Commissioning and building turnover: the phase nobody prices and everybody blames

Why it shows up as a cost surprise

Commissioning is frequently specified in Division 01, referenced in every technical section, and priced by nobody. Subcontractors bid their scope without reading the commissioning requirements, and the GC discovers the gap at the worst possible moment — when the testing is supposed to start.

What subs routinely fail to carry in their number:

• Labor to perform the functional tests, which is not the same as startup labor and is often several days per system.
• Return trips for retesting after a deficiency is corrected.
• Instrumentation and temporary provisions required to demonstrate a scenario.
• Training delivery, including preparing materials and recording the sessions.
• Documentation: prefunctional checklists, startup reports, and the closeout package assembly.

The fix is in the buyout conversation, not in the field. When scope is leveled, every affected subcontract should name the commissioning obligations explicitly — testing labor, retest trips, training hours and documentation — so nobody has a defensible claim later.

The sequence that has to hold

Commissioning has a dependency chain, and skipping a link does not save time, it multiplies failures:

• 1. Design intent and specification review, early, before submittals close.
• 2. Submittal and shop drawing review against the commissioning plan.
• 3. Installation verification — prefunctional checklists completed by the installing contractor, honestly.
• 4. Startup by the manufacturer's representative, documented.
• 5. Testing and balancing, completed and reported before functional testing begins.
• 6. Functional performance testing, witnessed by the CxA.
• 7. Deficiency correction and retest.
• 8. Training and closeout documentation.
• 9. Seasonal or deferred testing, for modes that cannot be demonstrated at the time of turnover.

Step 5 is the one most frequently jumped. Running functional tests before balancing is complete guarantees failures that are not failures — the system was never set up to perform yet — and each one burns a witnessed testing session that has to be repeated.

Putting it in the schedule honestly

The most common scheduling error is showing commissioning as a two-week bar at the end of the project. It is not an activity, it is a phase with its own dependencies, and treating it as a block is how projects miss substantial completion.

• Work backward from the required occupancy date, not forward from the last installation.
• Allow for the retest loop. On a first-time project team, plan for a meaningful percentage of tests to fail on the first pass. Zero retests scheduled means the schedule is fiction.
• Sequence by system, not by trade. A system can be commissioned when everything it depends on is complete, which rarely aligns with how trades finish.
• Protect the stable-conditions requirement. Many tests need the building enclosed, powered from permanent service and with temporary heat off. That is a real gate, and it needs a date.
• Separate seasonal testing formally. A cooling mode test in January is a deferred activity with its own scope, budget and responsible party — not something to quietly leave undone.

Why it is almost always the controls

Ask any commissioning authority where the deficiencies come from, and the answer is the same everywhere: sequences of operation. The equipment is usually fine. The problem is what it is told to do.

• Sequences written generically and never adapted to the actual equipment installed.
• Points not mapped or mislabeled, so the graphic says one thing and the field does another.
• Sensors uncalibrated or badly located, producing correct control of an incorrect reading.
• Setpoints left at factory defaults, often discovered a year later in the energy bill.
• Interlocks untested — the smoke damper, the fire alarm shutdown, the emergency power transfer — which are the tests that actually matter for occupancy.

The practical move for a GC is to treat the controls contractor as a schedule-critical partner from the submittal stage, not as a finishing trade. Getting the sequences reviewed and agreed early is the cheapest risk reduction available on the whole project.

Training is a deliverable, not a favor

The owner's staff will operate this building for decades, and the handover usually gets thirty rushed minutes in a mechanical room. Then the callbacks begin — and most of them are not defects, they are operation.

What a defensible training deliverable looks like:

• Scheduled sessions per system, with duration specified, delivered by someone who can actually teach.
• The right audience present, confirmed in writing. Training delivered to whoever happened to be on site does not count.
• Recorded video of each session, handed over with the closeout. Facilities staff turn over, and the recording outlives them.
• Attendance documentation, signed.
• The systems manual — not the same as the O&M binder. It explains how the building is meant to be operated, including sequences, setpoints and the reasoning behind them.

The turnover package

What the owner actually receives determines how the last payment goes. A complete package usually contains:

• The commissioning report, with test results, deficiencies found and how each was resolved.
• Prefunctional checklists and startup reports for every piece of equipment.
• The balancing report, final and approved.
• O&M manuals and the systems manual.
• As-built drawings, including the controls points list and final sequences.
• Warranty documentation, with start dates tied to beneficial occupancy rather than to installation.
• Training records and recordings.
• The deferred testing plan, with dates and responsible parties.

Commissioning in renovation and tenant improvement work

Most contractors first meet commissioning on a new building, but the harder version is the one that shows up in a renovation or a tenant fit-out, where new systems have to prove themselves while tied into equipment someone else installed years ago.

• The existing conditions are the risk. A new air handler feeding old distribution, a new panel fed from aging service, new controls talking to a legacy system. If the old side underperforms, the test fails and the argument about whose problem it is starts immediately.
• Establish a baseline before demolition. Documenting how the existing systems perform while they still run is the cheapest insurance available, and almost nobody does it. Without a baseline, every deficiency is presumed to be yours.
• Define the boundary in writing. The subcontract and the commissioning plan should state exactly where the new scope ends and what happens if the existing side fails a test. Without that sentence, the retest loop has no end.
• Occupied buildings limit when you can test. Interlock and failure-mode tests interrupt service, which means nights, weekends or coordinated shutdowns negotiated with the owner weeks ahead.

On this kind of work the commissioning conversation belongs in the estimate, because the exposure is not the testing labor — it is the possibility of being asked to fix a system you did not install in order to pass a test you agreed to perform.

What it costs and what it returns

A 62,000 square foot commercial project:

• Commissioning authority fee paid by the owner: $86,000.
• Contractor-side cost carried in the subcontracts — testing labor, retests, training, documentation: roughly $71,000 across all trades.
• Deficiencies identified: 148, of which 31 would have become warranty callbacks and 9 would have prevented occupancy if found during final inspection.
• The schedule number that decides everything: the project allowed 6 weeks for commissioning and used 5. The comparable project the year before allowed 2 weeks, used 7, and paid 5 weeks of extended general conditions at $19,000 per week — $95,000 for the privilege of not planning it.

That contrast is the entire business case. Commissioning is almost never what makes a project late. Pretending it will not take time is.

Mistakes that make it painful

• Not carrying it in the buyout. Guarantees change order fights at the worst moment.
• Prefunctional checklists signed without being performed. Every one of those becomes a failed witnessed test, which costs ten times more.
• Testing before balancing. Manufactures failures that are not failures.
• Treating the CxA as an adversary. They are the fastest route to a clean turnover; a team that cooperates gets flexibility on sequencing, and a team that fights gets the letter of the specification.
• Leaving seasonal testing undefined. It becomes a dispute after final payment, when leverage is gone.
• Assuming small projects are exempt. Energy codes increasingly require commissioning on projects far smaller than most contractors expect, and discovering that at inspection is expensive.

How to run it well from day one

Three habits separate teams that finish clean from teams that finish late. First, read the commissioning specification during preconstruction and price it into every affected subcontract by name. Second, put the phase in the schedule as a sequence of system-by-system activities with a retest allowance, worked backward from the occupancy date. Third, bring the controls contractor and the CxA into the same room at submittal stage, before anything is installed, and agree the sequences of operation in writing.

Do those three things and commissioning stops being the phase that ruins the end of the job. It becomes what it was designed to be: the process that lets a contractor hand over a building and actually walk away from it.

Frequently asked questions

What is the difference between startup, testing and balancing, and commissioning?

They are four distinct steps and confusing them is what causes most commissioning failures on a jobsite. Installation means the equipment is in place and connected. Startup means the manufacturer's representative energized it and confirmed the unit runs correctly on its own. Testing and balancing means airflow and water flow were measured and adjusted to the design values, and it produces a report. Commissioning is the structured verification that the systems work together, as designed, under real operating conditions, including the scenarios nobody wants to be surprised by, such as a sensor failure, a loss of power, a fire alarm signal or a shift between occupied and unoccupied mode. That is why a building can pass every individual inspection and still fail commissioning: commissioning is the first time anyone tests the connections between trades rather than the trades themselves.

Who pays for commissioning on a construction project?

There are two separate costs and they are paid by different parties. The commissioning authority's fee is normally paid by the owner, since the CxA is hired by the owner and is often independent from both the design team and the construction team. The contractor side carries a second cost that is frequently missed entirely at bid time: the labor for subcontractors to perform the functional tests, which is not the same as startup labor and can run several days per system, return trips for retesting after deficiencies are corrected, any instrumentation or temporary provisions needed to demonstrate a scenario, the delivery and recording of training, and the assembly of prefunctional checklists, startup reports and closeout documentation. The fix belongs in the buyout conversation: every affected subcontract should name the commissioning obligations explicitly so no one has a defensible claim later.

How much time should be scheduled for commissioning?

Enough to include the retest loop, which is the part almost every schedule omits. The common error is showing commissioning as a two-week bar at the end of the project, when it is actually a phase with its own dependency chain: prefunctional checklists, documented startup, completed and reported balancing, then witnessed functional testing, then correction and retest, then training and closeout. Running functional tests before balancing is finished manufactures failures that are not really failures, and each one burns a witnessed session that has to be repeated. The right method is to work backward from the required occupancy date, sequence by system rather than by trade, allow a realistic percentage of first-pass failures, protect the requirement for stable conditions such as permanent power and an enclosed building, and formally separate any seasonal testing that cannot be demonstrated at turnover.