The metric that actually tells you if the job is on track
Every scheduling tool on the market will bury you in numbers if you let it. Percent complete, earned value curves, a critical path that recalculates every time somebody breathes on it. Most of it is noise on a fast-moving jobsite. If you're running a weekly work plan and trying to keep six trades from stepping on each other, you don't need forty metrics. You need four or five that tell you the truth about next week and let you do something about it before Friday.
That's what a KPI actually is: not a report you print for the owner, but a number that changes what you do Monday morning. If a metric doesn't drive a decision, it's decoration. This article is about the handful of schedule indicators worth watching, what good and bad look like for each, and the traps that make the pretty ones lie to you.
Percent Plan Complete — the one lean guys got right
If you only track one number on a short-interval schedule, make it Percent Plan Complete (PPC). It comes out of the Last Planner System and it's brutally simple: of the tasks you committed to this week, what percentage got fully done? Not started, not "80% there" — done, the way you promised the following trade they'd be done.
You count it as a yes/no per commitment. Ten tasks planned, seven completed as promised, PPC is 70%. The power isn't the number itself, it's the honesty it forces. A task that's half-finished counts as a zero, because a half-finished task is exactly what wrecks the next crew's week.
Rules of thumb from the field:
- New teams often start in the 40–50% range. That's not failure, that's a baseline. If you're honest about it you'll climb.
- A crew consistently hitting 80–85% is running a genuinely reliable plan. Above that and I start wondering if they're sandbagging the commitments — padding tasks so they always look good.
- The trend matters more than any single week. A jobsite drifting from 75% down to 55% over a month has a problem brewing, even if 55% still sounds okay on paper.
The gold isn't the percentage — it's the reasons for non-completion you log next to every missed task. "Waiting on inspection," "material didn't show," "prerequisite trade wasn't done," "manpower short." Tally those over a few weeks and the pattern jumps out. Nine times out of ten it's not the crews, it's a broken constraint-removal process upstream. That's the finding you take to the next planning meeting, and it's the whole reason PPC beats percent complete.
Schedule variance and SPI — useful, and easy to misread
Schedule variance is the plain-English question: are we ahead or behind, and by how much? Compare planned progress to actual as of today. On a look-ahead you can read it straight off the wall — the work that should be done by this line versus the work that actually is.
The formal cousins are earned value metrics. Schedule Performance Index (SPI) is earned value divided by planned value: above 1.0 you're ahead, below 1.0 behind, 0.90 means you're getting about ninety cents of planned progress for every dollar of schedule you should have burned. It's a clean normalized number and it's fine for a monthly owner report.
But know its blind spot before you lean on it. SPI drifts back toward 1.0 as a job finishes, even on a late job, because near the end almost all the planned value has been earned regardless of timing. A project that's three weeks behind on the critical path can still show an SPI of 0.98 in month ten because the non-critical work all got done. That's exactly when it's most dangerous — it reads healthy while the finish date slips. SPI tells you about volume of work; it does not tell you whether the right work happened. For that you look at the critical path directly.
Critical path stability — watch the change, not the length
Everyone tracks the critical path. Fewer people track how much it's moving, and that's the more useful signal. A stable critical path running through the trades you expect — structure, then dry-in, then the long-lead MEP rough — means the plan is holding. A critical path that jumps to a different chain of activities every update means your logic is soft or your durations are guesses, and you can't trust the completion date it's spitting out.
Two things to watch:
- Near-critical paths. Anything with only a few days of total float is a critical path waiting to happen. One bad inspection or a late delivery and it goes critical overnight. Keep a short list of near-critical chains and treat them like the real thing in your look-ahead.
- Critical path length trend. If the number of days on the longest chain to completion is growing week over week, you're losing the job in slow motion. That's the number to graph.
Float consumption — your early-warning gauge
Total float is the cushion — how long an activity can slip before it moves the finish. The KPI isn't how much float you have, it's how fast you're burning it. A path that had fifteen days of float last month and has four this month is telling you something loud, even though four days still sounds like breathing room.
Plot float against time for your key chains and you get an early-warning gauge that leads the completion date by weeks. By the time the finish date itself moves, it's too late to do anything cheap about it. Float burn shows up first. A steep downward slope on a chain feeding a milestone is your cue to go find the constraint now — chase the submittal, expedite the material, add the crew — while it still costs you an email instead of overtime and a change order.
Constraint aging and lead-time hit rate
The best crews I've worked with track their constraints as hard as their production. Two numbers earn their keep:
- Constraint aging. How long has each open constraint — RFI, submittal, missing material, pending inspection, undecided detail — been sitting? An RFI that's twenty days old and feeds work three weeks out is a fire you haven't noticed yet. Sort your constraint log by age and the top of the list writes your Monday to-do list.
- Look-ahead hit rate. Of the work you pulled into the three-week window as "ready to go," how much was actually ready when the crew arrived? Low numbers here mean your screening is weak — you're planning work that can't happen, which quietly tanks PPC downstream.
Neither of these shows up in a classic CPM report, which is exactly why they're valuable. They measure whether the schedule you're about to execute is real.
Update timeliness — the metric behind every other metric
This one's unglamorous and it undoes all the others when it's ignored. If the schedule gets updated whenever someone remembers, every number above is fiction. A KPI that reads healthy off three-week-old data is worse than no KPI, because it buys false confidence.
Pick a cadence and measure adherence to it. Short-interval planning lives on a weekly rhythm: field progress captured by a set day, next week's plan committed in the same meeting, constraints reviewed every time. Track how often that cadence actually happens on schedule. A team that updates on time every week has KPIs worth trusting. A team that doesn't, doesn't — no matter how green the dashboard looks. Software helps here mostly by making the update fast enough that field crews will actually do it; a plan that takes an hour to update won't get updated. This is where a tool built for weekly work plans, LookAheadWall included, earns its keep — the update is the point, not an afterthought.
Setting targets and thresholds that mean something
A KPI with no target is trivia. Before the numbers help you, decide what good, watch-it, and act-now look like, and pin red/yellow/green thresholds to each. Rough starting points you can tune to your jobs:
- PPC: green above 80%, yellow 65–80%, red below 65% — and always read it alongside the trend and the non-completion reasons.
- SPI: yellow below 0.95, red below 0.90 — with the end-of-job caveat above firmly in mind.
- Float on key chains: flag any milestone-feeding path that drops below roughly a week of total float, and flag any chain losing float faster than the calendar is moving.
- Constraint aging: escalate anything open past your standard turnaround — often two weeks for a submittal, faster for an RFI feeding near-term work.
Set thresholds to trip before the problem is expensive, not after it's already cost you the schedule. A red that only lights up once the finish date has moved is a smoke alarm that goes off after the house burns down.
Keep the dashboard honest
A wall of gauges feels like control. It isn't. The failure mode I see most is a beautiful dashboard nobody acts on — twenty metrics, all watched, none driving a decision. Green-light theater. Everything reads fine right up until the day it obviously isn't, because the numbers were chosen to look good rather than to catch trouble early.
So put a short test on every metric you display: if this goes red, what specifically do I do? If you can't answer in one sentence, cut it. PPC red means chase the non-completion reasons in Monday's meeting. Float red means go find the constraint on that chain today. Constraint aging red means make the phone call before the work window arrives. Those are KPIs. The rest is wallpaper.
You don't need a data science degree to run a jobsite well. You need five honest numbers, updated on time, each tied to a move you'll actually make — and the discipline to act on the yellow instead of waiting for the red. Do that consistently and the schedule stops being a document you defend in meetings and starts being a tool that tells you where to point your crews next week. That's the whole game.