What Institutional Work Does to a Shop That Is Not Ready

Why this matters

Winning a district, a hospital or a campus feels like the end of the hard part. It is the start of a different one. Institutional work does not stress the thing shops check before saying yes, which is whether they can perform the work. It stresses two things almost nobody measures: how much of your billed output sits unpaid at any moment, and how many paid hours it takes to deliver a billed one on a site with badges, escorts, tickets and permits.

A shop can be entirely capable of the work and still be broken by the award. That is not a warning against taking institutional work, which is some of the best work available to a small shop. It is an argument for sizing it against the two ratios that actually bind, instead of against the crew.

What follows is one shop's first year on a school district contract, in the order things went wrong.

The shop and the award

Seven technicians. Averaging about 6 billable hours a technician a day across roughly 21 working days a month, which is about 126 billed hours per technician per month and about 882 hours a month shop-wide. A residential and light commercial book, collecting in about 7 days because most of it is paid at the truck.

The award is mechanical service across a district's buildings, expected to run about 220 billed hours a month. That is roughly 25% of the shop's billed hours. Terms are net 60, and the district's actual payment behaviour, which they were honest about when asked, averages about 68 days.

The owner ran the capacity question and got a clean answer: 220 hours a month is under two technicians' worth of output, the crew has the skills, the buildings are close. Everything about that reasoning is correct and none of it is the binding constraint.

Month 2: the money was never in the account

At steady state, 68 days of float is about 2.3 months, so about 500 billed hours of district work is outstanding at any moment. Against 882 hours a month of shop-wide output, that is roughly 0.57 months of everything the shop produces, sitting in one customer's accounts payable.

Compare it to the same 220 hours on the residential book, which collects in about 7 days, or about 0.23 months, tying up around 51 hours. Same work, same crew, same hours sold, and the institutional version ties up roughly ten times the working capital per hour. That multiple is the whole difference between the two customer types, and it has nothing to do with price.

Steady state is not what broke though, and this is the part shops get wrong when they model it. The first district payment arrived 68 days after the first month's work, which is partway through month 3. Months 1 and 2 were funded entirely by the shop, at full volume, while payroll ran every two weeks. The ramp carries the whole float before a single payment lands, and the ramp is short and unavoidable.

The shop drew on its line of credit for payroll in month 2 for the first time in nine years. Nothing was wrong with the account. It was doing exactly what net 60 does.

Month 4: an outage window is not a scheduling preference

Institutional systems come down when the building can spare them, which on a school district means summer, winter break and spring break, and on a hospital means almost never without a formal process. Those windows are set by an academic calendar published a year in advance and are not movable by anyone you will ever speak to.

The shop's residential peak is the same summer. Two of seven technicians were committed to district work during the eight weeks that produce the highest-margin residential volume of the year, and the residential backlog stretched from three days to about two weeks. Some of it did not wait.

The lost work does not appear anywhere in the district account's job costing. It shows up as a soft quarter on the other book, which is why almost no shop ever attributes it correctly.

One thing inside that window is worth naming because it carries a hazard that the calendar pressure makes worse. A summer boiler outage frequently means entry into a space that meets the definition of a permit-required confined space, and the permit exists to protect the person inside from an atmosphere that can change after entry, with an attendant outside and a rescue plan that does not consist of another technician climbing in after them. That is 29 CFR 1910.146 on general industry work and 29 CFR 1926 Subpart AA where the work is construction, and which Part applies turns on the character of the job rather than on the building. Nobody enters on a schedule pressure argument, and the attendant is not a spare pair of hands for passing tools.

Month 6: the labor basis was about 12% light

A job-cost review found what the bid had missed. The 220 billed hours a month were arriving with about 41 hours a month of paid, non-billable time attached that the residential book does not carry:

• about 19 hours of sign-in, badge-in, escort waiting and access time, at roughly 25 minutes per site visit across about 46 visits a month
• about 9 hours of entering work into the district's own maintenance software, on top of the shop's own paperwork, at roughly 12 minutes per work order
• about 6.5 hours of certified payroll preparation, because the prevailing-wage portion of the contract requires payrolls to be submitted weekly under 29 CFR 5.5(a)(3)(ii), which converts a monthly office task into a weekly one
• about 2.5 hours of annual site orientation, spread across the year, at 4 hours per technician for all 7
• about 4 hours of badge renewals, background check paperwork and insurance certificate and endorsement handling

That is about 0.19 non-billable hours for every billed hour, against about 0.06 on the residential book. The bid was built on the residential figure, so it assumed about 1.06 paid hours to deliver a billed hour and actually needed about 1.19. The labor basis was light by about 12%, and the correction on that account's labor line is a factor of about 1.12.

Two cautions on reading those numbers. The 41 hours are absorbed effort and the 220 are gross sold work, which are two different currencies, so the ratio between them is a delivery overhead figure and not a return of any kind. And that 0.19 was measured on a district with distributed buildings and daytime access; a hospital with infection control expectations and escorted access to clinical areas runs higher, and a plant with a permit system for anything involving heat, entry or energized work higher again. The figure travels as a method, not as a constant.

Month 9: the friction landed on a technician

The most experienced technician on the district rotation gave notice. His reasons, in his own order: the sign-in and escort routine, being unable to start work while waiting for a room to be released, and after-hours calls into buildings where he could not get in without waking somebody.

None of that is a complaint about the work. It is a complaint about a job that changed shape without anyone acknowledging it, and it is the cost that never appears in a ratio. Shops that hold institutional accounts well tend to rotate the assignment deliberately and pay a differential for it, rather than assigning the account to whoever is most patient and hoping the patience holds.

The two ratios that decide readiness

Ratio A, working capital. Institutional billed hours outstanding at their actual payment behaviour, expressed as months of shop-wide billed output. Measured monthly, shop-wide, at steady state. A workable starting ceiling is 1.0 month of shop-wide billed hours, tuned to committed credit you can actually draw. When it crosses, the response is to cap new institutional hours until collections catch up, not to chase the collections, because the delay is structural and no amount of calling shortens a net 60 term.

Ratio B, delivery overhead. Non-billable hours per billed hour on the institutional book, compared to the same figure on your existing book. Measured per account, per quarter. When the institutional ratio exceeds your baseline by more than about 0.10, that account's labor line needs its own correction factor rather than a general price increase, and the factor applied is the measured one rather than a rounded guess.

Running this shop through its own test

Ratio A: about 500 hours outstanding against 882 hours a month, or 0.57 months. Under the 1.0 ceiling. It passes.

Ratio B: 0.19 against a 0.06 baseline, a gap of 0.13. Over the 0.10 threshold. It fails, and the correction is the measured 1.12.

So the ratio that failed was the pricing one, and the one that passed was the cash one, which is the opposite of how the year felt from inside the shop. Month 2 hurt more than month 6 did. That is worth sitting with: the cash pain was a ramp effect that would resolve on its own once the account was in steady rotation, painful and temporary. The labor basis error was permanent, silent, and compounding at about 12% of the labor on a quarter of the shop's hours for as long as the contract ran.

Reading the pain instead of the ratios would have led this owner to solve the wrong problem, most likely by tightening collections on an account that was paying exactly as agreed.

What would have made the same award work

Four things, all available before signing.

Take a subset of the buildings in year one. Institutions are usually willing to phase an award, and a district that has just replaced a vendor is more willing than at any other time. Half the buildings halves the ramp float and gives you a real measurement of Ratio B before you are committed to the rest.

Price the mobilization separately. Orientation, badging, background checks and the first insurance endorsements are a one-time block of hours that recurs with every new technician. Put them in as a mobilization line rather than burying them in the hourly rate, which also makes them visible as a switching cost the next time somebody suggests re-bidding.

Ask what the payment behaviour actually is, not what the terms say, and model the ramp on the answer. This district answered honestly when asked. Most will.

Set the labor correction before the first invoice, not at the first renewal. A factor introduced at month 1 is a bid assumption. The same factor introduced at month 13 is a price increase, and it will be treated as one.

References

• 29 CFR 1910.146 and 29 CFR 1926 Subpart AA, permit-required confined spaces in general industry and construction work
• 29 CFR 5.5(a)(3)(ii), weekly certified payroll submission on covered prevailing-wage contracts
• See related: Net Terms and What They Do to a Small Shop's Cash Position, What Prevailing Wage and Certified Payroll Actually Require, Badging, Background Checks and Getting Onto the Site
• See related: What an Outage Window Really Costs Both Sides, How to Get Set Up as an Approved Vendor