Every builder has a gut feeling about which jobs will run late. We followed thousands of them from first slab to punch list to find out what that instinct is picking up on. It turns out the shape of a job gives it away long before anyone is ready to admit there is a problem. Based on 979 completed projects and roughly 47,300 progress readings that OnsiteIQ documented start to finish.
The momentum a project builds in its opening stretch is the single strongest tell for how the whole job will run. Line up the progress curves and almost every project lands in one of three groups, set early and rarely broken: teams that mobilize fast and keep the site dense with work stay ahead, while a slow, thin start compounds into a late, crowded finish.
By the midpoint of the schedule the gap is already decisive. Pattern 1, off to a strong launch, is 80% complete. Pattern 2 sits at 59%. Pattern 3 has stalled at just 13%, and every remaining trade is now stacked into the back half of the calendar — the shortest path to a blown completion date.
Structure and rough-in move quickly, then finishes drag. 80% complete at the midpoint, 344-day median. Reaches 50% at 30% of the timeline.
y = 100 / (1 + e^(−7.6(x − 0.30)))
The textbook S-curve: quiet mobilization, busy middle, orderly closeout. 59% at the midpoint, 389-day median. Reaches 50% at 47%.
y = 100 / (1 + e^(−8.4(x − 0.47)))
Flat for months, then everything at once. 13% at the midpoint, 422-day median, and the group most likely to slip. 50% at 64%, and the steepest climb of the three (highest k).
y = 100 / (1 + e^(−12.4(x − 0.64)))
Back-loaded jobs are not a handful of disasters dragging an average around. The entire spread moves right. If you only ever look at the average, you miss that.
Retail turns fastest at 170 days. Mixed use runs almost four times that. Median days from first slab to final walk, by what was being built.
Measure every active project against the ones that came before it. We will walk your portfolio with you and show you which jobs are tracking late.
