A data centre fit-out costs millions of euros, and launch day should be flawless. In our experience, most handovers include at least one "we didn't think about that" moment. Some include several. Here are ten ways things commonly go wrong — and the documentation discipline that prevents each one.
1. Missing or incomplete Inspection & Test Plans (ITPs)
Commissioning teams arrive on site and key systems have no ITPs — or the ITPs are missing acceptance criteria, witness authority, or test procedures. Testing ends up ad-hoc, defects go undocumented, and the final sign-off is weak because the rigour was never there.
- Require all ITPs before commissioning begins — non-negotiable.
- Have ITPs independently reviewed for completeness, clarity, and standards alignment.
- Include sign-off authority and witness requirements in every ITP.
- Link ITPs to commissioning gates: no Level 2 start until Level 1 ITPs are complete and signed.
2. Electrical defects — non-compliance with SFS 6000
Earth fault loop impedance out of tolerance, protective devices that don't coordinate, cables routed with mechanical damage risk, labelling missing. Found late, these halt commissioning for weeks of remediation and re-certification.
- Witness SFS 6000 Part 6 testing during construction, not after it.
- Enforce strict labelling and traceability — every circuit accounted for.
- Verify test certificates exist for continuity, insulation resistance, and Zs before accepting the pack.
3. Cooling systems that don't perform to specification
Data halls power up and outlet temperatures run high, consumption exceeds design, racks sit in thermal imbalance — because performance was never verified at part-load and full-load conditions before handover, and containment details were never checked against design.
- Include staged performance testing (25/50/75/100% load) in the commissioning plan.
- Verify BMS setpoints and calibration as part of the documentation pack.
- Require FAT and SAT records before load ramp.
4. Undocumented design changes
Field teams deviate from approved design — a valve added, a circuit re-routed — without telling the design engineer or recording the change. The as-built documentation then doesn't match the installed reality, and maintenance teams inherit drawings that are quietly wrong.
- Enforce change management: every deviation needs documented approval.
- Audit change records against the as-built pack during review.
- Verify installed reality matches the as-built record before sign-off.
5. Control systems configured but never truly tested
"It compiles and doesn't crash" is accepted as commissioning. Sensor calibration is unverified, alarm logic untested end-to-end, integration incomplete. The facility then operates without trustworthy environmental visibility.
- Require FAT (vendor demonstrates function) and SAT (site verification with real sensors) evidence.
- Demand calibration records for every sensor in the handover pack.
- Test alarm chains end-to-end — annunciation through escalation.
6. O&M manuals that don't match the building
Manuals arrive incomplete, outdated, or describing equipment that was value-engineered out months ago. The operations team is flying blind, and the first incident becomes chaos because the procedures are wrong.
- Review O&M manuals during commissioning, against what is actually installed.
- Incorporate as-built changes into the final documents before acceptance.
- Check language and clarity match the operations team who will use them.
7. Power distribution that isn't right-sized
UPS overload alarms in normal operation, phase imbalance across PDUs, transfer switches under-rated for peak transients — discovered after handover, when the fix requires downtime.
- Verify capacity calculations against the actual load profile and expansion plan.
- Witness load bank testing of UPS, ATS, and distribution at design load.
- Check the test evidence is in the pack — not promised for later.
8. Fire and life safety treated as an afterthought
Suppression uncommissioned, emergency lighting untested, alarm chains never proven end-to-end. Beyond the catastrophic risk, this is regulatory non-compliance — and insurers can decline claims where commissioning evidence is missing.
- Put life safety systems in the commissioning plan with named responsibility.
- Require witnessed testing of all detection and suppression before operational handover.
- File the certificates in the handover pack, traceable to system and date.
9. No-load commissioning
Systems are commissioned at partial load "to save time", with full-load operation deferred to live service. Issues then surface in production, where diagnosis is slow and reputations are on the line.
- Plan full-load commissioning before handover — use load banks if the IT load isn't there yet.
- Require Level 4 performance testing in the commissioning plan.
- Treat "we'll verify it in operation" as the red flag it is.
10. Weak handover sign-off
"The system is ready" — with no definition of ready. Open defects acknowledged but never formally dispositioned. The owner accepts responsibility for things they don't know about, and warranty disputes follow.
- Define handover gate criteria in writing before the project starts: all ITPs passed, NCRs closed or accepted, O&M reviewed, full-load testing passed.
- Base the go/no-go decision on an independent readiness assessment, not the contractor's self-assessment.
- Make the sign-off document list explicitly what is and is not being accepted.
The common thread
Every one of these failures leaves a documentation fingerprint long before it becomes a site problem. A missing ITP, an uncertified test, an as-built that doesn't match reality — these are all catchable in a structured documentation review, weeks before the main contractor or owner finds them.
That is exactly what we do: independent, standards-traced review of ITPs, test packs, and handover documentation, with a clear Pass/Fail recommendation. See how a review works, or tell us what you're preparing.