What You Take On When You Touch Somebody Else's Work

Why this matters

The moment your name is on the last invoice, you are the shop that was there. Not the installer, not the two companies before you, you. Whether that is fair is beside the point; what matters is that the amount you actually inherit is not fixed by fairness or by how long you were on site, and it is not fixed by what you charged. It is fixed by how deep you reached into the system, and by whether you said, in writing, what you did not evaluate.

Depth, not duration, sets what you inherit

Three depths, and the customer's reasonable belief about what you approved widens at each one. A shop that stays at depth one for twenty visits inherits less than a shop that goes to depth three once.

Depth 1, component replacement. You replaced a defined part with a like part and returned the system to the state it was in before that part failed. Nothing about the rest of the system was evaluated, and no reasonable person thinks otherwise unless you invited them to.

Depth 2, subsystem intervention. You adjusted, charged, balanced, re-piped, re-terminated or re-programmed something. You now own the settings you established and the correctness of the work you performed, and customers, adjusters and their counsel routinely read the job as your acceptance of that subsystem for that work, and you should assume the file will be read that way.

Depth 3, commissioning or re-commissioning. You measured the system against its design intent and reported it operating. You have asserted a whole-system condition. Everything you measured is yours to have measured correctly, and everything you did not measure is now yours to have said you did not measure.

The ladder is the same whether the system is mechanical, hydronic, electrical or drainage. A panel change is depth 2 and a load calculation is depth 3. Cutting in a fixture is depth 2 and pressure testing the branch is depth 3.

What you deliberately do not take on

This is the part shops get wrong, because the instinct is to disclaim everything vaguely rather than to name specific things precisely. A vague disclaimer does not communicate anything the customer can act on, which is the job it was written to do. These are the exclusions worth naming, each with the reason it is legitimate.

Design adequacy you did not evaluate. Replacing a component does not make you the designer. If you did not run the load, the sizing, the fixture-unit count or the circuit calculation, you did not approve the design and you should say so in one sentence. What kills shops here is the accidental assertion: telling a customer "it should be fine now" after a depth-1 repair is a design opinion you did not earn.

Code compliance of work you did not perform. You are responsible for your own work meeting the code your jurisdiction has adopted, in the edition it has adopted. You are not the inspector of record for the previous installer. But note the boundary carefully, because it is not "say nothing": you do not certify their work, and you do report what you find. Those are compatible and shops treat them as opposites.

The prior shop's warranty. Touching a system does not make the prior shop's obligation yours. Whether their warranty survives your intervention depends on its own terms and on state law, so read it before you touch anything if the customer still has it. Whether the prior workmanship warranty survives your intervention depends on its own terms, so read it before you touch anything if the customer still has it.

Parts you did not supply. A component someone else furnished carries whatever warranty its supplier gave. Who that warranty runs to is set by your state's version of the sales code and varies between states, so do not assume it reaches your customer or stops at them. Your labour to install it is yours. Keep those two separate on the invoice at the moment of installation, not afterward.

Conditions you documented and handed back. This is the strongest exclusion available to you and the one that requires actual work. A defect you found, wrote down, quoted or recommended, and the customer declined, is a condition they chose to carry. A defect you found and mentioned verbally is a condition you will be presumed to have accepted.

And the one you can never exclude: a dangerous condition you saw. No depth, no clause and no declined quote removes the duty to state a life-safety finding plainly and in writing. Combustion products entering occupied space, a compromised flue, an unsupported or corroded fuel line, scorched conductors, a missing bonding path, an inoperative relief device. If you smell gas at any time, everyone leaves the building immediately, no switches or lights are touched, no phone is used inside, and the gas utility emergency line is called from outside once clear. The commercial conversation waits, and an article that treated that as one option among several would be wrong.

The sentence that holds the line

Everything above collapses into one line on the invoice, written at the depth you actually worked:

"Scope was replacement of the failed [component]. The balance of the system was not evaluated, tested or certified as part of this visit."

Or at depth 3:

"System was measured against design at the readings listed. Conditions noted below were found and are outside this scope."

That second sentence is doing more work than it looks, because at depth 3 the readings are the boundary. What you measured, you own. What you listed as a finding, they own. What you neither measured nor listed is the gap, and it is the only place a real argument can live.

Worked example: three visits and a compressor

A shop is called to a packaged rooftop unit installed years earlier by a company no longer in business.

Visit 1, depth 1. Failed contactor, replaced like for like, 1.2 hours. Invoice reads: contactor replaced, balance of system not evaluated. Nothing else is asserted.

Visit 2, depth 2, three months later. Not cooling. The shop recovers, weighs in the correct charge and verifies operation. Now they own the charge they established and the joints they made. They are also presumed to have found the refrigerant circuit suitable to charge, which is a genuine assertion, and their invoice says so honestly: "Charge weighed in per nameplate. Circuit leak-checked at accessible joints. Coils and non-accessible sections not inspected."

Two hazards live inside that paragraph and neither is covered by the entry safety block. Recovering and charging a regulated refrigerant requires a technician certified under Section 608 of the Clean Air Act, 40 CFR Part 82 Subpart F, and it is not a task to hand a helper. Refrigerant contact freezes skin instantly, so gloves and full eye protection go on before a hose is cracked, and the work stays outdoors or in ventilated space because a heavier-than-air refrigerant release displaces oxygen at floor level in an enclosed one.

Visit 3, depth 3, at the customer's request to "get it right." Full commissioning. Airflow, static, charge, controls and safeties. Readings recorded:

That last row is the finding. 0.9 against a rated 0.5 is 80% above the external static the blower was selected for, which means the unit is moving less air than design across the evaporator. State the mechanism and the direction together: reduced evaporator airflow lowers suction saturation temperature and raises compression ratio, and if it goes far enough it collapses superheat and returns liquid to the compressor. At the other end of the range, excess airflow raises suction saturation and superheat, which is a different problem entirely and is not what these numbers say. That 0.9 reading is a duct system defect, and the duct system is not something this shop installed, quoted or touched.

Before drilling static test ports, the unit is opened at its disconnect, locked and tagged under 29 CFR 1910.333(b)(2) for the electrical hazard, and the terminals are proved dead with a meter checked live on a known source immediately before and after the test (the live-dead-live sequence is NFPA 70E-2021, 120.5, which binds through your employer's electrical safety program or your contract rather than of its own force). Ports are drilled with the blower stopped and guards in place; power is restored only after hands are clear of the wheel, because a blower that starts while a hand is in the cabinet does not care what the meter said.

The commissioning report states the 0.9 in wc measurement, identifies it as a duct restriction outside the scope of the equipment work, recommends duct correction, and records that the customer declined it that day.

Five months later the compressor fails. The customer's position is reasonable on its face: you commissioned this unit and said it was operating.

Because the report exists, the argument has a shape. The shop asserted the readings it took, and the readings it took included the one that predicts this failure. It named the duct restriction, priced the correction, and recorded the decline. The condition was handed back and stayed handed back.

Now delete that one row from the report and re-read the visit. Same three visits, same 1.2 hours, same charge, same certification. But at depth 3 the shop asserted a whole-system condition and produced no finding, so a system that failed in a way a static measurement would have predicted looks like a system that was measured badly. The difference between owning that compressor and not owning it is one line of a table, written five months before anybody needed it.

What widens your exposure without you noticing

• Verbal reassurance. "You should be good now" after a depth-1 swap is a depth-3 statement with no measurements behind it, and it is the single most common way techs give away a boundary for free.
• The free look. Walking a system you were not paid to walk and saying nothing about it. Either evaluate and report, or do not evaluate and say you did not.
• Repeat presence. No single visit widens depth, but a shop that has been the only company on a system for four years will find that everyone in the conversation, including the customer's insurer, starts from the assumption that it knows that system. That is not a reason to leave; it is a reason to keep depth-honest invoices every single time.
• Fixing a symptom whose cause you identified and did not write down. You now know something the customer does not, and knowing it without recording it is the worst of both positions.

How to verify you got this right

• Pull ten recent invoices on systems you did not install and mark each with the depth you actually worked. Then read what the invoice asserts. Any invoice asserting more than its depth is an exposure you created with a pen.
• Check whether your depth-3 reports list at least one thing outside scope. A commissioning report with no findings on somebody else's system is either an exceptional system or an incomplete report, and the second is far more common.
• Search your recent job notes for reassurance phrases ("should be fine", "good to go", "all set"). Each one is a free assertion.
• Ask whether a declined recommendation from last year could be produced in under five minutes. If it lives only in a tech's memory or a phone photo, it is not handed back, it is just known.

References

• 29 CFR 1910.333(b)(2), selection and use of work practices for electrical work; NFPA 70E-2021, 120.5, live-dead-live verification, binding through an employer electrical safety program or contract
• 40 CFR Part 82 Subpart F, Clean Air Act Section 608 technician certification for refrigerant handling
• See related: Prior Fix by Another Company Rework vs Flag vs Decline Decision Tree
• See related: The Pre-Existing Condition and How to Hand It Back
• How far a scope limitation or disclaimer reaches is set by state law and differs for consumer and commercial customers; have your own attorney review language you use repeatedly