Anonymised, illustrative composite. An Ontario general contractor kept losing foremen within a year of hiring them, and exit interviews never named a real cause — until someone checked the dispatch logs against the rest periods the Employment Standards Act actually requires.
At a glance
A general contractor running several concurrent sites kept an 8-person foreman roster, one per active project. Over a 12-month stretch, five of the eight left, most citing vague burnout language in exit interviews rather than anything the firm could act on directly. Recruiting and standing up a replacement foreman on an active site is disruptive on its own, and the pattern kept repeating.
A scheduling audit of six months of dispatch logs across the eight foremen — roughly 1,040 shift assignments — found 156 instances where a foreman was scheduled to start a new shift with fewer than 11 hours off since the end of the last one, and 9 instances where a foreman went two or more consecutive weeks without a full rest day. Ontario’s Employment Standards Act sets at least 11 consecutive hours off work each day as a requirement that cannot be waived by agreement, on top of 24 consecutive hours off each week or 48 hours off in every two-week period. The violations were not evenly spread — they clustered specifically around weeks when two or more sites launched or hit a milestone at the same time, and foremen were the position most often pulled to cover the gap.
The eight foremen were, on average, the most experienced and hardest-to-replace people on any given site. Comparing the departed five against the three who stayed showed no obvious skill or tenure difference — what differed was how often each one had been assigned into the 156 short-rest shifts, with departures concentrated among the foremen who had absorbed the most of them.
The firm built a hard-blocking rule directly into its scheduling system: no assignment may leave a foreman with fewer than 11 consecutive hours off before a new shift, and no assignment may leave a foreman without a full 24-hour rest day at least once in a work week, matching the ESA floor exactly rather than a softer internal target. Unlike the old process, a scheduler cannot override the block by exception — a site launch competing for the same foreman has to be resolved by reassigning the work, not by shaving the rest period.
In the twelve months after the rule went live, short-rest assignments fell to zero — the block does not allow them to be scheduled in the first place, so there was nothing left to audit after the fact. Foreman turnover over that same period fell to 1 of 8 (12.5%), down from 5 of 8 (62.5%) the year before. The industry backdrop makes the stakes clear: construction employs more than 1.6 million people and contributes approximately 7% of Canada’s GDP nationally — an experienced foreman is not a role a firm replaces from a deep bench.
For the retention question this sits inside, see retaining field staff in a tight market, and for the hiring side of replacing a departure, see building a hiring pipeline for the trades.
On the firm’s own project-margin tracking, a foreman departure cost roughly six weeks of reduced site productivity while a replacement got up to speed, on top of recruiting time. Against five departures in the prior year, that is thirty weeks of degraded site performance absorbed across the firm’s active projects — a cost that never appears as a line item, because it is spread invisibly across schedule slippage and rework on whichever sites lost their foreman that quarter. It would also have kept repeating: nothing about the old scheduling process would have self-corrected without the audit, so the same 62.5% turnover rate was the realistic baseline for every year that followed, not a one-off bad year the firm could expect to age out of.
The tell is turnover concentrated among the people a firm can least afford to lose, paired with an explanation — “burnout” — too vague to act on. Before accepting that as the final answer, pull the actual dispatch records against the ESA’s daily and weekly rest floors. A rule that “cannot be waived by agreement” is not a target a good scheduler can quietly work around during a crunch; a schedule that keeps landing inside it during exactly the weeks multiple sites compete for the same people is the pattern worth auditing first.
A 30-minute call is enough to tell you whether AI pays for itself here.
