S-curve and earned value (EVM) on the jobsite: how to know a project will slip before it slips

The project S-curve is the tool most people in construction have on their computer and fewest actually use. Picture the scene: at the monthly meeting, the engineer projects the S-curve slide, the actual line sits right on top of the planned line, everyone nods, "we're on schedule," next project. Three weeks later that same project blows through its schedule and nobody can explain where the delay came from — because the curve on the slide was a snapshot from a month ago, thrown together the night before with whatever number the site engineer remembered off the top of his head.

It's not the fault of whoever built the curve: it's the nature of an S-curve that lives in a spreadsheet, disconnected from the jobsite — beautiful to present, blind to warn.

I spent my career on industrial pulp and paper projects, undertakings in the billions of reais, where no schedule decision gets made without earned value on the table. And I learned something that applies just as much to a Suzano plant as to your R$30 million building: the S-curve isn't there to prove the project is on schedule — it's there to catch, early, that it's about to fall off it. Most people use it for the first thing; it's the second that saves margin.

What the S-curve shows (and what it hides)

The S-curve plots, over time, how much of the project should be complete (planned progress) against how much is actually complete (earned progress). It gets its "S" shape because every project starts slow, accelerates through the middle, and decelerates at finishing. As long as actual tracks planned, the reading is "we're on pace."

The problem is what that reading hides. The classic S-curve tells you how much you progressed; it doesn't tell you what it cost to get there. And that's where the disguised delay lives.

Two projects can have the exact same physical-progress curve — 60% at month 6, exactly as planned. Except on the second one, that 60% only exists because the engineer doubled the crew, paid overtime, and bought materials in a panic to keep the work front from stalling: same curve on the slide, margin bleeding underneath, a wall coming up fast. The physical-progress curve, alone, measures progress and ignores the price of that progress. To see the real project, you cross progress against cost — and that's exactly what earned value is for.

Earned value (EVM) explained without the jargon

EVM — Earned Value Management — has a reputation for being complicated because of the alphabet soup of acronyms. I'll give you the concept without any of them. Earned value answers the question the S-curve alone doesn't: is what I delivered worth what I spent to deliver it?

Think of three numbers, all in reais:

1. What I planned to spend by today. Per the schedule and budget, I should have disbursed, say, R$ 3 million.
2. What I actually spent by today. The real tally: R$ 3.2 million.
3. What I actually delivered, valued at budget prices. I take what's physically complete on site and ask: "what did the budget say this portion would cost?" Say R$ 2.8 million. That's the earned value.

Two calculations decide everything. Ahead or behind schedule? What I delivered (R$ 2.8M) against what I planned to deliver (R$ 3M): I delivered less, I'm behind schedule — and the delay is measured in reais, not "I think maybe two weeks." Under or over cost? What I delivered (R$ 2.8M) against what I spent (R$ 3.2M): I spent more than what I delivered is worth, I'm over budget. Both objective numbers.

It's measuring progress on the ruler of money, so delay and overrun become a number instead of a gut feeling. And it gives you a head start: at month 6 of a twelve-month project, with both deviations already visible, you don't wait until month 11 to find out it won't close on target. The project warns you — if someone is measuring. And there's the problem: almost nobody measures in time, because the data lives trapped in different places.

Why the spreadsheet curve is never up to date

To work as an early warning, the S-curve and earned value need three pieces of information that are cross-referenced and fresh: the schedule (what should be complete today), the actual physical measurement (what's actually complete on site), and the realized cost (how much cash has actually gone out, including the invoice that arrived yesterday). Each one lives in a different place, in a different person's hands, in a different format.

The schedule sits in the planner's MS Project file. The measurement lives in the site engineer's head, in a field notebook, or scattered across WhatsApp photos. The cost sits in finance's ERP, which only records the invoice after someone keys it in. To build the updated curve, someone has to go to all three places, extract, reconcile, and recalculate everything in a spreadsheet — half a day of tedious work, at minimum. That's why it only happens once a month, the night before the meeting, and it's already stale by the time the slide goes up. Between one meeting and the next, the project runs thirty days in the dark, and the deviation that started in week two only shows up in next month's report, by which point it's already become an emergency purchase, overtime, and a stalled work front.

The enemy was never your team. It's trapped information — the schedule in the planner's file, the measurement in the site engineer's head, the cost in the ERP until someone posts it. Until those three come together on their own, every day, the S-curve stays a snapshot that arrives after the problem already happened. It's no coincidence that industry surveys point to nearly 88% of construction delays being linked to materials and supply issues — exactly the kind of deviation a curve disconnected from execution only sees too late. I wrote about how that delay builds up day by day in Construction delay: why your project slips a little every day (and nobody notices).

A live schedule connected to execution

The shift isn't having a prettier S-curve. It's having a living S-curve — one that updates itself because it's plugged into all three sources, instead of waiting for someone to build it the night before.

Physical measurement enters through the project's own daily routine. When the team reports progress on each work front as part of the routine — a voice note while walking the site, a photo, a quick answer on their phone — that progress already feeds the curve. Nobody "stops to update the spreadsheet," because the day's record is the update. The report stops being a nighttime chore and becomes a byproduct of work already done — worth reading more on this in Construction daily report: how to stop writing reports at night after hours.

Cost enters through the invoice, automatically. When the invoice arrives, it's read and posted to the right cost center without anyone typing it in. "How much have I spent so far" stays fresh at all times, without waiting on month-end close.

The schedule ties it all together. With planned on one side and actual physical progress and cost on the other, the S-curve and earned value recalculate continuously. The question "am I ahead or behind, under or over cost?" has an answer today, not three weeks from now — the difference between driving while looking in the rearview mirror and driving while looking through the windshield. Connecting schedule, measurement, and cost into a cycle that doesn't depend on heroics is the heart of Planning and Project Control (PCP): the method that separates the project that delivers from the project that blows up.

From indicator to an alert that demands action

This is where most tools stop halfway. Having a living S-curve is already a leap, but an up-to-date indicator, however accurate, is still passive: it waits for you to look at it. And nobody's going to open the screen on a Tuesday night to discover the earned value on the structural work front slipped — you're busy putting out three other fires at once.

That's why our thesis is that your project doesn't need one more indicator to check. It needs active coordination — an agent that watches on your behalf and acts when the number deviates. In schedule terms: it monitors progress on every work front against plan and, the day a front drifts from the curve — not a month later — alerts whoever needs to know and demands a mitigation plan: "The masonry front has underperformed for the second week running; at the current pace, the slab milestone will slip. What's being done to recover?" It's the question a good project manager would ask, if they had time to cross-check eight work fronts every single day — which no human does.

Worth being honest here: this cross-referencing only works if the data actually shows up — if the day's measurement gets reported and the invoice gets posted. That's why we come in with method and engineers during onboarding, not just software. Jobsite routines don't change because of a new screen; they change because of people who've lived the job designing the process alongside your team.

Frequently asked questions

1. I already build an S-curve at my company. What's the real difference? It comes down to whether it's alive or dead. The curve you throw together the night before the meeting is a snapshot of last month: useful for reporting, useless for acting. The curve connected to execution updates with the progress reported that day and the invoice posted automatically, and shows the deviation the moment it's born, not thirty days later. Add earned value — measuring progress on the ruler of cost — and you stop discovering overruns only at the end.

2. I already pay for Sienge/TOTVS. Doesn't the ERP give me this? The ERP records what already happened well — the invoice that came in, the cost, the measurement entered — and you should keep using it. But it's passive: it doesn't cross-reference cost with physical progress and schedule to warn you, today, that a work front is going to miss its milestone, let alone demand action from whoever needs to act. That's why we integrate with the ERP instead of replacing it: we pull cost from there and close the loop it can't close on its own.

3. Won't setting up EVM add more work for my already overloaded team? Manually, yes — that's why almost nobody does it continuously. The key is having the data feed the calculation on its own: measurement comes in through the day's report (including by voice), cost through automatic invoice reading, and the calculation redoes itself without anyone opening a spreadsheet. That tedious half-day before the meeting disappears: your team reports what they'd report anyway, and the curve builds itself from that.

4. Is R$ 10 thousand a month expensive for this? Compare it to what a delay discovered late costs: the emergency purchase to recover the front, the overtime, the contractual penalty, the idle crew. A single course correction made in time — because the living curve warned you in week three instead of the following month — usually pays for months of the platform. It's half the salary of a mid-level engineer for a project manager that cross-references schedule and cost across every work front, every day, and never forgets to demand mitigation.

5. How much exactly would I recover with real-time schedule and cost control? It depends on your project — and I'd rather be honest than sell you a number I haven't measured on your job. In our estimation model, bottom-up and conservative, the invisible waste that active coordination attacks falls in the range of 3.6% to 7.6% of project value, combining crew capacity, purchasing, and rework. I'll stress this is a model, not a measured client result: our pilot is less than a month old. What I can state today is the capacity effect — our Communication Hub gives back roughly 10 hours a week to each project manager. The rest is a defensible estimate, and the exact number is your project's to calculate.

How TrackBuild tackles invisible project delay in practice

Everything this article described comes from the same root: schedule, measurement, and cost trapped in three places that don't talk to each other. That's exactly what TrackBuild attacks:

• Budgeted vs. actual in real time — cost enters through automatic invoice reading and cross-references with the day's reported progress, keeping the S-curve and earned value alive without anyone building a spreadsheet.
• AI Project Manager — monitors every work front against plan and, when progress drifts, alerts whoever needs to know and demands mitigation the day the deviation is born, not in next month's report.
• Engineers in onboarding — the living curve depends on the day's measurement arriving and the field routine being solid; that's why implementation comes with method and engineering people designing the process alongside your team.

And because billing is per module, with unlimited users, the planner, the site engineer, finance, and the subcontractor all feed the same curve — no per-seat toll and no data trapped in anyone's WhatsApp. Already have Sienge or TOTVS? Even better: we integrate and close the loop it doesn't close on its own.

Find out how much invisible delay is costing your project

An S-curve disconnected from execution is just one of the ways a project bleeds in the dark — and earned value done in time is what brings those deviations to light while there's still time to act. If you want your project's number before deciding anything, the Leak Diagnostic is a free 30-minute call where we calculate, using your own data, how much your company loses per month to delays discovered late, panic purchases, and scattered information — no strings attached.

Talk to us

---

Felipe Barreto is co-founder of engineering at TrackBuild. A civil engineer with a master's in civil engineering from UFMG, an MBA from IESE Business School, and a PMP certification, he managed plant construction for Suzano, Klabin, and other pulp and paper industry leaders — projects in the billions of reais. TrackBuild puts AI agents and specialized engineers into jobsite routines, with unlimited users, so information stops getting trapped and margin stops leaking.