The Last Planner System: How to Run It on Site, Week by Week

Key takeaways

• The Question: How does the Last Planner System work in practice, from the master plan down to the daily huddle, and how do you know its measures are telling the truth?
• The Answer: The Last Planner System moves planning down through five horizons: milestones, pull-planned phases, a six-week lookahead where constraints are removed, a weekly plan of tasks that are genuinely ready, and a learning loop built on percent plan complete and reasons for variance. It works when each commitment is checked against what was physically completed.
• The YelloSKYE Angle: PPC is only as honest as the definition of "done". Construct IQ lets the team check committed tasks against dated 360°, drone and fixed-camera imagery before the weekly score is recorded.

The Last Planner System (LPS) is a collaborative planning and control method in which the people who direct the work, typically trade foremen, plan it and commit to it, once the obstacles to that work have been removed. It sits under the master schedule and turns milestones into reliable weekly promises through pull planning, a rolling lookahead, a constraint log, a weekly work plan and daily check-ins, with percent plan complete (PPC) used to learn why promises were missed. Verifying completion with evidence from Construct IQ keeps that learning loop accurate.

Our lean construction guide introduces LPS as one of several lean practices. This guide goes into the detail a project team needs to run it: what happens at each level, how to keep a constraint log, what makes a good weekly assignment, how to calculate and read PPC, and how to check that "complete" really means complete.

The idea behind it: should, can, will, did

The system was created in the 1990s by Glenn Ballard and Greg Howell, who had observed that conventional site plans described what should happen, while crews did whatever could happen given the materials, information and access available that day. The gap between the two was rarely recorded, so nobody learned from it.

LPS separates four questions and handles each one on purpose:

The discipline is that "will" may only be chosen from "can". A foreman should never commit to work that still has an open constraint.

Master plan and milestones

The master plan sets the contract dates, the main phases and the key milestones: structure complete, building weathertight, permanent power on, each floor ready for fit-out. It is the logic-linked programme the project manager already maintains. LPS does not replace it; it gives the master plan a reliable way to be delivered.

The master plan should be detailed enough to define phases and milestones clearly, but not so detailed that it dictates daily sequences the trades have not agreed.

Phase planning by pulling from a milestone

For each phase, the trades involved meet to build the plan backwards from its milestone. The session works best with some preparation.

Before the session

• Define the milestone precisely, including what "complete" means for it and who accepts it.
• Invite the foremen who will run the work, not only commercial managers.
• Bring the current master plan dates, design status and known long-lead items.

During the session

The team starts with the milestone at the right-hand end of a wall or digital board. Each trade asks what it must receive from others before it can do its own work, and places those hand-offs to the left. Each card names the task, the trade, the duration agreed by that trade and what the next trade needs from it. When the chain reaches the present, the team checks whether the dates fit the time available.

If they do not, the team negotiates: overlapping work, splitting zones, changing the sequence or adding crews. Any spare time is shown openly as a phase buffer before the milestone, rather than left hidden in individual durations.

After the session

The planner enters the agreed sequence into the programme so the phase plan and the master plan say the same thing, and the phase tasks begin feeding the lookahead.

The six-week lookahead and making work ready

Each week, the lookahead rolls forward one week. Every task entering the window is screened for constraints, and someone is made responsible for removing each one before the task reaches the weekly plan.

Tasks whose constraints are all cleared move into a workable backlog: work that could be done if something else stalls. A healthy backlog gives crews somewhere productive to go when a planned task falls through, instead of standing idle or starting work out of sequence.

The constraint log

The constraint log is a simple register, but it is the engine of the system. Each row needs:

• the task it blocks and its planned start date,
• a plain description of the constraint,
• one named owner, not a company,
• a promised removal date,
• status: open, cleared or escalated.

Review it every week in the same meeting. A constraint whose removal date has passed is escalated, and if it cannot be cleared before the task's start, the task stays out of the weekly plan. Over time the log shows which kinds of constraint keep recurring, often design information and material deliveries, and where the project needs earlier attention.

The weekly work plan

In the weekly planning meeting, each last planner chooses next week's tasks from those that are free of constraints and commits to completing them. Ballard's research set out quality criteria for a good assignment, which can be summarised as:

• Defined: specific enough that anyone can tell when it is done, with location and quantity.
• Sound: all constraints removed, materials on site, information issued, prerequisite work complete.
• Sequenced: chosen in the order the phase plan requires, not by convenience.
• Sized: matched to the crew's realistic capacity for the week.
• Learning: missed tasks are tracked so the reasons can be studied.

The commitment is personal: the foreman is saying "my crew will do this". That is why it must be voluntary and why the plan should be built from tasks the foreman considers achievable.

The daily huddle

A short stand-up each morning, usually ten to fifteen minutes at the work face or a board near it, keeps the weekly plan on track. Each foreman says what was done yesterday, what is planned today and what might stop it. Problems are noted and owners named, with the discussion taken offline. The huddle is not a progress meeting; its job is to catch a blocked task the day it becomes blocked.

Measuring PPC and learning from variance

At the end of the week, each committed task is marked complete or not complete. Percent plan complete is:

PPC = tasks completed as committed ÷ tasks committed × 100

A task is complete only if all of it is finished; there is no partial credit. Here is an example week.

The figure is useful, but the reasons column is the point. Group reasons by category every week and plot them over time. If prerequisite work keeps appearing, the phase plan's hand-offs need attention; if materials dominate, the lookahead is not screening deliveries early enough. Two supporting measures help: tasks anticipated, the share of weekly tasks that were in the lookahead three weeks earlier, and tasks made ready, the share of lookahead tasks whose constraints were actually cleared on time.

Remember what PPC does not measure. A team can score 100 per cent while the project falls behind, simply by committing to very little. PPC measures how reliably promises are kept; physical progress against the programme needs a separate measure, explained in our guide to percent complete in construction.

Checking completion with site evidence

The weakest point in the learning loop is the word "complete". In most weekly meetings, tasks are marked done because the foreman says so. That is quick, but it invites two problems: tasks marked complete with a few rooms or a small area left, and tasks marked complete that the next trade later finds unready. Both inflate PPC and hide exactly the constraints the system is meant to expose.

A more reliable approach checks commitments against a dated record before the weekly score is recorded.

• Write the definition of done into the task. "Level 3 east corridor ceiling grid installed and ready for tiles" can be checked; "work on ceilings" cannot.
• Capture the work zones on a fixed day. A 360° walk of each floor at the end of the week, or fixed-camera views of priority zones, gives a consistent view of every committed task.
• Review disputed items, not every item. The team looks only at tasks where the evidence and the claim disagree.
• Feed confirmed status into the programme. The same evidence that settles PPC can update the master schedule.

Construct IQ supports each of these steps. Any site worker can carry out a 360° walk without a specialist technician, and the record is organised by location on the DEX platform. In the interior fit-out use case, teams resolved snags five times faster and cut coordination time by 70 per cent once they worked from that record. Fixed cameras give a constant timestamped view of chosen zones, and our fixed-camera monitoring work has reduced physical site visits by as much as 80 per cent. DEX's Time Travel Playback lets the weekly meeting step back to any day in the past week to see when a task actually finished.

Common reasons the system fades

Getting started: a practical checklist

• Pick one phase with a clear milestone and run a pull planning session with the foremen who will do the work.
• Set up a six-week lookahead and a constraint log with one named owner per constraint.
• Agree written criteria for a sound assignment and a definition of done for common task types.
• Hold the weekly planning meeting at the same time every week and limit it to commitments, constraints and learning.
• Run a short daily huddle at the work face.
• Calculate PPC weekly with no partial credit, and record a reason category for every missed task.
• Verify committed tasks against dated site evidence before recording the score.
• Review the top recurring reasons each month and change the lookahead or phase plan to remove them.