Here's the split that sinks a lot of otherwise solid jobs: the schedule lives in one system and the money lives in another, and nobody reconciles the two until the monthly owner meeting when it's already too late to do anything about it. The superintendent is chasing float and the PM is chasing the budget, and the first time they actually compare notes is when a cost report shows a code 30% overspent on work that's only 15% complete. By then you're not managing the problem — you're explaining it.
Good scheduling and good cost control are the same conversation. Every activity you sequence has labor hours, material, equipment, and a sub behind it, and every one of those has a dollar figure attached. The point of tying cost tracking to your schedule isn't to generate prettier reports. It's to see the overrun coming while there's still a week of the pour left to fix your crew mix, not after the concrete's cured. This is what people mean when they say a schedule should be "resource-loaded" — and why the practical version of it usually starts at the short-interval level, in your weekly work plan, not in the master CPM.
Why the master schedule is the wrong place to track cost
A lot of teams try to cost-load the master P6 schedule and then wonder why the numbers never match reality. The master schedule is a contract instrument and a milestone map. It's too coarse and too slow. By the time a master-schedule activity rolls up its actual cost, the crew has moved on and the money's spent.
Cost variance is born and dies at the crew-day level, and the crew-day level is where look-ahead scheduling lives. Your three-, four-, and six-week look-aheads are where you commit actual crews to actual work in actual locations. That's the granularity where labor productivity is either won or lost. If you want cost signal early enough to act on it, you attach it to the short-interval plan — the weekly work plan and the trade-flow sequence — not the Gantt bar that spans a whole floor.
Load labor first — it's where the money moves
On most self-perform and multi-family work, labor is 30 to 50 percent of the cost and it's the only line item that moves every single day. Material is a purchase order that either landed at the price you bought it or it didn't. Labor is a bet you re-place every morning, and the schedule is the bet slip.
The number that matters is unit rate: hours per unit of installed work. Not "how much did we spend," but "how many man-hours did it take to hang how many sheets, pour how many yards, hang how many feet of ductmain." When you plan a week's work, you're implicitly forecasting a labor cost — five carpenters times five days times eight hours is 200 hours, and the estimate says that 200 hours should install a certain quantity. Write that expectation down. At week's end, compare what you actually put in place against the 200 hours you spent.
That comparison is the whole game, and it's dead simple: if your production rate slips against the estimate two weeks running, you have a cost problem right now, not at closeout. The crews that get burned are the ones who feel "a little behind" for six weeks and never once put a number on it. A rule of thumb that has saved me more than once: any activity trending past about 1.15 on its labor unit rate — 15% more hours than estimated per unit — gets a hard look this week, not next month. That's usually the threshold where a bad access sequence, a missing material, or a crew that's too big for the work is quietly eating the job.
Earned value, minus the jargon
Earned value sounds like a controls-department buzzword, but stripped down it answers three plain questions: How much work did I plan to have done by now? How much have I actually done? And what did it cost me to do it? You're comparing planned progress, actual progress, and actual spend on the same timeline.
The mechanics only work if progress and cost share a spine, and the schedule is that spine. When you mark an activity 60% complete in your look-ahead, that percentage should pull its budgeted value with it — that's your earned value for the period. Set it beside actual hours and you immediately see the two failure signatures:
- Behind schedule, on rate: you're producing at estimate but you didn't get to enough of it — a sequencing or manpower problem. Add crew or fix the access, the unit cost is fine.
- On schedule, over rate: you hit the quantity but it cost more hours than it should have — a productivity problem. More bodies won't fix it; something about how the work is being done is wrong.
Those two get treated identically by a team that only looks at "are we behind," and they need completely opposite responses. That's the practical payoff of connecting cost to the schedule: it tells you which lever to pull.
Where the schedule and the cost report actually disagree
The disconnects are always in the same handful of places. Watch these:
- Rework hours booked to the original activity. The crew fixes something and the hours land on the same cost code as the good work, so your unit rate looks worse than your actual first-run production. If you can, carry rework as its own line — it's the single most useful number for a superintendent's own defense at closeout.
- Front-loaded material, back-loaded install. The PO hits the cost report the day the steel or the switchgear lands. If your schedule shows that material installing over the next six weeks, the cost report will scream "overspent" against near-zero earned value until the crews catch up. That's a timing artifact, not a problem — but only if you can see the install schedule next to the spend.
- Change-order work performed before the change is executed. Everybody's favorite margin-killer. The crew is out there doing directed extra work, the hours are real and booked, and the added budget doesn't exist in the system yet. Your codes go red on paper. Flag the schedule activity as change-related the day you start it, so those hours are quarantined and not silently dragging your base-contract unit rates down.
Subs, milestones, and cash flow
For subcontracted scope, the schedule is what makes a payment application defensible. Progress against your look-ahead is the evidence behind the percentage you certify. When a sub bills 40% complete on drywall and your weekly work plan shows them one floor into a four-floor building, you've got the conversation you need before you pay ahead of the work — which is the thing that actually strands you when a sub walks.
Cash flow is the same idea run forward. Your schedule is a spend-timing forecast whether you treat it as one or not. Cost-load even roughly and the near-term look-ahead becomes a projection of what you'll pour into the job over the next month and roughly when the billable milestones land. Owners and lenders care about the S-curve; the S-curve is just your schedule with dollars on it. A four- to six-week horizon is usually the sweet spot — long enough to give the office real cash-flow warning, short enough that the crew commitments behind it are actually reliable.
Forecast off actual production, not gut feel
The estimate-to-complete is where optimism goes to die. The honest way to forecast remaining cost is to take the unit rate you're actually achieving and run it across the quantities you have left. If you're running 1.2 on labor through the first third of a repetitive scope — podium decks, unit rough-ins, whatever repeats — the other two-thirds will run about 1.2 too, unless you change something concrete about the work. Assuming you'll magically "make it up later" on identical repeating work is how a small early overrun becomes the number that eats the fee.
The upside of the same discipline: repetitive work is where the schedule teaches you. Track unit rates floor over floor and the learning curve shows up as a real, bankable trend. You can tighten both the crew size and the durations on later floors with data behind you instead of hope, and pull the saved hours forward into your forecast.
Making it real without drowning in data entry
All of this dies the moment it becomes a second full-time job for the field. The trap is a parallel cost-tracking spreadsheet that the super updates on Sunday night and abandons by week three. The version that survives is the one where the cost signal falls out of work you're already doing.
You're already building a weekly work plan. You're already sequencing trade flows and committing crews to locations. Tools built for short-interval scheduling — LookAheadWall among them — let you attach expected hours or budget to those planned activities and mark real progress against them as the week runs, so the plan-versus-actual comparison is a byproduct of planning the week, not a separate chore. Time captured through a crew-leader's phone in the field flows to the same activities, so the hours land where the work was planned instead of in a shoebox for the office to reconcile later. The schedule stops being just a coordination tool and becomes the place variance shows up first.
You don't need earned-value software or a controls department to start. You need three numbers per major activity every week — hours planned, hours spent, quantity installed — anchored to the schedule you already keep. Do that consistently and you'll catch the overrun that matters while there's still a crew on it and a week of work left to fix it. That's the entire point: not to report the loss accurately after the fact, but to see it coming and never take it.