Primers library/Practice notes
Practice notes
Three applied walkthroughs using the compiled record's real aggregates. Each case shows the arithmetic a working supplier actually runs — no hypotheticals, no invented numbers.
Two invitations, one proposal team
The situation. An eight-person consultancy can staff exactly one proposal this month. Two invitations are live: a federal application-maintenance RFP around $180,000, and a subcontract quote on a municipal road-resurfacing contract around $600,000.
The reading. Category competition data first: services tenders average 1.78 bidders with a 75.4% single-bid rate; works average 3.14 bidders with 34.6% single-bid. Cold baselines: about 56% win probability in the services room, about 32% in the works room — before any differentiation.
The adjustment. The works room is populated by serial bidders with estimation data from thousands of prior bids (Eurovia Québec alone: 3,159 bids, 798 wins). A first-time roads subcontractor starts below baseline there. The services room's 75.4% single-bid rate means the realistic field is one or two — so the mandatory-criteria checklist, not the technical narrative, is the contest. The consultancy's cleared personnel and prior federal references clear those gates.
The call. Bid the services RFP with a compliance-matrix-first response (Primer 03); pass on the works quote and revisit roads only after building delivery references at smaller values (Primer 07's ladder). One proposal team, deployed where the field is thin and the gates are clearable.
The quiet buyer and the contested one
The situation. An engineering firm opening a Québec office must choose where to concentrate its first year of bidding: transit authorities or infrastructure buyers.
The reading. The buyer table splits cleanly. The Société de transport de Montréal ran 5,866 tenders at 1.44 average bidders — 84.9% single-bid. The Société québécoise des infrastructures ran 3,147 tenders at 2.86 average bidders — 41.0% single-bid. The transport ministry's main tendering office: 6,569 tenders, 2.74 average bidders, 37.3% single-bid.
The interpretation. STM's numbers describe appointment procurement — quiet rooms with expected guests (Primer 10's diagnostics). SQI's and the ministry's describe contest culture: real fields, rotating winners, entries genuinely open. For a newcomer, expected first-year value is a function of contest frequency times entry probability, and only the second factor is improvable — by choosing contested rooms.
The plan. Concentrate 80% of bid capacity on SQI and ministry solicitations, priced to build references rather than to win margins in year one. Monitor STM's category postings for the structural breaks — method changes, new qualification rounds — that reopen quiet rooms (Primer 06), and enter then, not before.
The mandatory criterion you failed
The situation. A supplier loses a $220,000 federal services solicitation at the screening stage: a mandatory personnel-clearance requirement it believes the eventual winner also failed to meet at bid time. Counsel estimates six weeks of procedural work. File or not?
The reading. The registry's shape sets expectations: 2,043 complaints since 1990, running 38–108 per year recently — a remedy that is rare relative to the tens of thousands of annual awards, and concentrated among serial filers who keep documentation as routine (the top complainant alone filed 16). Outcomes split across 777 decisions, 719 determinations, and 547 orders — many matters end at screening rather than a merits finding.
The frame. Expected value = probability of a favourable finding × acceptable remedy (re-tender, compensation for bid costs), minus legal cost, six weeks of principal attention, and a relationship discount on this buyer's next solicitation. The alleged breach here is the classic merits pattern — unequal application of a pass/fail gate — but only if provable from documents: the solicitation text, both firms' submissions, and the screening notices. A supplier that ran a compliance matrix has that file already (Primer 03); one that did not is reconstructing.
The call. If the screening notices document the unequal application, file: breach, proof, and material value all hold (Primer 09's three conditions). If the evidence is inference — "we believe they weren't cleared" — request a debrief, fix the clearance for the next round, and keep the six weeks for delivery.
Run these frames against live opportunities with
Cite this page
The Procurement Institution. (2026). Practice notes — applied procurement cases. https://procurementinstitution.org/practice
@misc{pi-2026-practice,
author = {{The Procurement Institution}},
title = {Practice notes — applied procurement cases},
year = {2026},
url = {https://procurementinstitution.org/practice},
note = {Release 2026.10}
}