Every job I've run had a wall of metrics somewhere — a dashboard, a spreadsheet, a laminated sheet in the trailer that nobody looked at after the second week. Most of it was noise. Manpower counts, activity totals, percent-complete curves that only ever went up because we kept adding scope to the denominator. None of it told me whether next week was going to hold together.
The metrics that actually matter in short-interval scheduling are a small handful. They measure one thing: can your team predict and deliver the next week or two of work. That's the whole game. If you can reliably say what you'll get done and then get it done, the master schedule takes care of itself. If you can't, no Gantt chart is going to save you. Here's what to track, what good looks like, and — more importantly — what to do when the number goes sideways.
Percent Plan Complete (PPC) — the one that keeps you honest
PPC is the anchor. Take the tasks your crews and subs committed to for the week, count how many were fully complete by Friday, and divide. Ten committed, seven done, you're at 70%. Simple math, and it cuts through everything.
The rule that makes PPC work is binary scoring. A task is done or it isn't. "80% done" counts as zero. This feels harsh the first time a foreman argues that his guys got the whole floor except one closet, but that's exactly the discipline you want — because the trade stacked behind him can't start in "mostly finished" drywall either. Partial credit lets everyone lie to themselves.
Realistic targets: a crew new to weekly work planning usually starts around 50%. That's normal, don't panic. With a few months of honest commitment reviews you should climb into the 65–75% range, and a mature team running clean holds 80%+ week over week. If you're sitting above 90% consistently, don't celebrate yet — that usually means your foremen are sandbagging, only committing to work they already know is bulletproof and holding back the rest. A sandbagged 95% is worse than an honest 75%, because you've stopped surfacing the real constraints. You want commitments that stretch a little.
The trend matters more than any single week. One bad week from a weather day or a late material drop is noise. Three declining weeks in a row is a signal that your make-ready process is falling behind the field.
Reasons for variance — where the actual gold is
PPC tells you that you missed. The reason codes tell you why, and this is the single most valuable data you'll collect. Every task that didn't complete gets a reason, captured live in the weekly meeting while people still remember. Keep the categories short and consistent, something like:
- Prerequisite work — the trade in front of us wasn't finished
- Materials — not on site, wrong material, or damaged
- Information — waiting on an RFI, submittal, or a decision from the owner/architect
- Manpower — sub didn't field the crew they promised
- Equipment — lift, crane, or tool not available
- Directive change — we got told to do something else
- Weather / conditions
One miss in a category is life on a jobsite. The same category showing up week after week is a broken process, and now you know exactly where to aim. If "prerequisite work" is your top reason three weeks running, your trade-flow sequencing is optimistic — you're committing to follow-on work before the predecessor realistically clears. If "information" dominates, your RFI turnaround is the bottleneck and that's a conversation to have with the design team, backed by data instead of a hunch. I've walked into an OAC meeting with eight weeks of reason codes showing 40% of our misses traced to slow submittal reviews. That's a very different conversation than "we feel like reviews are slow."
Constraint resolution — the leading indicator
PPC and variance reasons are lagging indicators. They tell you what already happened. To see trouble before it lands, watch your make-ready process — the work of clearing constraints off future tasks so they're actually ready to commit.
A task in your look-ahead window isn't schedulable until its constraints are gone: design answered, materials confirmed on site, predecessor complete, permit in hand, equipment lined up, crew available. The job of the make-ready process is to find those constraints two to six weeks out and knock them down before the task hits the commitment week.
Two numbers worth watching here. First, the percentage of constraints cleared before the task reaches its commitment week — you want the vast majority, north of 90%, resolved ahead of time. If tasks are still constrained the week you need to run them, you're firefighting, not planning. Second, constraint age: how many days a constraint sits open from the day someone flagged it to the day it's resolved. A constraint that's been open 30 days and counting is a bottleneck wearing a disguise, usually pointing at one person or one vendor who isn't responding. Sort your open constraints oldest-first and the problem names itself.
This is where a real look-ahead tool earns its keep versus a spreadsheet. Tracking constraints, aging, and who owns each one by hand across a six-week window is miserable and it decays fast. Software like LookAheadWall keeps the constraint list attached to the tasks and surfaces what's aging, so the make-ready review is a five-minute scan instead of an archaeology dig. But the discipline comes first — a tool just makes the discipline cheaper to sustain.
Tasks Made Ready — throughput of the pipeline
Related to constraint resolution is a flow number: how many tasks you move from "constrained" to "ready to commit" each week. Think of it as the output of your make-ready pipeline. If you're burning through, say, 25 ready tasks a week in the field but only clearing 15 into the ready pool, you can see the shortfall coming three weeks out. You're going to run dry. That's an early warning to push harder on constraints now, before the crews are standing around next month.
PPC by trade — accountability that holds up
Roll your PPC up by subcontractor and you get an honest, defensible scorecard. The framer sitting at 85% and the plumber at 45% are two very different conversations, and the numbers keep them from turning into shouting matches. It's hard to argue with eight weeks of your own committed-versus-delivered record.
Watch two things per trade: their PPC trend, and whether they're a net creator or resolver of constraints. Some subs consistently show up ready and clear their own path. Others generate constraints for everyone downstream — leaving a mess for the next trade, or never confirming material until the week of. That pattern is worth knowing at buyout time on the next job, and it's worth raising with the sub's PM now, calmly, with the data in front of you.
Forecast accuracy and schedule variance
Look further out and ask a bigger question: when your look-ahead said a milestone would land, how close was it? Track your six-week-out completion forecasts against actuals over time. Early in a job, or with an immature planning process, you'll swing wildly. As constraint management tightens, the forecasts converge on reality. Improving forecast accuracy is one of the clearest signs your whole planning system is maturing.
On schedule variance, watch both directions. Chronic lateness is obvious trouble. Chronic earliness is a quieter problem — it usually means durations are padded, which means crews finish and float while the next trade isn't sequenced to move up. Tighten those and you compress the whole job. Random variance around zero is healthy; consistent bias in either direction is a planning defect to hunt down.
The metrics to ignore
Some numbers look productive and teach you nothing. A few to leave off the wall:
- Activity counts. More tasks in the system isn't better planning. It's often just noise that buries the tasks that matter.
- Login frequency or "system usage." How often people open the software says nothing about whether the field is running well. Plenty of dead schedules get opened daily.
- Raw percent-complete on the master schedule. Useful for the owner's progress billing, useless for running next week. It tells you where you've been, never where you're about to get stuck.
- Perfect PPC. As above — 100% almost always means sandbagging, not excellence.
Putting it on the wall
Keep the dashboard small. If it takes more than about fifteen seconds to read, nobody reads it. I'd rather see four numbers with trend lines than twenty with none — every metric should show its direction over the last six to eight weeks, because the slope is what tells you whether you're getting better or sliding. And every number on that board should have an owner and an implied action. If PPC drops, someone opens the reason codes. If a constraint ages past a threshold, someone gets a call. A metric that doesn't trigger a next step is decoration.
Practically, start narrow. Pick PPC and variance reasons, run them clean for a month to get an honest baseline, and only then set targets and layer in constraint tracking. Teams that try to measure everything on day one usually measure nothing well and quit by week three.
The bottom line
Good schedule metrics answer one question over and over: can this team say what it will do, and then do it. PPC keeps you honest about delivery. Variance reasons tell you why you fell short and where to fix it. Constraint resolution and made-ready throughput warn you before the field goes dry. Everything else is commentary.
Measure that small handful honestly, look at the trends instead of the single weeks, and — this is the part most teams skip — actually act on what the numbers tell you. A wall full of metrics nobody responds to is just expensive wallpaper. Track the few things that matter, and the schedule starts holding.