The Cost Curve That Never Resets: Design Errors, Long-Tail Liability, and the QA Record
A dimension flagged during design costs an hour to fix. The same error found while tenders are out re-prices a bid. Discovered mid-construction, it buys a change order, a schedule hit, and an argument about who owns it. Discovered after the building is occupied, it can mean remediation, a professional-liability claim, and a mark on the firm's name. Nothing about the error itself changed from the first day to the last — only the cost of unwinding it.
For an engineering or architecture firm, that single fact is the entire economics of quality assurance. The cost curve keeps climbing well past the day the project closes. The implication is straightforward: the discipline is not about doing more QA — it is about doing it as early as possible, and keeping the record complete long after the drawings are sealed.
Why the curve steepens with each phase
The escalation is not mysterious. In design, correcting an error means changing a line and re-checking a calculation — nothing downstream is committed. At tender, the error distorts pricing and scope for all bidders. In construction, materials are ordered and crews mobilised, so the fix carries not just its own cost but the cost of everything built on top of the mistake. After occupancy, it is no longer a construction problem at all, but a performance failure, warranty dispute, or safety question argued under a limitation clock that can run for years.
Each phase adds committed cost the last did not. That is why the earliest catch is always the cheapest, and why QA dollars spent before tender are worth a multiple of QA dollars spent during construction.
The exposure is real enough to have its own institutions
Pro-Demnity, the Ontario Association of Architects' professional-liability insurer, described its purpose in its February 2026 annual update as protecting designed work, defending architects' professional judgment, and advancing practices in an increasingly complex risk environment. A dedicated insurer exists for a reason: a design decision can return as a claim long after the drawings were sealed, and when it does, a firm's defence is only ever as strong as the file it can put on the table years later.
The gap most firms actually carry
Many engineering and architecture firms run QA/QC checklists before drawings go out. The checklist governs the moment a set ships. The liability lasts far longer than that moment. The gap most firms carry is not between intent and execution — it is between 'we reviewed it then' and 'we can prove what we reviewed, and why, years from now.'
A calculation without the review that checked it, or a revision without the RFI that drove it, is half a defence. A complete, time-stamped project file is the other half — it shortens disputes, supports the insurer, and often keeps a weak claim from becoming an expensive one. Coverage and record are not substitutes. The record is what makes the coverage work.
What a defensible QA record contains
Drawings and their full revision history — so the current set is unmistakable and QA runs against the right version.
Calculations with the review that checked them — not just the result, but the reasoning behind each decision.
RFIs and the review comments that closed them — tracking why each revision was made, not only what was changed.
Approvals with timestamps — so the sequence of decisions is reproducible under a limitation clock that may run for years.
Four disciplines for firms that want to stay on the cheap side of the curve
Push detection as far left as it will go. Every phase an error survives multiplies the cost of fixing it. The cheapest QA dollar is the one spent before tender.
Treat the record as part of the deliverable, not overhead. The sealed set is the product and the defence at once — budget for keeping it complete the way you budget for producing it.
Capture why, not only what. A revision without the RFI that drove it is an incomplete record under examination.
Make QA a system, not a hero. Consistent deliverable review should not depend on one diligent principal — the discipline needs to travel with every set, not with whoever is most meticulous that week.
The bottom line
The cost of a design error is set less by the error than by when it is caught. The meter does not stop at closeout — a sealed drawing can return as a claim years down the road. Limitation periods run for years, and a defence assembled from scattered emails after a demand letter is the weakest position a firm can be in. The firms that protect both their margin and their name are not the ones with the cleverest tools; they are the ones whose quality was caught early and whose record was never in doubt. Deliverable QA is not paperwork around the work. For a consulting firm, it is how the work stays defensible for as long as the duty of care runs.
This content was generated by AI.
Comments