Every superintendent I've worked with can walk a floor and tell you inside of ten minutes whether the crew is having a good day or a bad one. That instinct is real and it's valuable. But instinct doesn't tell you that your drywall hangers averaged 38 sheets a day in March and 29 in April, or that the slip happened the week the material hoist went down. It doesn't tell you which of your three electrical crews is quietly carrying the job. Field management software doesn't replace the walk. It puts numbers under it, so that when you sit down with a foreman who's behind, you're both looking at the same picture instead of arguing about whose memory is right.
Here's what productivity tracking actually looks like when it's done well on a jobsite, what it measures, where the numbers lie to you, and how to use them without turning your crews into a spreadsheet they resent.
What "productivity" actually means on a jobsite
Productivity is output over input. Simple enough on paper. The problem is that construction output comes in a hundred different units and the input is almost never as clean as "hours." A concrete crew's output is cubic yards placed. A framer's is board feet or wall length. An electrician's is device count, home runs, linear feet of pipe, or fixtures hung, depending on the phase. You can't roll all of that into one number and call it "site productivity" without lying to yourself.
So the useful software doesn't try. It tracks productivity at the level where it's meaningful: units of specific work against the labor hours that produced them. Everything above that — the crew average, the trade average, the project curve — is built up from those honest small measurements. If your tracking starts at the project level and tries to divide down, it's noise. If it starts at "the tile crew laid 420 square feet in 8 man-hours in the master bath group" and rolls up, it's information.
Planned versus actual: the first honest number
The cheapest, most durable productivity signal you already have is planned duration against actual duration. You committed an activity to three days in the weekly work plan. It took five. That gap is a productivity flag before you've bought any fancy analytics.
The value of a look-ahead schedule here isn't just that it holds the plan — it's that it forces you to commit to a number before the work starts, when you can't rationalize. In a short-interval schedule, you're making that commitment weekly, at the level of "this crew, this location, this week." When the actuals come back, the variance is small and specific enough to investigate. A tool like LookAheadWall is doing this quietly every time you mark an activity complete against its planned window; the planned-versus-actual delta is the raw material for everything else.
One caution that trips people up: a blown duration is not automatically a productivity problem. If the crew sat idle two of the five days waiting on an inspection or a material delivery, their productivity while working may have been fine. The activity was slow; the crew wasn't. Keep those two ideas separate, because if you chew out a foreman for a delay that was really a logistics failure on your end, you've just taught him to pad his next estimate.
Unit-based rates are where the real coaching lives
Duration variance tells you something went sideways. Unit rates tell you what and by how much. When your foreman logs "1,200 linear feet of MC cable pulled, 3 electricians, full day," the software turns that into a rate — roughly 50 feet per man-hour — that you can compare against the same crew last week, against your other crews, and against what you estimated in the bid.
Track it over a few weeks and patterns show up that no walk-through catches. Maybe the crew's rate is strong on the open floors and collapses in the tight mechanical rooms — that's a coordination and access problem, not a skill problem. Maybe every crew's rate on a given assembly is below your estimate, which means the estimate was wrong, not the crews, and you'd better fix it before the next phase or the next bid. The number doesn't just grade the crew; it points at the cause.
A rule of thumb for setting daily targets: take your reliable historical rate, not your best-day rate. The best day is a fluke you'll spend the whole job failing to repeat. Plan to the honest average and you'll hit your weekly work plan far more often, which matters more for the sequence than any single heroic day.
Percent Plan Complete: reliability before speed
If you take one metric away from this, make it Percent Plan Complete. PPC is dead simple: of the tasks you committed to this week, what percentage did you actually finish? Committed to twelve, finished nine, that's 75 percent. It comes straight out of the Last Planner mindset, and it measures something more important than raw speed — it measures whether your plan can be trusted.
Here's why it beats a pure productivity number. A crew can be fast and still wreck the job if nobody can predict what they'll finish, because every trade downstream is planning around their commitments. A team running 90 percent PPC at a modest pace will beat a team running 55 percent PPC full of sprints, every time, because the sequence holds and the follow-on trades aren't constantly getting stood up at the door. When you track PPC weekly and it climbs, productivity almost always follows, because reliable planning kills the waiting and rework that eat real hours.
The gold in PPC is the second question: for every task you didn't complete, why not? Log the reason every single week — prerequisite work not done, material late, RFI open, manpower short, weather, changed conditions. After a month you'll have a ranked list of what's actually stealing your production, and it's almost never "the crew is lazy." It's usually the same two or three upstream failures over and over. Fix those and the whole board speeds up.
The numbers that will lie to you
Productivity data is only as honest as the conditions you record alongside it. A few traps worth naming:
- Weather and access. A masonry crew's rate in a 40-degree wind isn't comparable to a still 70-degree day, and everyone knows it — but only if it's written down. Log the conditions or you'll draw unfair conclusions from fair data.
- Learning curve. The first repetitive unit — first floor of a repeating layout, first unit of a multifamily stack — is always the slow one. Judge a crew on unit three onward, not unit one, or you'll condemn a crew that's actually on track to beat your estimate by the top floor.
- Mixed work under one line. If "rough-in" lumps together the easy corridors and the ugly mechanical penthouse, the blended rate hides both the strong performance and the real problem. Break the activity down far enough that the number means one thing.
- Gaming the units. If you reward raw output with no eye on quality, you'll get output with no eye on quality. The punch list and the inspection pass rate belong right next to the production rate, or you're optimizing for the wrong thing.
Real-time beats the post-mortem
A monthly productivity report is a nice history book. What actually saves a day is seeing the number move while you can still do something about it. If your target was 40 units and the morning huddle already shows the crew at a pace that lands them at 30, you have until lunch to figure out why — material short, a tool down, the layout crew behind them — and shuffle people to recover. Weekly-only tracking finds that same shortfall on Friday, when it's a loss you book instead of a problem you solve.
This is exactly where a mobile field tool earns its keep. When the crew leader logs production from the floor on his phone instead of scribbling it for the trailer at end of day, the number is current enough to act on. Same-day visibility is the difference between managing production and reading its obituary.
Comparing crews without starting a war
Crew-to-crew comparison is powerful and it's the fastest way to poison morale if you handle it clumsily. The rule: never compare raw rates without normalizing for the work. Put your slowest crew on the tightest, most congested scope and of course they'll trail — that's the scope talking, not the crew. Comparison is fair only when the difficulty and conditions are close, or when you've accounted for them.
Used right, though, this is how you actually get better. When one crew consistently outproduces the others on comparable work, go watch them. Usually it's not that they move faster — it's that they stage material better, keep a cleaner work face, or hand off to the next trade without leaving a mess to argue over. That's transferable. And there's a scheduling payoff hiding in the data: once you know which crews are genuinely strong on which assemblies, you put your best crew on the critical-path activity and let the schedule slack absorb the rest. Deploying crews by their real numbers instead of by habit can pull days out of a phase without adding a single man.
Feed the numbers back into the schedule and the bid
The whole point of measuring is to make the next plan and the next bid more honest. Durations built from your own recent actuals beat durations copied from a generic productivity manual or last decade's job every time. When you build the next look-ahead, pull the real rates your crews are hitting on this job and this scope, and your weekly work plan stops being a wish list. Trade-flow sequencing gets tighter too, because you're spacing the follow-on crews by durations you've actually seen rather than ones you hoped for.
It runs all the way up to estimating. The productivity history from your completed work is the most defensible number you can put in a bid — it's what your people actually do, on your kind of work, with your kind of supervision. Bid to fantasy rates and you either lose the job or win it and bleed the margin back on the schedule. Bid to your tracked reality and you protect both.
Measure, then actually do something
None of this matters if the data just piles up in a dashboard nobody opens. The cycle that pays off is short and unglamorous: measure the rate, look at the variance, find the cause — usually in your PPC reasons, not in the crew — fix the cause, and check next week whether the number moved. Do that loop honestly, week over week, and you'll find most of your production problem was never speed at all. It was waiting, rework, and bad handoffs, and those are all things a superintendent with good numbers and a reliable weekly plan can grind down.
The instinct that let you read the floor in ten minutes is still the most valuable thing you bring. Productivity tracking just makes sure that when your gut says something's off, you can prove it, name it, and fix it — instead of arguing about it until the job's over.