still cost it clients it took years to earn.
A complaint can be answered with reassurance. A log has to be answered line by line. Either the times are wrong, or the assumption is. Here the times were checkable in the platform’s own system.
She took the log to the platform’s regional vice president. The ask was narrow, framed as a rebuilding, not a rejection. During declared weather events, local dispatch authority returns to the branch. The platform system remains the record — every loss logged there, nothing off the books — but it stops being the gate work must pass through before anyone can move. The branch assigns; the system records. One exemption, bounded by a verifiable condition and a visible end.
What the exemption bought, and what it cost
The exemption was granted. In the next surge, assignment times went back to minutes, because the person assigning again knew the crews and the ground. The platform kept its record; the branch kept its response time. The new pattern — central system as record, local authority as gate during events — was the snowmobile in practical form.
The log became a standing item in the branch-platform review, the more durable change. That review is orientation maintenance at platform scale: a place where new field information can alter the concept before it fails somewhere expensive. Without it, the exemption would be forgotten in the next reorganization.
None of this was free.
An exemption makes the branch a permanent special case. Integration reporting now carries an asterisk next to it, and future standardization has to work around it. The synergy model took a real dent, too — the headcount savings assumed intake would be centralized, and during the highest-volume weeks, in this market, it now is not. That is money the deal model counted and will not fully get.
She also spent political capital she may need later, for staffing or the brand conversion ahead. And mismatch logs make people uncomfortable in proportion to how accurate they are: a right log is also a record that the original concept was wrong.
Those costs are the reason to keep the log factual, narrow, and tied to times. The point is to catch the moment a concept stops matching the work, before clients have to report it.
A green dashboard is not evidence that nothing is wrong. Sometimes it is only evidence that the dashboard is watching something else.
The practice
- Keep a mismatch log during the next surge, not after it. Two columns — the plan’s assumption, the observed fact — with a time on every entry. Contemporaneous times separate a log from a recollection.
- Name the clock the work actually runs on. Put assignment time, report to crew in minutes, next to milestone reporting during an event week. If the clocks disagree, believe the building’s clock first.
- Ask for a bounded exemption, not a reversal. Name the trigger (a declared event), what returns locally (dispatch authority), and what stays central (the system of record). A narrow ask is easier to grant and defend.
- Make the log a standing agenda item after you win. Close or renew each entry out loud in the regular review. A one-time log wins one exemption; a standing log keeps the concept matched to the field.
From the Boyd & Toyota series at Restoration Intel — what two old disciplines, tempo and standard work, look like inside a restoration shop.