Procurement planning and strategy

Procurement planning answers four questions before any invitation to bid goes out: how the work will be divided into packages, which packages the contractor will perform itself, where each package will be sourced, and when each must be awarded so that the material or the trade arrives when the schedule needs it. The answers become the procurement plan, and at its core is the procurement schedule. A good plan is short, specific and owned by the project manager; a bad plan is a generic checklist nobody reads. This lesson builds the Larch Street plan step by step.

Packaging the work

A package is the unit of buying: one scope, one contract, one responsible firm. Most packages follow the trades a market is organized around, so the starting point is usually the estimate's trade breakdown, which follows MasterFormat divisions. But the contractor still makes choices: whether to combine related scopes, split large ones, separate supply from installation, or carve out items that create interface risk.

The right answer depends on the local market. In a busy market with few qualified mechanical contractors, splitting mechanical into plumbing and HVAC may bring in firms that could not take on the whole scope. In a slow market, combining scopes may attract strong bidders looking for volume. The contractor's own capacity matters too: a firm with a strong carpentry crew might self-perform doors and hardware installation and buy only the supply.

The responsibility matrix

Once packages are chosen, the contractor must make sure that every item of work belongs to exactly one package. The tool is a responsibility matrix, also called a division of work or scope matrix: a table with items down the side and packages across the top, with one mark per row. It is built from the specifications, the drawings and experience of what falls between trades. Items that commonly fall into gaps include firestopping, blocking and backing in walls, temporary heat and protection, final electrical connections to mechanical equipment, roof curbs and penetrations, access panels, and cleaning.

# Check a responsibility matrix: every item must have exactly one owner
matrix = {
    "Rooftop unit curbs, supply":            ["Mechanical"],
    "Rooftop unit curbs, flashing":          ["Roofing"],
    "Power wiring to rooftop units":         ["Electrical"],
    "Control wiring for rooftop units":      ["Mechanical", "Electrical"],
    "Firestopping, electrical penetrations": ["Electrical"],
    "Blocking for wall-hung fixtures":       [],
    "Access panels, install":                ["Drywall"],
}
for item, owners in matrix.items():
    if len(owners) == 0:
        print(f"GAP      {item}")
    elif len(owners) > 1:
        print(f"OVERLAP  {item}: {', '.join(owners)}")
print("Items with exactly one owner:",
      sum(1 for o in matrix.values() if len(o) == 1), "of", len(matrix))

OVERLAP  Control wiring for rooftop units: Mechanical, Electrical
GAP      Blocking for wall-hung fixtures
Items with exactly one owner: 5 of 7

The check above finds one gap and one overlap in a draft matrix. Control wiring is a classic overlap: the mechanical specification says the controls contractor provides it, and the electrical specification says all wiring is electrical. Unless the matrix settles it before bids, both trades may price it (and the contractor pays twice) or both may exclude it. The matrix is issued with the invitation to bid so that bidders price the same scope, and it is attached to each subcontract.

Self-perform or subcontract

Many contractors keep some trades in house: concrete, rough carpentry, general labour, sometimes drywall or millwork. The decision is a make-or-buy comparison. Self-performing can earn a margin that would otherwise go to a subcontractor and gives direct control of critical-path work, but it also brings labour productivity risk, equipment costs, supervision and safety exposure.

# Make-or-buy: self-perform concrete formwork and placing, or subcontract it?
self_perform = {
    "crew wages (incl. burden)":    1_920_000,
    "formwork materials and rental":  610_000,
    "equipment (pump, tools)":        240_000,
    "supervision (foreman, share)":   180_000,
    "risk allowance (productivity)":  145_000,
}
sub_quote = 3_260_000          # best adjusted subcontract quote, same scope
markup_on_sub = 0.04           # fee Halyard can earn on subcontracted work
sp_cost = sum(self_perform.values())
print(f"Self-perform cost:         {sp_cost:>12,}")
print(f"Subcontract quote:         {sub_quote:>12,}")
print(f"Saving from self-perform:  {sub_quote - sp_cost:>12,}")
print(f"Saving as % of quote:      {(sub_quote - sp_cost) / sub_quote:>11.1%}")

Self-perform cost:            3,095,000
Subcontract quote:            3,260,000
Saving from self-perform:       165,000
Saving as % of quote:             5.1%

On these illustrative numbers, self-performing the podium formwork and placing is about $165,000 cheaper than the best subcontract quote, including an allowance for productivity risk. Halyard also values the control: the podium is on the critical path, and its own crews can adjust sequence daily. But the decision is not only about cost. If Halyard's concrete crews were fully booked on another project, the cheaper option on paper would mean hiring unknown labour, and the risk allowance would need to be much larger.

Sourcing strategy

For each package the contractor decides where it will look for bidders and how the material will reach site. Local trades know the inspectors, unions or labour pool and the weather; regional or national firms may bring capacity and buying power. For materials, the choice is often between buying through the installing subcontractor (simplest: one responsibility) and buying direct from the manufacturer or distributor (sometimes cheaper, and it secures the material early, but the contractor takes on supply risk).

The procurement schedule

The procurement schedule works backward from the date each package is needed on site. The need-by date comes from the construction schedule (CON-210). From it, the planner subtracts the time for each step of the procurement chain: delivery, fabrication, submittal approval (often several review cycles), submittal preparation, contract execution, award and bidding. The result is the latest date to issue the invitation to bid. If that date is already past, the package is late before it starts.

# Backward procurement schedule for Larch Street long-lead packages (weeks after award)
pkgs = {
    #                need_by  ship  fab  review_cycle cycles  prep  award  bid
    "Switchgear":     (56,     2,    40,       2,        2,     4,    2,    3),
    "Elevators":      (52,     3,    30,       2,        2,     5,    2,    3),
    "Windows":        (40,     2,    18,       2,        2,     5,    2,    3),
    "Rooftop units":  (50,     2,    26,       2,        1,     3,    2,    3),
}
float_weeks = 2
for name, (need, ship, fab, rev, n, prep, award, bid) in pkgs.items():
    chain = ship + fab + rev * n + prep + award + bid + float_weeks
    issue_by = need - chain
    flag = "LATE" if issue_by < 0 else ("urgent" if issue_by <= 2 else "ok")
    print(f"{name:14} need-by wk {need:2}  chain {chain:2} wk  issue bid package by wk {issue_by:3}  {flag}")

Switchgear     need-by wk 56  chain 57 wk  issue bid package by wk  -1  LATE
Elevators      need-by wk 52  chain 49 wk  issue bid package by wk   3  ok
Windows        need-by wk 40  chain 36 wk  issue bid package by wk   4  ok
Rooftop units  need-by wk 50  chain 40 wk  issue bid package by wk  10  ok

The schedule shows why switchgear and elevators must be the first packages out. Switchgear, with an illustrative fabrication time of 40 weeks, shows as late on the day of award: its latest issue date is a week before award, so even the float is already used. Elevators must go out by week 3 and windows by week 4; rooftop units have more room. Lead times for electrical distribution equipment, transformers and generators have been unusually long in recent years: in mid-2026, industry sources reported medium-voltage switchgear at roughly 52 to 80 weeks and pad-mount transformers at about 40 to 65 weeks, with some utility substation equipment taking two years or more. The durations here are illustrative; a real plan uses current quotes from suppliers, which can change from month to month.

Writing the procurement plan

The procurement plan pulls these decisions into one document, typically five to fifteen pages for a mid-size project. It is reviewed with the superintendent, estimator and senior management in the first week after award, updated as packages are bought, and filed as part of the project record.