Anonymised, illustrative composite. An Alberta electrical subcontractor spent five months taking verbal instructions from a site superintendent on a school retrofit, and found out at closeout which of them the general contractor was willing to pay for.
At a glance
The site superintendent ran a fast-moving retrofit and preferred to direct changes in person — walk the space, point at a wall, tell the crew what to add or move. The electrician did the work as instructed and kept its own field notes, assuming the paper would sort itself out at closeout.
It did not sort itself out, because the paper was largely missing on one side of the ledger. CCDC’s own guide to this exact situation is the CCDC 16 – 2024 Guide to Changes in the Contract, which CCDC describes as explaining “the philosophy to changes embodied within CCDC documents, including key terms, the change process, methods for adjusting the contract price, and claims”. That process exists precisely because a price adjustment needs something to check the scope and price against — and a verbal instruction with no follow-up record gives a GC’s project manager nothing to verify at closeout, which makes it the easiest kind of claim to simply refuse.
Across the project, the electrician tracked roughly 80 verbal instructions totalling $238,500 in claimed extra cost. At closeout, 57 of them had some contemporaneous record beyond the electrician’s own notes — a confirming email, a marked-up drawing, a written site instruction — representing $189,300.
The remaining 23 had nothing but the electrician’s field notes, representing $49,200. The GC’s PM disputed those 23 outright, and without any independent record to point to, $189,300 of $238,500 — 79% — was what the electrician was able to substantiate and recover.
The dividing line was not whether an instruction was verbal — almost all of them were — it was whether a verbal instruction was ever converted into some written record close enough in time to serve as backup. CCDC 16’s change process assumes a paper trail behind every price adjustment; where one existed in any form, the claim held. Where the only record was the contractor’s own after-the-fact notes, it did not.
The electrician recovered $189,300 and wrote off $49,200. The firm now confirms every verbal instruction by email within 24 hours, copying the GC’s PM, so even instructions given verbally on site generate a contemporaneous record from that point forward. See change order management that avoids disputes and AI estimating for electrical contractors. A related dispute over who could legally replace a terminated electrical sub is in how a GC replaced a terminated sub without losing the schedule.
Had every one of the 80 verbal instructions been confirmed by email within 24 hours from the start of the project, the electrician's recoverable total would very likely have been the full $238,500 rather than the $189,300 actually recovered — the $49,200 written off was not disputed on its merits, it was disputed for lack of any contemporaneous record to check it against, a gap a same-day email would have closed for the cost of a few minutes per instruction.
Measure what share of your change-order value has any written record beyond your own field notes. 57 of 80 instructions here, representing 79% of claimed value, had one; the 21% that did not is roughly the share of every verbal-instruction job likely to be written off at closeout unless the habit changes before the next one starts.
$189,300 of $238,500 is exactly 79.4% — a recovery rate that reflects the documentation gap precisely, not a rounded estimate. The firm's new 24-hour email confirmation habit does not require converting every verbal instruction into a formal change order at the time it is given; CCDC 16's own change process still runs on its own timeline for pricing and approval. What the confirmation email does is create the contemporaneous record that determines, at closeout, which side of the 79%/21% line a given instruction falls on — the confirmation is evidence a claim exists, not the claim itself.
A same-day confirmation email takes a few minutes to write and costs nothing to send; the $49,200 written off represents what those same few minutes, multiplied across 23 instructions, were actually worth in hindsight. That asymmetry is the whole argument for making the habit automatic rather than judgment-based — nobody on site was deciding which verbal instructions were “important enough” to confirm and which were not; the ones that got confirmed were simply the ones somebody happened to follow up on, which is not a reliable filter for a claim worth pursuing months later.
A 30-minute call is enough to tell you whether AI pays for itself here.
