1. The "Furnished By/Installed By" Trap
The most common margin-killer in commercial estimating happens at the borders between CSI divisions. The mechanical engineer specifies a roof-top unit (RTU) with a variable frequency drive (VFD). The mechanical schedule says "VFD furnished with unit." The electrical drawings show a power drop to the RTU but nothing for the VFD control wiring.
Who owns the low-voltage control wiring The mechanical sub assumes electrical. The electrical sub assumes mechanical (or controls). On bid day, neither carries the number. Record the control-wiring boundary as an open RFI until the project team assigns it.
Record shared-interface questions
For fire/smoke dampers, reconcile the mechanical schedule, power connections, controls and fire-alarm interface. Ask the designer to assign missing connections before pricing, and retain the response with the bid revision.
2. Leveling the Subcontractor Bids
Illustrative bid comparison: three prices for the same core-and-shell package may have different inclusions. These amounts are hypothetical, not completed-project records:
- Bidder A: $1.2M
- Bidder B: $1.35M
- Bidder C: $850k
Price alone does not identify an omission. Compare drawing revisions, temporary power, equipment quotes, exclusions, alternates and schedule assumptions. Ask each bidder to clarify differences before choosing the bid basis.
Build a bid comparison register
Use columns for drawing revision, measured scope, furnished/installed responsibilities, exclusions, allowances, quote dates and unresolved RFIs. Keep answers traceable to the project documents.
Inspect sample deliverables3. Reflected Ceiling Plan (RCP) Clashes
Compare architectural reflected ceiling plans with mechanical routing and lighting layouts. Record potential conflicts and request designer clarification. A drawing review for estimating does not establish a coordinated model or replace engineering approval.
Changed routing, fittings, access or ceiling requirements can change quantities and fabrication scope. Price only the accepted revision, and identify unresolved changes as assumptions rather than applying a universal percentage.