05 · Procurement Basics¶
Most projects buy something — contractors, software licences, hardware, professional services. The moment money crosses an organisational boundary, the project manager's informal authority disappears and is replaced by a contract. This module covers the contract types and what each one does to risk allocation, how to run a defensible vendor selection, and the arithmetic of incentive contracts that catches out PMs who sign a fixed-price-incentive deal without knowing where its ceiling bites.
The procurement lifecycle¶
| Phase | Key activity | Output |
|---|---|---|
| Plan | Make-or-buy analysis, choose contract type | Procurement management plan |
| Conduct | Issue solicitation, evaluate, negotiate, award | Signed contract, SOW |
| Control | Manage performance, changes, payments, disputes | Performance reports, change orders |
| Close | Verify deliverables, settle claims, release retention | Closure certificate, lessons learned |
The three solicitation documents are not interchangeable, and using the wrong one wastes weeks:
| Document | Use when | You are asking for |
|---|---|---|
| RFI — Request for Information | You don't yet know what the market offers | Capability information, no commitment |
| RFQ — Request for Quotation | The requirement is fully specified and standard | A price |
| RFP — Request for Proposal | You know the problem, not the solution | A proposed approach and a price |
If you issue an RFQ for something you cannot fully specify, you will receive cheap quotes against a specification that turns out to be wrong, then pay for every gap through change orders. Issue an RFP.
Contract types and where the risk sits¶
This is the central table of the module. The contract type is a risk allocation decision, not a payment mechanism.
| Type | Buyer pays | Cost-overrun risk | Use when |
|---|---|---|---|
| FFP — Firm Fixed Price | A fixed sum | Seller | Scope is precisely defined and stable |
| FPIF — Fixed Price Incentive Fee | Cost + fee, sharing over/underrun, capped by a ceiling | Shared, then seller above ceiling | Scope defined, but you want a cost incentive |
| FP-EPA — Fixed Price with Economic Price Adjustment | Fixed, indexed to inflation/FX | Shared on macro factors | Multi-year deals with commodity or FX exposure |
| CPFF — Cost Plus Fixed Fee | All allowable costs + a fixed fee | Buyer | R&D; scope genuinely unknown at signature |
| CPIF — Cost Plus Incentive Fee | Costs + fee that varies with performance | Buyer, partly shared | Unknown scope, but cost control matters |
| CPAF — Cost Plus Award Fee | Costs + subjective award fee | Buyer | Long services deals judged on quality |
| T&M — Time and Materials | Agreed rates × hours + materials | Buyer | Staff augmentation, small or urgent work |
Three practical rules follow from this table.
A fixed price does not remove risk; it converts it into price and adversarial behaviour. The seller prices in a contingency you cannot see, and the moment your requirement moves, every clarification becomes a change order with a margin attached. FFP is right when scope is genuinely stable — and expensive when it is not.
Cost-plus contracts require you to audit. You are paying actual costs, so "allowable cost" must be defined in the contract and someone on your side must check invoices. A CPFF contract with no cost verification is an open cheque book.
T&M must always be capped. Include a not-to-exceed value and a review point, or it becomes cost-plus with no fee ceiling. T&M is for small or urgent work; if it runs longer than a quarter, convert it.
Worked example: where an FPIF contract actually bites¶
You award an FPIF contract on these terms:
| Parameter | Value |
|---|---|
| Target cost | $800,000 |
| Target fee | $90,000 |
| Target price | $890,000 |
| Ceiling price | $1,000,000 |
| Share ratio (buyer / seller) | 80 / 20 |
The share ratio means that for every dollar of overrun, the buyer absorbs 80 cents and the seller 20 cents — until the ceiling, above which the seller absorbs everything. The seller's fee is:
| Actual cost | Fee calculation | Fee | Final price |
|---|---|---|---|
| $750,000 | 90,000 + (800,000−750,000) × 0.20 | $100,000 | $850,000 |
| $800,000 | 90,000 + 0 | $90,000 | $890,000 |
| $850,000 | 90,000 + (800,000−850,000) × 0.20 | $80,000 | $930,000 |
| $937,500 | 90,000 + (800,000−937,500) × 0.20 | $62,500 | $1,000,000 |
| $980,000 | 90,000 + (800,000−980,000) × 0.20 | $54,000 | ~~$1,034,000~~ $1,000,000 |
The number that matters is the Point of Total Assumption (PTA) — the actual cost above which the seller absorbs 100% of every further dollar:
PTA = ((Ceiling price − Target price) / buyer share) + Target cost
= ((1,000,000 − 890,000) / 0.80) + 800,000
= 137,500 + 800,000
= $937,500
The fourth row confirms it: at an actual cost of exactly $937,500 the final price hits the $1,000,000 ceiling precisely. At $980,000 the formula would give $1,034,000, but the ceiling holds the buyer at $1,000,000 — the seller eats the $34,000.
Why you must know your PTA. Below it, the seller loses only 20 cents per dollar of overrun, so cost discipline is weak. Above it, the seller loses a full dollar per dollar — and a seller heading past PTA on a large contract is a seller who may cut quality, staff the job down, dispute scope aggressively, or in the worst case walk away. PTA is not just a pricing number; it is an early-warning threshold. Track the seller's actual costs against it and treat approaching PTA as a risk trigger with a named response, exactly as in module 04.
Vendor selection: weighted scoring¶
Awarding on price alone is how projects acquire cheap vendors who cannot deliver. Publish the criteria and weights before you open the responses.
| Criterion | Weight | Vendor A | Vendor B | Vendor C |
|---|---|---|---|---|
| Technical approach | 0.30 | 9 → 2.70 | 7 → 2.10 | 8 → 2.40 |
| Relevant experience | 0.20 | 8 → 1.60 | 6 → 1.20 | 9 → 1.80 |
| Price | 0.25 | 6 → 1.50 | 9 → 2.25 | 7 → 1.75 |
| Delivery timeline | 0.15 | 7 → 1.05 | 8 → 1.20 | 6 → 0.90 |
| Support model | 0.10 | 8 → 0.80 | 7 → 0.70 | 9 → 0.90 |
| Weighted total | 1.00 | 7.65 | 7.45 | 7.75 |
Check Vendor C: (0.30×8) + (0.20×9) + (0.25×7) + (0.15×6) + (0.10×9) = 2.40 + 1.80 + 1.75 + 0.90 + 0.90 = 7.75.
Vendor C wins, and Vendor B — the cheapest — comes last. B scored a 9 on price, the single highest score anywhere in the table, and still lost because price carries only a quarter of the weight while its technical approach and experience were weakest. That is the model working as designed.
Two disciplines make this defensible rather than decorative. Set the weights before seeing the bids, or you will unconsciously tune them toward a preferred vendor. And note the spread: 7.75 versus 7.65 is 1.3% — too close to call on the scores alone. When the top two are within a few percent, say so, and break the tie with a reference call, a paid proof-of-concept, or a risk assessment. Also apply screening criteria (mandatory pass/fail: insurance, certifications, financial stability) before scoring, so an unqualified bidder cannot score its way in.
The contract is the scope baseline once it is signed
Anything not in the statement of work is a change order, priced by the seller when they have no competition left. Before signature, walk the SOW against your WBS line by line and confirm every deliverable, acceptance criterion, and interface is named. An hour spent here is worth more than any amount of contract management afterwards.
How It Actually Works¶
The FPIF (Fixed Price Incentive Fee) payout curve is a piecewise linear
function of actual cost, and its mechanics explain exactly why contract
type shifts risk: below the Point of Total Assumption (PTA), the buyer and
seller share cost overruns at the negotiated share ratio (e.g. 80/20 — buyer
absorbs 80 cents of every dollar over target), so Seller payment = Target
Cost + Target Fee − ShareRatio_seller × (Actual − Target) up to the ceiling
price; once actual cost crosses the ceiling, the contract flips to
effectively Firm-Fixed-Price and the seller absorbs 100% of every
additional dollar. PTA = ((Ceiling − Target Price) / Buyer's Share) +
Target Cost is the exact cost level at which that flip happens, and it's a
number every vendor calculates privately before signing — which is why
vendor behavior often changes sharply as actual cost approaches PTA (increased
scrutiny of change orders, resistance to scope additions) even though nothing
in the contract language itself changed. Weighted vendor scoring is a
straightforward linear model, Score = Σ(criterion_weight × criterion_score),
but its hidden failure mode is weight sensitivity: because scores are
bounded (e.g. 1–5) while weights are continuous, a vendor's total ranking can
flip based on a 5-percentage-point change in a single weight — always
re-running the model with ±10% weight perturbations before finalizing a
selection reveals whether the decision is robust or an artifact of one
weighting choice.
Exercise¶
Plan a real procurement for a project of your own.
- Write a make-or-buy analysis for one significant component, with at least three factors on each side and a stated recommendation.
- Choose a contract type from the table and justify it by naming who carries the cost-overrun risk and why that is the right allocation here. State what would have to change for you to choose differently.
- Draft the solicitation: state whether it is an RFI, RFQ or RFP and why, and write a statement of work with at least five deliverables, each with a testable acceptance criterion.
- Build a weighted scoring matrix with at least five criteria, weights summing to 1.00, and three hypothetical vendors. Compute every weighted score and the totals. State the winner, and whether the margin is decisive or too close to call.
- Set up an FPIF deal of your own with a target cost, target fee, ceiling and share ratio. Compute the PTA, then compute the final price at three actual-cost levels — one below target, one between target and PTA, and one above PTA. Confirm your PTA is right by checking that the price at that cost equals the ceiling exactly.
- Name the one contract clause you would most want in place before signing, and what it protects you from.