A building site looks like a place where things get built. Spend a morning standing on one and it becomes clear that it is mostly a place where people wait for information.
The framer waits on the pour. The electrician waits on the framer. The inspector waits on all of them. And everybody waits, at some point, on a drawing, which may or may not be the current version of that drawing.
The most expensive object on site is an old drawing
A revision goes out on a Tuesday. Five of the eight trades see it. By Thursday a wall exists in a place the current design does not have a wall. Nobody was careless. The materials were right, the crew was competent, and the work was done properly. The information was simply out of date.
Rework of this kind rarely shows up as a line item. It hides inside labor hours and material orders, and it is one of the reasons a job can be run tightly and still finish over. The industry’s traditional fix, a paper set in the site trailer and a lot of phone calls, half works, which is the worst amount of working.
A single shared repository solves a surprisingly large share of it. Drawings, permits, specifications and revisions live in one place, and the version on the phone in someone’s pocket is by definition the current one. Modern construction project management software exists largely to make the stale drawing impossible rather than merely discouraged.
A schedule is a chain, not a list
Construction schedules are unusual in that almost nothing in them is independent. A two-day delay on a concrete pour is not a two-day delay. It is two days pushed through every trade that follows, plus whatever it costs to get those crews back on site once their window has closed.
Two things make that chain manageable. The first is being able to see the dependencies at all, so the effect of Tuesday’s weather on next month’s handover is visible on Tuesday rather than next month. The second is being able to reassign work the moment the chain moves, to the crew rather than to a supervisor who then has to find the crew.
Budgets leak quietly
Overruns are rarely one dramatic event. They are an accumulation of small unrecorded things: a variation agreed verbally at nine in the morning, an extra day of plant hire nobody logged, an invoice reconciled against a milestone that was not actually finished.
The discipline that stops this is unglamorous. Costs get logged against the budget as they happen rather than at month end. Change orders get approved and signed before the extra work starts, not after it is already built. Subcontractor invoices get checked against completed milestones rather than against optimism.
The gap between what a job cost and what anyone thought it cost is almost always made of things that were true on site and never written down.
The office and the site stop guessing at each other
Most friction on a build is not conflict. It is two groups of competent people working from different pictures of the same day. The office thinks the second floor is closed in. The site knows it is not, and has known since Monday.
Closing that gap takes very little technology and a great deal of habit. A daily log with photographs attached. A request for information that goes into a queue with a name on it instead of into a voicemail. Progress notes written by the people doing the progressing.
The punch list decides how the job ends
Every build finishes with a list of small wrong things, and the handover takes as long as that list takes to close. The traditional version is a supervisor walking the floor with a clipboard, writing descriptions that have to be translated back into locations by somebody else a week later.
Pinning a defect to a point on the floor plan, with a photograph, removes the translation step entirely. The subcontractor gets the location, the image and the required fix at the same moment, which is usually the difference between a handover that slips and one that does not.
The building is the easy part
None of this makes a building. Concrete, steel, timber and skilled trades make buildings, and no amount of software has ever squared a corner.
What the coordination layer does is stop good work from being undone by bad information. On a well-run site the drawings are current, the dependencies are visible, the costs are known while they are still being incurred, and the defect list shrinks instead of circulating. That is not a technology outcome. It is what it looks like when everyone is finally working from the same page.
Edited by Ben VanderVeen · About Moss & Fog
