The Last Planner System gets described in a lot of ways that make it sound like a management fad. It isn't. Strip away the lean-construction vocabulary and what you're left with is a discipline for making the schedule tell the truth. That's the whole game. A CPM bar chart tells you what should happen if the world cooperates. The Last Planner System is built around the fact that the world doesn't cooperate, and that the people best positioned to say what will actually get done next week are the foremen assigning the crews — not the PM who drew the pretty schedule six months ago.
If you've ever stood in a Monday meeting listening to a super read off a three-week look-ahead that everybody in the room privately knows is fiction, you already understand the problem this system solves. Below is how it actually works on a jobsite, what the moving parts are for, and where teams get it wrong.
The five levels, and why the horizon shrinks as certainty grows
The Last Planner System nests five planning horizons inside each other. The trick — and the whole point — is that the level of detail increases as the time horizon shrinks, because certainty increases as you get closer to the work.
- Master schedule. Milestones. Substantial completion, TCO, the dry-in date, when the tower crane comes down. This is strategy, not sequence. Nobody should be pulling durations for individual drywall zones off the master.
- Phase (pull) planning. You take a milestone — say, "level 3 ready for finishes" — and the trades work backward from it, each one declaring what has to be true for them to hand off to the next. This is where the real sequence gets built, by the people who own the work.
- Look-ahead planning. A rolling 3-to-6-week window where you take the phase plan and screen every activity for constraints. This is the make-ready engine. Its only job is to arrive at the week with clean, workable tasks.
- Weekly work planning. The commitments. What each crew promises to complete in the next seven days, and only tasks that survived the constraint screen are eligible.
- Learning. Every week you measure what you promised against what you delivered, and — this is the part everyone skips — you find out why the misses happened.
Most crews that "try Last Planner" and quit only ever implemented the fourth level. They started tracking commitments without doing the make-ready work that makes commitments keepable, then wondered why their numbers stayed bad. The look-ahead is not optional connective tissue. It's the part that does the work.
Pull planning: build the sequence backward, in the room, with tape on the wall
Pull planning is a session, not a document. You get the trades whose work overlaps into one room — GC, MEP, framer, drywall, whoever owns the phase — and you start at the finish line. The last trade in the sequence posts a card: "I need X done before I can start." The trade responsible for X posts their card and their duration and their predecessor. You walk it backward until you hit the day you're standing on.
What you learn in that room you will not learn any other way. The framer says he needs two days between top-out and the drywall hanger starting, and the hanger says he was told he'd get the wall the same afternoon. That collision — invisible in a Gantt chart because somebody's software auto-linked the tasks finish-to-start with zero lag — surfaces in about ninety seconds when two humans are staring at the same wall of cards.
A few things that make pull planning actually work:
- Durations come from the trade, not the scheduler. If the electrician says rough-in on that floor is four days, it's four days. The moment you "adjust" his number to fit the milestone, you've turned his commitment back into your assumption.
- Capture the handoff conditions, not just the bars. "Drywall starts" is useless. "Drywall starts when framing inspection passes, MEP rough-in is signed off, and the floor is broom-clean" is a handoff you can actually verify.
- Digitize it after, not instead. The sticky-note wall is the thinking tool. Once the sequence is agreed, it needs to live somewhere every trade can see it and update it — that's exactly the job a look-ahead tool like LookAheadWall does well, turning the wall into a shared, location-based plan the subs can actually pull up in the field. But the software captures the decision; it doesn't make it.
The make-ready process: screen for constraints or the promise is a lie
Here's the mechanism that separates Last Planner from wishful thinking. Every activity in the look-ahead window gets run through a constraint screen before it's ever allowed onto a weekly work plan. The seven usual suspects:
- Predecessor work — is the prior activity actually complete, or just "mostly"?
- Materials — on site, or "the PO went out"? Those are not the same thing.
- Information — RFI answered, submittal approved, latest drawing revision in hand?
- Labor — the right crew, the right size, actually assigned?
- Equipment — lift booked, hoist time reserved, pump scheduled?
- Space — is the area clear, or is another trade still working in it?
- External conditions — permit, inspection, weather window.
Each open constraint gets an owner and a date — a name and a day, not a shrug. "Marcus, mechanical submittal approved by Thursday." That's a made-ready action, and its due date has to land before the week the activity is supposed to start, with buffer. If a constraint is still open when the weekly planning meeting comes, the activity does not go on the plan. It goes into the workable backlog — the pile of tasks that are clean and ready — and something from that pile gets pulled forward to fill the gap.
That backlog is the secret weapon nobody talks about. When a crew's planned work falls through because a constraint blew up, an experienced super doesn't send them home — he reaches into the backlog for the next ready task. Keeping a stocked backlog is what lets you protect production against the inevitable surprise.
Reliable promises: a commitment is a specific, conditional, resourced thing
The weekly work plan is nothing but a set of promises, and a promise only counts if it's reliable. "We'll get to the electrical rough-in" is not a promise. "We will complete electrical rough-in in zones A and B by end of day Thursday" is a promise, because you can hold it up on Friday and say plainly whether it happened.
A promise you can actually rely on has four things behind it: a scope both sides understand the same way; the crew and materials genuinely available to do it; every prerequisite already satisfied; and a foreman who honestly intends to deliver it rather than nodding to end the meeting. When a foreman says "I can't promise that yet, the deck's not poured," that is not a failure — that is the system working exactly as designed. The whole point is to surface the "no" on Monday instead of discovering it on Friday.
PPC: the one number, and how not to fool yourself with it
Percent Plan Complete is the signature metric and it's dead simple. Of the tasks you promised this week, what fraction did you fully complete? Ten promised, eight done, PPC is 80 percent. Partial credit doesn't exist — a task 90 percent done is a zero, because the trade waiting behind it can't start on 90 percent.
New teams tend to land somewhere in the 50s. That number feels bad until you realize the traditional schedule was hiding the same reality behind rolled-up bars; Last Planner just made it visible. A mature crew running the system well lives in the 80 to 85 percent range. And here's the part that trips people up: 100 percent PPC is a warning sign, not a trophy. If you hit every promise every week, you're sandbagging — under-committing so the number looks good. The value isn't in the score. It's in the misses, because the misses are where the learning is.
That's the trap the moment PPC becomes a stick to beat crews with. Foremen aren't dumb; if they get punished for a low number, they'll promise less next week and the number climbs while actual production flatlines. PPC is a diagnostic for the planning process, not a performance grade for the crew. Say that out loud, repeatedly, or you'll train your best people to lie to you.
Variance analysis: the misses are the whole point
Every unkept promise gets a reason code. Over a few weeks the codes stop being anecdotes and start being a pattern, and the pattern tells you exactly what to go fix. Common categories:
- Prerequisite not complete — the handoff before yours slipped.
- Materials — not on site, wrong material, or staged where the crew couldn't reach it.
- Information / design — RFI, changed direction, wrong revision in the field.
- Labor — crew pulled to another job or short-handed.
- Equipment — lift double-booked, hoist backed up.
- Weather / external — the honest one, and usually the smallest bucket.
- Over-commitment — the crew simply promised more than a week could hold.
Read the tallies. If a third of your misses are "materials," your problem isn't the field, it's procurement and staging — fix the make-ready lead time on long-lead items. If "prerequisite not complete" dominates, your look-ahead screening is too shallow; you're letting activities onto the plan whose predecessors were never truly clean. If "over-commitment" keeps showing up for one foreman, that's a coaching conversation about honest promising, not a discipline problem. The reason codes turn a vague sense that "the job's behind" into a specific, actionable list of what to change.
Where the software actually earns its keep
You can run the Last Planner System with painter's tape, index cards, and a spreadsheet — plenty of great jobs have. The reason to reach for a tool is that the manual version falls apart the moment your team is distributed, your subs need the plan on their phones, or you want to trend PPC and variance across months instead of retyping it every Friday.
Good look-ahead scheduling software should do a handful of things and stay out of the way otherwise: give you a visual, location-based board that mirrors the pull-plan wall; let you attach constraints to activities with an owner and a due date; make the weekly commitments and their status visible to every trade, including in the field; and calculate PPC and variance for you so the learning conversation is about causes, not arithmetic. That's the lane LookAheadWall sits in — the shared weekly work plan, the trade-flow sequences from the pull session, and the crew-leader mobile view so the foreman is looking at the same plan on his phone that you're looking at in the trailer. The software is worth having exactly to the degree that it makes the disciplined behavior easier than skipping it.
What it can't do is make the promises for you. No app screens a constraint — a person does. No dashboard decides that framing really is ready for the hanger. The tool captures and surfaces and trends; the judgment still comes from the people doing the work, which was the whole premise to begin with.
How to start without blowing it up
Don't roll out all five levels on Monday. The teams that make it stick start narrow and add layers as the habit forms:
- Start with weekly commitments and PPC. Just get foremen making specific promises and honestly scoring them. Two or three weeks of this and the conversations in the trailer already change.
- Add the look-ahead constraint screen. Once weekly planning is a habit, push the horizon out three to six weeks and start screening for constraints. This is where PPC finally starts climbing, because now you're clearing obstacles before they hit the week.
- Add pull planning at the next phase. When a new phase or a new project comes up, run a real backward-pass session with the trades. By now the crew trusts the shorter-horizon planning enough to invest the time.
The technology is the easy part; the culture is the hard part. You're asking foremen to make public promises and admit publicly when they miss — that takes trust, and trust takes a few weeks of nobody getting hammered for an honest low number. Protect that, keep the workable backlog stocked, read the variance codes like they matter, and the schedule slowly stops being a wish and starts being something you can actually run a jobsite on. That's the entire promise of the Last Planner System, and it's a real one.