The Report Was Written Before Anyone Went to the Site

Give me a commissioning report and thirty minutes and I can usually tell you whether the building was tested. You don't need site access and you don't need to be a mechanical engineer. You need to know what a real record of work looks like, and drive-by reports fail that test in consistent, boring ways.

Owners should learn this, because the report is the only artifact you actually receive. Everything else — the site visits, the meetings, the tests — you are taking on faith.

Read the issues log first. Skip the narrative.

The narrative sections of a commissioning report are marketing. The issues log is the evidence. Open it first and look at four things.

Count. Divide the number of issues by the building area in thousands of square feet. Under about 0.5 issues per 1,000 sf on new construction, start asking questions. This is a soft heuristic, not a standard — a small, simple, well-run project with a strong controls contractor can legitimately come in low. But low counts should prompt a conversation, not a signature.

Distribution. Sort the issues by category. A real log is dominated by controls and sequence findings: economizer changeover setpoints wrong, minimum outdoor air damper position not verified, VAV boxes with reversed heating/cooling loops, static pressure reset not implemented, occupancy schedules left at factory defaults, hot water reset curves that don't match the design. Those findings only exist if somebody sat at a workstation with the sequence of operations printed out and drove the system through its states. A log dominated by "install missing gauge," "provide access door," "label conduit," and "submit O&M manuals" is a punch list. Punch lists are a general contractor's job. You paid for something else.

Closure. Every issue should have a resolution, a resolution date, and a retest verification by the CxA. "Closed — contractor reports complete" is not verification. If more than a handful of items closed without a retest, the log is decorative.

Discovery dates. If ninety percent of the issues were discovered within the same week, you did not have a commissioning process. You had a site visit.

Then check the test procedures against the controls submittal

Ask for two documents side by side: the approved sequence of operations from the controls contractor's submittal, and the functional performance test scripts.

A real test script is derived from that sequence, line by line. If the sequence has fourteen distinct operating modes and failure responses, the script should have test steps that force the system into each of them and record the observed response against the expected response. It should have blanks that got filled in with actual values — actual damper position, actual discharge air temperature, actual time to stable.

Drive-by scripts are recognizable instantly. They are checkbox forms with "Pass / Fail" and no recorded values. They use the phrase "verify proper operation." They are identical across every system type in the building. And most tellingly, they are identical to the scripts in the same firm's report for a completely different project — which you can sometimes confirm by asking a peer for a sample report and comparing.

Recorded values are the crux. A test with no numbers in it did not measure anything, and a test that did not measure anything is an opinion.

Boilerplate residue

Search the PDF for the names of other projects. It happens more than you'd think. Search for system types the building doesn't have. Search for "TBD," "insert," "XX," and "[Project Name]."

Check the document properties: the PDF's creation date and, if the metadata survived, the original template filename. A report whose underlying document was created three days before it was issued, for a nine-month construction period, contains nine months of nothing.

None of this is clever. It's the sort of thing an owner's rep does in an afternoon and almost nobody does.

The specific lie of the photo log

Every drive-by report has a photo appendix. Forty pages of equipment photographs with captions like "AHU-3, typical." Photo logs feel like evidence. They are not evidence of testing; they are evidence of attendance. A photograph of an air handler establishes that the air handler exists and that a person stood near it.

What would be evidence: trend logs. Screenshots from the building automation system showing the actual response of the actual system to the actual test, timestamped, over a period long enough to show behavior rather than a moment. Discharge air temperature and outdoor air temperature and damper position on the same axis across an economizer changeover. A static pressure reset trend across an occupied day. Twenty minutes of a chilled water plant staging up.

Trend data is hard to fake and easy to interpret, which is precisely why drive-by reports don't contain it. If I could push one change into how owners review commissioning deliverables, it would be this: require trend logs as a condition of final payment, and require that the trends span real operating conditions rather than a forced test.

Why this matters more than it sounds

There is a temptation to treat all of this as a documentation-quality complaint. Sloppy paperwork, who cares, the building works.

But the building often doesn't work, and the report is the reason nobody finds out. An accepted commissioning report closes the question. It ends the owner's inquiry, releases retainage, satisfies the code official, earns the LEED prerequisite, and starts the warranty clock running toward expiration. It is the instrument by which an untested building becomes, officially, a tested one.

That's not a paperwork problem. That's the whole problem, wearing a paperwork costume.