Chapter 27 — Lean Construction and the Last Planner System: Pull Planning, Flow, and Making Work Ready

The Hook: Fifty-Four Percent

Friday, February 6 of Year 2. Month twelve on the Northgate Outpatient Pavilion. Wei Chen had run the monthly schedule update the day before and the answer was on the whiteboard in the trailer in Wei's tidy handwriting:

Substantial completion forecast: September 21, Year 2. Contract date September 18. Variance: −3 CD.

Three calendar days behind. CD means calendar days; WD, when it shows up later, means work days. Three days on a 565-day job is nothing. Three days is a rainy week and a half. Ray looked at it and felt fine.

Margo Deacon looked at it and said, "That's wrong."

"It's not wrong," Wei said. "That's what the network calculates. I updated 340 activities against actual starts and finishes. The longest path runs through interior rough-in on Level 3 and it comes out three days late."

"I believe your arithmetic. I'm telling you the number is wrong." Margo had her hard hat under her arm and mud to the knee. "I walked the building this morning. Cardinal had six sheet-metal guys on Level 3 standing around at seven-forty because Halcyon hadn't pulled their tray out of the corridor. Apex had a hanger crew show up Wednesday for a zone that wasn't inspected. Vantage has been telling me for two weeks they're going to finish the Zone A main and they have not finished the Zone A main. Every one of those is a day somebody didn't work. Your schedule doesn't see any of it, because your schedule updates once a month and the job breaks every morning."

"Then give me the data and I'll model it."

"That's the problem," Margo said. "I don't have data. I have twenty-nine years of standing in that building knowing it's worse than three days, and no way to prove it to you or to Pri Sethi or to anybody else. It's a feeling. Feelings lose arguments."

Ray had been in this argument before, from both sides. What he said next was the only genuinely good idea he had that month.

"So let's measure it. Not the schedule — the promises. Every Monday, every trade foreman writes down what they will finish that week. Not what they'll work on. What they will finish. Then every Friday we go down the list and mark each one done or not done. Binary. No partial credit, no 'we're basically there.' And when it's not done, the foreman tells us why, and we write the why down."

"They'll hate it," Margo said.

"For about three weeks."

"What do we change while we're doing it?"

"Nothing," Ray said. "That's the point. If we start fixing things we'll never know what the baseline was. Six weeks. We measure and we change nothing."

They ran it for six weeks — the weeks of February 9 through March 16, Year 2. Fourteen trade foremen. 187 commitments made. 101 completed.

Percent plan complete: 54%.

Just over half. Not half of what a scheduler in an office guessed. Half of what the people who do the work — who know the crew, the material, the sequence, and who wanted very badly to be right in front of thirteen other foremen — said out loud on Monday that they would finish by Friday.

Then Dani Okonkwo sorted the 86 misses by reason. This is the table. Everything else in this chapter is commentary on it.

(Percentages rounded; they sum to 100% within rounding.)

Look at what is not on that list. Nowhere in 86 misses does anybody say "my crew was slow." Nobody says "we don't know how to do the work." Nobody says "the drawings are impossible."

Fifty-five percent of the misses are the top three lines — the work in front of us wasn't done, the material wasn't there, or nobody answered the question. Every one of those is knowable in advance. Every one of those is somebody's job in an office. And none of them appear anywhere on a critical path method (CPM) schedule, because a CPM network assumes that the moment a predecessor finishes, the successor is ready to go.

That assumption is false. On a real job it is false more often than it is true.

🏃 Fast Track: If you have run pull plans and weekly work plans before, skim §27.1 and §27.2, then read §27.5 (constraint screening — the part most teams skip), §27.6 (why the binary rule and the reason codes matter more than the score), and §27.9 (where it fails) carefully. Those three are where experienced teams still lose the system.

🔬 Deep Dive: The CPM machinery this chapter builds on is Chapter 14 and Appendix B. Look-ahead forms, constraint logs, and weekly-work-plan templates are in Appendix D. What you do with schedule variance once you have measured it is Chapter 29. Model-based sequencing that feeds pull planning is Chapter 35.

27.1 Why the CPM Schedule Is Not Enough — and Why That Is Not an Attack on Chapter 14

Let me be careful here, because this section is where lean construction is usually taught badly.

You will hear people say the critical path method is "obsolete," or "an office fantasy," or "a document produced for lawyers." That is nonsense, and if you believe it you will lose a delay claim and deserve to. Everything in Chapter 14 is true and necessary. CPM is genuinely excellent at four things:

1. Proving the work fits in the time. You cannot bid a 565-day job without a network that demonstrates 565 days is achievable.
2. Finding the longest path. The critical path is a calculated result, not a management opinion — you cannot eyeball it on a job with 340 activities.
3. Analyzing delay after the fact. Entitlement, causation, and concurrency arguments (Chapter 33) live or die on a defensible network.
4. Communicating with an owner and a lender. Pri Sethi's board does not want sticky notes. It wants a milestone chart and a variance number.

Now the honest part. CPM is poor at one specific thing, and it happens to be the thing a foreman needs on Tuesday morning.

Here is the sentence to carry out of this section:

CPM answers "does the work fit in the time?" It does not answer "is this specific work ready to be done by this specific crew on Tuesday?" Those are different questions and they need different tools.

They are complements, not competitors. On Northgate, Wei Chen's P6 network and Margo Deacon's weekly work plan sat side by side on the same trailer wall for the last six months of the job, and neither one replaced the other. The network told them where the longest path ran and what a slip was worth. The weekly plan told them what to do about it before Friday.

🔄 Check your understanding. Wei Chen's February update said the job was three calendar days behind. Margo believed it was worse. Both of them were looking at the same building. What, precisely, was the CPM model unable to see?

27.2 A Short, Honest Grounding in Lean Thinking

You need enough theory that the practices make sense, and no more. Here is enough.

Where it came from

"Lean" is a name that Western researchers gave, around 1990, to the production system that Toyota Motor Corporation developed after the Second World War — the Toyota Production System (TPS), most closely associated with the engineer Taiichi Ohno. The term "lean production" was popularized by researchers at MIT's International Motor Vehicle Program, whose 1990 book The Machine That Changed the World (Womack, Jones, and Roos) put it in front of a general audience.

Construction picked it up in the early 1990s. The usual academic starting point is Lauri Koskela's argument that construction management theory recognized only transformation — turning inputs into outputs — and had no concept of flow (how work moves between transformations) or value (what the customer actually gets). Shortly after, Glenn Ballard and Greg Howell developed the Last Planner System (LPS) and went on to co-found the Lean Construction Institute (LCI), a nonprofit that publishes practitioner guidance and training on lean methods in construction. "Last Planner" is a registered trademark of LCI. The research community organizes itself largely through the International Group for Lean Construction (IGLC), whose annual conference proceedings are freely available and are the best place to find out whether a claim about lean is supported or merely repeated.

And now the caveat that most lean training skips

Construction is not a factory. Anyone who tells you otherwise has not built anything. Three differences are structural and permanent:

That third one is the big one, and we will come back to it in §27.9, because it is the reason most lean adoptions fail. In a Toyota plant, the person who removes a constraint and the person who benefits work for the same employer. On your job, they usually do not.

So do not try to import the factory. Import the three ideas that survive the translation: waste, flow, and pull.

Idea one: value and waste

Lean starts by asking what the customer is paying for — value — and calls everything else waste. Ohno's original list had seven categories; an eighth (unused human talent) was added later by others and has become standard in practice. Translated into a construction site, with real Northgate examples:

Notice something about that table. Waste 1 through 7 are things you can see if you walk the building with your eyes open. Waste 8 is invisible, costs the most, and is free to fix.

Idea two: flow, and why variability is the real enemy

This is the one that matters, and it deserves the space.

🧩 Productive struggle. Five trades hand off in a chain: layout, overhead rough, framing, in-wall rough, drywall. Each one is 90% reliable — nine times out of ten, when it says it will hand off Friday, it hands off Friday. That is a good crew. That is better than most.

What is the probability that the fifth trade starts on time? Take three minutes and estimate before you read on. Most people say "around 85%." Write down your number.

🔍 Why this works. Here is the mechanism, and it is the single most useful idea in this chapter.

A construction schedule is a chain of handoffs. Every trade knows the chain is unreliable, because every trade has been burned. So every trade does the rational thing: it protects itself. It quotes a duration longer than it needs. It keeps a second job warm so it has somewhere to send the crew. It shows up late on purpose because it has learned that showing up on the scheduled date means standing around. It hoards its own private float and does not tell you it exists.

Every one of those responses is locally intelligent and globally catastrophic. The buffers do not cancel out — they stack. Each trade's protection consumes calendar. Each trade's late start pushes the next one. The sum of twenty locally optimized subcontractors is a globally slow project, and nobody in it is behaving badly.

Variability, not slowness, is what destroys construction productivity. Your crews are not lazy. Your trades are not incompetent. Your handoffs are unreliable, and unreliability compounds. Fixing the compounding is worth vastly more than pushing anyone to work faster — and unlike pushing people to work faster, it does not get anybody hurt (Chapter 24).

Idea three: push versus pull

Push planning says: the schedule says start Tuesday, so start Tuesday. Work is released into the system on a date calculated by a central authority, whether or not the downstream trade is ready to receive it.

Pull planning says: the successor needs this handoff by Thursday, so we work backward from the handoff and release the work only when the successor can actually take it.

What changes in practice:

Pull does not mean "no schedule." It means the last few weeks before a crew arrives are planned by the people who will be standing there, working backward from what the next crew needs.

27.3 The Last Planner System: Four Levels and a Scoreboard

The Last Planner System is the working method that makes all of the above operational. Its name comes from a precise idea: the last planner is the person who makes the final assignment before work is executed — the trade foreman who tells a crew where to be at 6:30 tomorrow. Everyone above them plans; only the last planner commits.

The system is four nested levels of planning plus a measurement, and the classic shorthand is worth memorizing because it is the whole system in four words:

SHOULD → CAN → WILL → DID

📊 Diagram (described). The system is a funnel. Work enters at the top as a large set of things that should happen according to the master schedule. It is filtered at the look-ahead level down to what can happen — the readiness screen. Only what survives the screen becomes a commitment about what will happen. Then you measure what did happen and feed the reasons back to the top. Rendered as a stack:

  LEVEL 1 · MASTER SCHEDULE            "SHOULD"   Milestones, whole job     Owner: Wei Chen
       │   CPM network (Chapter 14)                Updated monthly
       ▼
  LEVEL 2 · PHASE / PULL PLAN          "SHOULD"   One phase, 6–12 weeks     Owner: the trades
       │   The wall of sticky notes                Rebuilt for each phase
       ▼
  LEVEL 3 · LOOK-AHEAD + MAKE-READY    " CAN "    6 weeks, rolling          Owner: make-ready mgr
       │                                           Updated weekly
       │   ✂  ◄── THE SCREEN. Activities with an open constraint are REMOVED here.
       ▼
  LEVEL 4 · WEEKLY WORK PLAN           " WILL "   1 week                    Owner: each foreman
       │   Commitments, by name, out loud          Made Monday morning
       ▼
  LEVEL 5 · MEASUREMENT                " DID "    PPC + reason codes        Owner: everyone
           PPC, Pareto, corrective action          Scored Friday morning
                    │
                    └──────────► feeds back into Levels 2 and 3

Most teams that "do Last Planner" do Levels 1, 2, and 4. They build a CPM, they run a fun pull-planning session with sticky notes, and they hold a Monday meeting where foremen say what they will do. Then they skip Level 3 and Level 5 — the screen and the scoreboard — because those two are boring, unglamorous, and require someone's actual time.

That is exactly what Kestrel did on Northgate. They ran a beautiful pull plan on January 8. They held the Monday meeting every week. And their percent plan complete was 54%.

Levels 3 and 5 are the system. Levels 2 and 4 are the meeting.

27.4 Level 2 — Pull Planning: The Session, the Wall, and the Sequence Nobody in the Office Thought Of

What a pull-planning session actually is

A phase pull plan is a working session in which the foremen who will build a phase of the project design the sequence of that phase themselves, backward from its end milestone.

Not a presentation. Not a review of the CPM. A working session, with a blank wall, in which the plan does not exist when people walk in and does exist when they walk out.

The setup:

• A phase and its end milestone. One phase — not the whole job. Six to twelve weeks is the useful range. Northgate's was: enclosure and dry-in, ending at the canonical milestone building dried-in, March 28, Year 2.
• The people who will do the work. Foremen and superintendents, not project managers reading from a laptop. If a company sends its PM instead of its foreman, the plan that comes out is a guess.
• A wall. Butcher paper, gridded into weeks. Sticky notes, one color per trade. Fat markers so writing is legible from ten feet.
• A facilitator who is not the boss. On Northgate this was Dani Okonkwo, deliberately — a field engineer, junior, with no authority to overrule anyone. A facilitator's job is to enforce the process, not to decide the sequence. When the general superintendent facilitates, the plan is the general superintendent's plan with sticky notes on it.
• Three to four hours. It is not a lunch meeting.

The rules, posted on the wall:

1. Write on a note only what your own crew will do. You do not plan other people's work.
2. Every note has: trade, deliverable, duration in work days, and the crew size.
3. Every note states what you need from the trade before you in order to start. That sentence is the whole method.
4. Durations are stated by the person who will perform the work. Nobody negotiates your duration down. They can ask what would have to change to make it shorter.
5. Build right to left — from the milestone backward — the first time through. Then walk it left to right to check it.
6. Disagreements are resolved at the wall, by the two trades involved, not by escalation.

The Northgate enclosure pull plan

Thursday, January 8, Year 2. Eleven people in the trailer conference room. The milestone note went up on the far right first:

▶ BUILDING DRIED-IN — March 28, Year 2 Definition: roof watertight and flood-tested; all wall openings permanently closed; all penetrations flashed and sealed; field water test passed at curtain wall; temporary enclosure removed where permanent enclosure is complete.

That definition took twenty minutes to write and was the most valuable twenty minutes of the session. Before it, "dried-in" meant six different things to six different trades. Delia Krupa (Summit Roofing) thought it meant membrane down. Renny Kovach (Solstice Glazing Systems) thought it meant glass in. Aaron Vogt (Coastline Waterproofing) knew it meant sealed, which is a different date entirely and about two weeks later than either of the other two thought.

💡 Aha moment. Half the disagreements on a construction site are not disagreements about sequence. They are disagreements about what "done" means, discovered three weeks after everyone assumed they agreed. A pull plan surfaces those in the first hour, in a room, for free — instead of in the field, at the handoff, for money.

Then they built backward. Here is the wall.

   PULL PLAN WALL — NORTHGATE ENCLOSURE / DRY-IN PHASE
   Session: Thursday, January 8, Year 2   ·   6 trades, 11 people   ·   Facilitator: D. Okonkwo
   Built RIGHT TO LEFT from the milestone. Shown here left to right for reading.
   Each █ block = work in that week.  5 WD per week.

   TRADE            │ W1  │ W2  │ W3  │ W4  │ W5  │ W6  │ W7  │ W8  │ W9  │W10  │W11  │
   MONDAY OF        │Jan12│Jan19│Jan26│Feb 2│Feb 9│Feb16│Feb23│Mar 2│Mar 9│Mar16│Mar23│ ▶ DRY-IN
  ══════════════════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪══ Mar 28
   Cardinal (mech)  │█████│███  │     │     │     │     │     │     │     │     │     │  RC-1 roof curbs set
   Halcyon (elec)   │  ███│███  │     │     │     │     │     │     │     │     │     │  RP-1 roof penetrations
   Granite Ridge    │█████│█████│█████│███  │     │     │     │     │     │     │     │  PC-1 panels set, N+E
     (precast)      │     │     │     │  ███│█████│     │     │     │     │     │     │  PC-2 joints caulked
   Solstice         │     │  ███│█████│█████│█████│███  │     │     │     │     │     │  CW-1 units hung, S+W
     (curtain wall) │     │     │     │     │     │  ███│█████│     │     │     │     │  CW-2 glaze + wet seal
   Summit (roofing) │     │     │█████│█████│███  │     │     │     │     │     │     │  RF-1 insul + membrane
                    │     │     │     │     │  ███│███  │     │     │     │     │     │  RF-2 edge metal, curbs
   Kestrel carp.    │     │     │     │     │  ███│█████│█████│█████│     │     │     │  BW-1 backup wall + AVB
   Coastline (wp)   │     │     │     │     │     │     │  ███│█████│███  │     │     │  WP-1 corner interfaces
                    │     │     │     │     │     │     │     │     │  ███│█████│     │  WP-2 penetration seals
   ALL              │     │     │     │     │     │     │     │     │     │  ███│█████│  DI-1 water test + punch

And here is what the foremen actually wrote on the notes — the part that never survives into a Gantt chart:

The moment the plan got shorter

Ninety minutes in, the wall was a mess and the finish date was landing in April.

The office CPM — Wei Chen's December update — had precast and curtain wall alternating by floor: both trades on all four elevations, floor by floor, sharing the crane. Forecast dry-in: April 8, Year 2, eleven calendar days past the contract milestone.

Then Tobias Ferrell and Renny Kovach, who had been arguing about the crane for twenty minutes, arrived at the same sentence from opposite directions.

Ferrell: "I don't need the crane all day. I need it from six to two. What I actually need is a face. Give me the north and the east and don't put anybody else on them and I'll set five panels a day and never stop."

Kovach: "That's my problem too. I'm unitized. My crew hangs six to eight stacks a day and a stack is four floors of one bay. I can't leapfrog around his panels. Every time I demobilize off an elevation and come back it costs me a day and a half. Give me the south and the west, whole, top to bottom, and I don't need his crane at all — I hang off davits on the slab edge."

Margo Deacon: "So you two are telling me the schedule split you by floor and you both want to be split by elevation."

Ferrell: "Yeah."

Kovach: "Yeah."

Margo: "Nobody ever asked you."

Kovach: "Nobody ever asked me."

They rebuilt the wall on the elevation split. Granite Ridge took north and east continuously; Solstice took south and west continuously; they never shared a face and never shared a crane hour. That change alone took 6 work days out.

Then Delia Krupa took two more. The CPM had roofing in two phases, split around MEP curb setting, with a four-day gap while Cardinal came back for the second batch of curbs. Krupa's offer: "Give me the whole deck once — every curb, every sleeve, every penetration finished and flashed before I bring a roll up — and I'll do 34,000 square feet in thirteen days instead of your fifteen plus the four-day gap. But if anybody drills my membrane after I roll it, you own the leak for twenty years." Cardinal agreed to set all 34 curbs in one 8-day push in weeks 1 and 2. Net: 2 more work days, and one fewer roof mobilization.

Total: 8 work days out of the sequence. Forecast dry-in moved from April 8 to March 27, Year 2 — one day inside the contract milestone.

🏗️ From the field. I have facilitated maybe forty pull-planning sessions and roughly two-thirds of them produce a plan shorter than the office schedule. Not because the office schedulers are bad — Wei Chen is excellent — but because a scheduler builds logic from drawings and a foreman builds it from having personally lost a day and a half demobilizing off an elevation. That knowledge exists only in the field, it is free, and the only way to get it is to put those people in a room and ask.

The corollary is uncomfortable: for every pull plan that finds eight days, there is a job somewhere that never had the session and is currently eight days late for no reason at all.

⚠️ Safety alert. Notice what Krupa's offer eliminated: a second roof mobilization, which means a second set of workers on a 34,000 SF roof at height, a second temporary edge-protection setup and teardown, and a crew drilling into a finished membrane with other trades already working under it. Sequence changes that remove mobilizations remove exposures. A plan with fewer handoffs is not just faster — it puts fewer people at height fewer times.

🔄 Check your understanding. Renny Kovach's bid duration for the curtain wall was 26 work days. In the pull-planning session he committed to 21. He did not get more people and he did not get more equipment. Where did the five days go?

27.5 Level 3 — The Six-Week Look-Ahead and the Constraint Log

This is the center of the chapter.

🚪 Threshold concept.

The schedule tells you what should happen; the constraint log tells you what can.

A CPM network assumes that when a predecessor finishes, its successor is ready. On a real job that assumption is false most of the time. Work is only ready when every constraint on it has been removed — and making work ready is a separate discipline from scheduling it.

Before you understand this, a look-ahead schedule is a zoomed-in slice of the CPM. You produce it by filtering the network to the next six weeks and printing it. It answers the question what is scheduled to happen soon? Its purpose is to inform people. When work does not happen, you conclude that the schedule was wrong, and you update the schedule.

After you understand this, a look-ahead is not a report — it is a screen, and it runs in the opposite direction. You start with what is scheduled, then you interrogate each activity against a fixed list of constraints, and activities that fail the interrogation are removed from the plan. Its purpose is not to inform. Its purpose is to produce a list of work that is genuinely ready and a second list of things somebody has to go fix. When work does not happen, you conclude that a constraint was missed, and you fix the constraint-removal process.

The shift is from predicting to preparing. You stop asking "when will this happen?" and start asking "what has to be true before this can happen, who is going to make it true, and by when?"

The eight constraint categories

Every activity, everywhere, on every project, is gated by some combination of exactly these eight things. Learn them as a list you can recite, because you will run through them dozens of times a week.

Compare that list to the reason-code Pareto in the hook. They are the same list. Nine of the ten Northgate reason codes map one-to-one onto a constraint category; the tenth is weather. That is not a coincidence — it is the design. You categorize the reasons for failure using the same taxonomy you use to screen for readiness, so that every miss tells you which part of your screen leaked.

The constraint log

The log is a table. It is not sophisticated. Its power is entirely in the fact that it exists and that somebody looks at it every week.

And then the rule that makes the whole thing work:

An activity does not enter the weekly work plan until every constraint on it has been removed.

Not "is likely to be removed." Not "the architect said probably Thursday." Removed. The RFI is answered and in the field. The material is on the floor. The inspection is scheduled with a confirmation number.

🔍 Why this works. The screen looks like it removes work from your plan, and in the short run it does. That feels like going backward. Here is what it actually does.

When you put unready work into a weekly plan, three things happen and all of them are expensive. First, a crew mobilizes to an area, discovers it can't work, and demobilizes — that is roughly a half-day of eight people, gone, and it does not show up on any report. Second, the crew that got displaced now needs somewhere to go, so it gets pushed into an area where another trade is working, and you have manufactured trade stacking (see the Spaced Review at the end of this chapter, and Chapter 19). Third — and this is the one that ruins the system — the foreman learns that the plan is fiction. Once a foreman believes the weekly plan is decorative, they stop planning against it, they stop committing honestly, and every future number you collect is garbage.

Screening protects the credibility of the plan, and credibility is the asset the whole system runs on. A short plan that is true beats a long plan that is aspirational, every single week.

📋 Try it — Build the Constraint Log and Screen the Look-Ahead

It is Friday, April 3, Year 2. You are the field engineer on Northgate. Your job this afternoon is to produce next week's weekly work plan for Level 3, Zone B — the east clinic pod, about 7,400 SF — and to build the constraint log for the six weeks ahead.

Here is your six-week look-ahead as it comes out of the CPM. Twelve activities.

And here is the state of the project, as of this afternoon:

1. RFI 214 — corridor 3-C ceiling height; the duct main elevation conflicts with a structural brace. Issued March 18. No answer from Halvorsen + Pike. The main runs the full length of Zone B, so its elevation gates the whole run.
2. RFI 219 — nurse-call device locations at exam-room casework. Issued March 27. No answer.
3. RFI 221 — extent of lead lining above the ceiling at Imaging 3-14: slab-to-slab, or 6 inches above ceiling? Issued April 1. No answer.
4. Submittal 08-1200-C, hollow metal frames for Zone B, was returned "Revise and Resubmit" on March 24. The supplier has not sent the resubmittal. Frames must be set during partition framing.
5. Material: lead-lined gypsum and lead-lined frames for Imaging 3-14. Purchase order placed. The vendor confirms delivery May 6, Year 2.
6. Inspection: the Zone B in-wall rough inspection is not scheduled. Rivermont requires three business days' notice, and will not schedule until the mechanical, electrical, plumbing, and fire-protection contractors have each signed the pre-inspection checklist.
7. Preceding trade: the Zone A sprinkler main — which the Zone B branch ties into at the corridor 3-C header — is 60% complete. Vantage has one four-person crew splitting between Zone A and Level 2.
8. Permit: the medical gas installation permit is applied for but not issued. The AHJ is waiting on Cardinal's brazer certifications.
9. Labor: Marisol Duarte says Apex has 6 hangers available the week of April 27 and 10 the week of May 4.
10. Equipment: three scissor lifts are on Level 3. One is committed to Vantage in Zone A.
11. Studs, track, and 4,200 LF of Zone B framing material were delivered March 30 and are staged in Zone B. Cable tray and branch conduit for Zone B were delivered March 25.

Your tasks:

1. Build the constraint log for the six weeks.
2. Screen the week-one candidates (A, B, C, D, E) into ready and not ready.
3. Produce the weekly work plan for the week of April 6.
4. Identify the one finding in this look-ahead that is not a next-week problem at all — and say what you do about it today.

Do the work before you open the answer. Give it twenty minutes and a pencil.

27.6 Level 4 and Level 5 — Commitments, PPC, and the Reasons That Actually Improve Things

The weekly work plan

A weekly work plan is a list of commitments. A commitment has four properties, and if any one of them is missing you do not have a commitment, you have a topic:

1. A named person who controls the crew makes it. Not a company. Not a PM relaying it.
2. A specific, finishable deliverable — "top track complete in Zone B, all 47 partitions," not "work in Zone B."
3. A date, which is this Friday.
4. It is made out loud, in front of the other trades, who are relying on it.

That fourth one does most of the work. A promise made to a project manager is an administrative item. A promise made to the four foremen whose work depends on it is a promise to peers, and people keep those. This is the same mechanism as the action-item discipline in Chapter 26, applied to production instead of to administration.

The meeting is short — 45 to 60 minutes for a job the size of Northgate — and it has exactly three parts: score last week, remove constraints for the next six weeks, commit for this week. In that order. Scoring first is deliberate, and we are about to see why.

Percent plan complete

PPC = commitments completed ÷ commitments made, expressed as a percentage.

Northgate's six-week baseline: 101 ÷ 187 = 54%.

Three rules govern the measurement, and all three matter:

Rule 1 — It is binary. No partial credit, ever. A commitment is complete or it is not. "We got 90% of it" scores zero.

🔍 Why this works. This rule offends everybody the first time they hear it, so here is the mechanism. The purpose of PPC is not to measure production — you already measure production, in the cost report (Chapter 28) and in earned value (Chapter 30). The purpose of PPC is to measure whether the plan can be relied on by the next trade. And reliability is binary from the receiver's point of view. If the framing crew is 90% done in Zone B, the in-wall rough crew cannot start. Ninety percent of a handoff is zero handoffs. The moment you allow partial credit, the score stops measuring what the next trade can count on and starts measuring effort — and effort is exactly the thing everyone already believes about themselves.

Rule 2 — The last planner scores their own commitment, and the score is not a performance review. The foreman says done or not done. Nobody argues. If a foreman feels the score is a disciplinary instrument, they will commit to less and less until the number is meaningless. See Case Study 2 for what that looks like.

Rule 3 — Track commitments made, not just PPC. This is the one almost everybody forgets and it is the single easiest way to catch a gamed system. PPC of 90% on nine commitments is much worse than PPC of 72% on forty. If your PPC is rising while your commitment count is falling, your team is not improving — it is learning to promise less.

What a normal PPC looks like. Practitioners commonly report first-measurement figures somewhere in the 40s to 60s, and teams that sustain the discipline commonly report the high 70s to high 80s. Treat those as practitioner folklore rather than research findings: PPC is not measured consistently enough across projects, companies, or trades to compare one job's number to another's. The only PPC number that means anything is your own, trended over time. A team that goes from 54% to 78% has learned something real. A team that reports 85% in month one has usually just defined "commitment" loosely.

And note that 100% is a warning sign, not a triumph. A team hitting 100% is committing only to what is certain, which means it is planning below its capacity and leaving production on the table.

Reasons for variance — the part that actually creates improvement

PPC is a thermometer. The reason codes are the diagnosis.

A thermometer tells you that you have a fever. It does not tell you what to do. Every serious team that has run this system will tell you the same thing: the score creates urgency and the reason codes create improvement, and a team that collects the score without the reasons has bought a thermometer and no medicine.

The discipline is:

1. Every miss gets a reason, in the foreman's words, recorded on the spot Friday morning.
2. Every reason gets a category — one of the eight constraint categories, plus weather, plus a genuine "other."
3. Categories are counted and trended, not just listed. Six weeks of data, sorted by frequency: a Pareto.
4. The top two or three categories get a countermeasure with a name and a date — not a discussion.

Here is Northgate's Pareto again, drawn out so you can see the shape of it. Each ▉ is one miss:

 REASONS FOR VARIANCE — NORTHGATE, 6 WEEKS (Feb 9 – Mar 20, Year 2)
 86 missed commitments out of 187 made.  PPC = 54%.

 Prerequisite work not complete  ▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉▉  21   ┐
 Material not on site            ▉▉▉▉▉▉▉▉▉▉▉▉▉▉         14   ├─ 47 of 86 = 55%
 Information (RFI / drawing)     ▉▉▉▉▉▉▉▉▉▉▉▉           12   ┘
 Labor availability              ▉▉▉▉▉▉▉▉▉▉             10
 Space and access                ▉▉▉▉▉▉▉▉▉               9
 Inspection                      ▉▉▉▉▉▉                  6
 Submittal / approval            ▉▉▉▉▉                   5
 Owner or design change          ▉▉▉▉                    4
 Weather                         ▉▉▉                     3
 Equipment                       ▉▉                      2
                                                        ───
                                                         86

Read the three bars at the top and ask who owns them.

• Prerequisite work (21). Owned by the general contractor's coordination and the look-ahead screen. Kestrel's, entirely.
• Material (14). Owned by procurement and by subcontractors' own buyout. Mostly manageable by the look-ahead — material need-by dates were never back-calculated to the room level.
• Information (12). Owned by the RFI process (Chapter 25). Kestrel's log, Kestrel's escalation, Kestrel's failure to push.

Fifty-five percent of Northgate's misses were caused by Kestrel, not by a subcontractor. That was the finding that changed the conversation in the trailer, and it is the finding on most jobs. It is also the reason the score has to be safe to report: the moment a general contractor uses PPC as a stick to beat subcontractors with, it loses access to the data that shows the general contractor is the problem.

💰 Money check — what is a PPC improvement actually worth?

Let us do this honestly, which means it comes out smaller than the sales pitch and more defensible.

Step 1 — Measure the loss, don't assume it. Over the six measured weeks (30 work days), Wei Chen's updates showed the longest path advanced 26 work days for 30 work days of calendar. Net loss: 4 WD ≈ 6 calendar days over six weeks — about 1 CD of slip per week. That is the number Margo could feel and could not prove. It is also three times worse than the −3 CD on the whiteboard, because the monthly update absorbed most of it inside remaining durations.

Step 2 — Project it. Twenty-six weeks remained to substantial completion on September 18, Year 2. At 1 CD per week: 26 CD of slip.

26 CD × $10,650/CD = $276,900 of exposure ($5,150 extended general conditions + $5,500 liquidated damages)

Step 3 — Measure the improvement, don't assume it. Over the twelve weeks after Kestrel staffed constraint removal, the observed slip rate fell from about 1.0 CD/week to about 0.3 CD/week. Projected over the remaining 26 weeks: 8 CD instead of 26.

Days avoided: 18 CD. General-conditions burn alone: 18 × $5,150 = $92,700 If those days would have crossed substantial completion: 18 × $10,650 = $191,700

Step 4 — Price the system. It is not free, and pretending it is free is how people get sold on lean and then abandon it.

Step 5 — The net.

Now the honest caveats, because a number without them is marketing:

1. You cannot prove the counterfactual. You never run the same job twice. The 0.3 CD/week rate is what happened; whether it happened because of the system is a judgment, not a proof.
2. Some of those meeting hours displaced other meetings that stopped happening, so the true incremental cost is lower than $75,184 — but by an amount nobody measured.
3. On general conditions alone, this roughly pays for itself and not much more. The real return is in the tail: it is insurance against the day the slip crosses the contract date and the liquidated damages switch on. Sell it that way and you will be believed.
4. The largest beneficiary is not Kestrel. Every one of those avoided crew-idle events is a subcontractor's labor productivity, and the subcontractors paid nothing toward the $75,184. That imbalance is the central commercial problem of lean construction, and §27.9 is about it.

🔄 Check your understanding. A project manager reports PPC rising from 62% to 89% over six months and asks for a bonus. What is the first question you ask, and why?

27.7 Making Work Ready Is a Job Function, Not a Meeting

Here is the sentence that separates the projects where this works from the projects where it does not:

A project that adopts the Last Planner System without giving someone the time to remove constraints has adopted a meeting, not a system.

Constraint removal is real work. On Northgate, at any given moment, the six-week look-ahead carried between 35 and 60 open constraints. Every one of them required somebody to make a phone call, chase a supplier, escalate an RFI, book an inspection, confirm a delivery, or physically walk a floor and verify that a prerequisite was actually finished rather than reported finished.

That is not something a project manager does "as well as" running a $47.5 million job. Somebody has to own it.

Who owns it. Practice varies. The three workable models:

How far ahead it runs. Six weeks is the common default and it is a reasonable one for building construction, because it is long enough to chase a material delivery and short enough that the plan is still meaningful. But six weeks is not sacred. The right horizon is the longest lead time among your typical constraints. If your dominant constraint is an RFI with a 14-day contractual review, six weeks is generous. If your dominant constraint is a fabricated item with an eight-week lead, a six-week look-ahead will find the problem two weeks after it became unfixable, and you need ten or twelve. Look at your reason codes and let them set the horizon.

What Kestrel actually changed after the six-week baseline. Three things, all boring, all cheap:

1. Dani Okonkwo was given 20 hours a week for make-ready and taken off two other assignments to create the time. Not asked to "also do" it.
2. The look-ahead was re-cut by room, not by activity. "Interior rough-in Level 3" became nine zone-level lines. Prerequisite-work misses fell immediately, because you can verify a room and you cannot verify an activity.
3. Material need-by dates were back-calculated to the room and put in the constraint log with the vendor's name and a phone number. Material misses were the second-biggest category and most of them were things sitting in a warehouse forty minutes away that nobody had asked for.

None of that is clever. All of it takes somebody's time. That is the whole finding.

27.8 The Rest of the Lean Toolkit — What Each One Is, When It Earns Its Keep, and What It Costs

The Last Planner System is the core. Around it sits a set of practices worth knowing. Take them one at a time and adopt none of them because they are "lean" — adopt them when the problem they solve is a problem you actually have.

Takt time planning

Takt is German for beat or rhythm. Takt time planning divides the work into similar-sized zones and moves a fixed sequence of trades — a "train" of "wagons" — through them at a fixed rhythm. Every trade gets the same amount of time in every zone. When the drumbeat is right, the building fills up like an assembly line running through a stationary product.

The worked calculation — Northgate levels 2 through 4 interiors.

The three clinic floors are similar in size and layout. Level 1 (lobby, café, imaging suite, ambulatory surgery) is not similar to anything and gets left out — which is itself a lesson.

The train takes longer than the sum of one pass, because the last wagon still has to travel through all nine zones after the first wagon has finished. Total duration in takts is:

Total takts = Zones + Wagons − 1 = 9 + 5 − 1 = 13 takts

So the takt time is:

Takt = Available time ÷ Total takts = 45 WD ÷ 13 = 3.46 WD

Round DOWN, never up. Takt = 3 work days.

Duration = 13 takts × 3 WD = 39 WD. Buffer remaining: 45 − 39 = 6 WD.

🔍 Why you round down. The takt is a drumbeat and every wagon must fit inside it. If any trade cannot finish a zone in the takt, the whole train stops — the delay does not stay local, it propagates to every zone behind it. Rounding down produces a schedule that finishes early with a buffer at the end. Rounding up produces a schedule that exactly consumes the window and has no room for the first zone that goes wrong. And a zone will go wrong.

Here is the train:

   NORTHGATE TAKT PLAN — LEVELS 2–4 INTERIOR ROUGH-IN
   9 zones · 5 wagons · takt = 3 WD · 13 takts · 39 WD · start Mon Mar 30, Year 2

   ZONE │ T1  │ T2  │ T3  │ T4  │ T5  │ T6  │ T7  │ T8  │ T9  │ T10 │ T11 │ T12 │ T13 │
   ═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪═════╪
   2-A  │ LAY │ OVH │ FRM │ INW │ DRY │     │     │     │     │     │     │     │     │
   2-B  │     │ LAY │ OVH │ FRM │ INW │ DRY │     │     │     │     │     │     │     │
   2-C  │     │     │ LAY │ OVH │ FRM │ INW │ DRY │     │     │     │     │     │     │
   3-A  │     │     │     │ LAY │ OVH │ FRM │ INW │ DRY │     │     │     │     │     │
   3-B  │     │     │     │     │ LAY │ OVH │ FRM │ INW │ DRY │     │     │     │     │
   3-C  │     │     │     │     │     │ LAY │ OVH │ FRM │ INW │ DRY │     │     │     │
   4-A  │     │     │     │     │     │     │ LAY │ OVH │ FRM │ INW │ DRY │     │     │
   4-B  │     │     │     │     │     │     │     │ LAY │ OVH │ FRM │ INW │ DRY │     │
   4-C  │     │     │     │     │     │     │     │     │ LAY │ OVH │ FRM │ INW │ DRY │

Read any column and you know exactly how many crews are in the building and where. Read any row and a foreman knows exactly when their zone opens and closes. That legibility is most of the value.

Where takt earns its keep: repetitive floors — hospitals, hotels, multifamily housing, student housing, offices, parking structures, anything with a repeating unit. Where it does not: irregular, one-off spaces. Northgate's Level 1 has a two-story lobby, a café, an imaging suite with a depressed slab, and an ambulatory surgery suite. There is no rhythm in that, and forcing one produces a schedule that is beautiful and wrong.

What it costs: the hard part is balancing the wagons. You have to size zones so that every trade can finish every zone in the same 3 days, which usually means resizing zones and sometimes means adding or removing manpower to fit the beat. That is a design exercise, it takes real effort with the trades, and if you skip it the train derails in the first week.

5S and workplace organization

Sort, Set in order, Shine, Standardize, Sustain (from the Japanese seiri, seiton, seiso, seiketsu, shitsuke). A discipline for organizing a work area so that everything has a location, the location is marked, and deviations are visible at a glance.

Earns its keep: in shops, laydown yards, gang boxes, and fixed work areas — and as a safety intervention, because most housekeeping violations and a great many slip-trip-fall exposures are 5S failures wearing a different name. Costs: the fifth S. Sorting a laydown yard once is a Saturday. Sustaining it is a daily habit that dies the moment a superintendent stops walking it.

First-run studies

Before the first repetition of a repetitive task — the first patient-room headwall, the first curtain-wall stack, the first typical bay of formwork — you plan it deliberately, perform it while observing (often on video), review what happened with the crew, and adjust the method before you do it four hundred more times. It is Plan-Do-Check-Adjust applied to a means-and-methods decision.

Earns its keep: anywhere a task repeats more than about twenty times. A 6% method improvement on 400 repetitions is worth vastly more than the half-day you spent finding it. Costs: a half to a full day of a crew's time plus a superintendent's attention, spent at the exact moment everyone feels most behind.

A3 problem solving

A structured one-page problem-solving format, named for the ISO A3 paper size (roughly 11 × 17 inches). One page holds: background, current condition with data, goal, root-cause analysis, countermeasures, plan, and follow-up. The constraint is the point — if it does not fit on one page, you have not understood the problem.

Earns its keep: on recurring problems that keep getting "solved" in meetings and keep coming back. Costs: discipline. The first three A3s a team writes are usually a description of the problem with a solution stapled to it and no root-cause analysis in between.

Visual management and the big room

Visual management means the state of the project is legible from the wall without opening a laptop: the look-ahead, the constraint log, the PPC trend, the reason-code Pareto, the zone map with today's crews, the safety metrics. Obeya — Japanese for "big room" — or co-location means putting the project team, and often key trade partners and designers, in one physical space.

Earns its keep: big room co-location is genuinely powerful on large, fast, complex projects where the cost of a two-day email exchange exceeds the cost of a desk. Costs: real money — space, travel, and people's time away from other projects — and it is usually the first thing cut when a job gets tight, which is exactly when it was working.

Choosing by advantages (CBA)

A decision method, developed by Jim Suhr, in which alternatives are compared by the importance of their advantages rather than by pro-and-con lists or weighted scoring. Its discipline is that you must state the factors, the attributes of each alternative on each factor, and then the advantage — the difference between the least-preferred attribute and each other one — and only then judge importance. Cost is deliberately considered last, after the advantages are understood.

Earns its keep: on contested multi-stakeholder decisions where a pros-and-cons list would let everybody argue past each other — enclosure system selection, mechanical system selection, phasing strategy. Costs: it is slower than a vote and it feels bureaucratic until the first time it stops a two-week argument in ninety minutes.

Target value design

Instead of designing a building and then estimating what it costs, target value design (TVD) sets a target cost from the owner's business case and designs to it, with the estimator and the trades in the room continuously. It converts cost from an outcome into an input. This is the mature form of what Chapter 11 describes as preconstruction services — and it is the difference between value engineering (cutting after the fact) and designing to a number in the first place.

Earns its keep: on negotiated work where the contractor is engaged early enough to influence design. Costs: it requires a delivery method and a fee structure that pay for early contractor involvement, which brings us to the next section.

27.9 Lean, the Commercial Terms, and Where It Fails

The IPD relationship, stated honestly

Integrated Project Delivery (IPD) — the multi-party agreement described in Chapter 3, with shared risk, shared reward, and often a waiver of claims among the parties — is the delivery method most closely associated with lean construction. The two grew up together, and IPD's commercial structure is genuinely well-matched to lean practice: when the owner, designer, and builder share a risk pool, every party's incentive points the same direction, and reliability improvements land in a pot everybody owns.

Now the honest note: most lean practice happens on ordinary design-bid-build and CM at Risk projects, not on IPD. IPD remains a small share of construction volume. Northgate is CM at Risk with a guaranteed maximum price. Willow Street is hard-bid, lump-sum, public design-bid-build. Both can run the Last Planner System, and one of them is where most of you will work.

You do not need IPD to run a pull plan. You do need to understand that without IPD's commercial structure, the benefits and the costs of reliability land on different parties — and that is the root of nearly every failure below.

⚖️ What the contract says

Three things worth knowing before you mandate any of this:

First: you can require participation, and you should require it in the subcontract, not in a meeting. A well-drafted subcontract can obligate the subcontractor to send a person with authority over the crew to a weekly planning meeting, to participate in phase pull-planning sessions, and to provide look-ahead information. If it is not in the subcontract, it is a request, and by month four a request is a suggestion. Put it in the scope sheet during buyout (Chapter 16), price it, and make sure the subcontractor knows it was priced.

Second — and this is the one nobody warns you about: a weekly work plan is a contemporaneous record. Every commitment a foreman writes down and every miss you code is a dated document about who said they would do what and what happened. In a delay claim it is evidence. It can help you enormously — a reason-code log showing twelve information-related misses traceable to a single unanswered RFI is a powerful exhibit (Chapter 33). It can also cut the other way, against you, when the misses are yours.

Subcontractors know this. A subcontractor who is reluctant to write down a commitment is not being obstructive — they are being appropriately cautious about creating evidence against themselves. If you want honest commitments, you have to be trustworthy with the record, and you have to say so out loud, in the first meeting, and then behave that way for eighteen months. The first time a general contractor waves a PPC printout in a negotiation, that job's data becomes worthless forever.

Third: never let a planning artifact substitute for contractual notice. A constraint logged on a look-ahead is not notice of delay. A commitment missed in a weekly work plan is not a written claim. Notice requirements are contractual, jurisdictional, and unforgiving (Chapter 5). Run both systems.

Where lean fails

This section is required reading and it is what will make you credible when you propose any of this. Six failure modes, all common, all survivable if you see them coming.

1. Adoption as ritual. The team runs the meetings and never removes a constraint. The look-ahead is printed, distributed, and not screened. Sticky notes go on a wall in January and are still there, curling, in June. This is the most common failure by a wide margin, it is what Case Study 2 is about, and its signature is a team that can describe the process fluently and cannot tell you how many constraints they closed last week.

2. A general contractor that demands commitments and makes none. You ask fourteen foremen to promise in public and be scored, and you — the party responsible for information, access, coordination, inspections, and clearing the area — promise nothing and are scored on nothing. Foremen notice this in about three weeks. The general contractor must put its own commitments on the same board and take the same score. On Northgate, Kestrel's own commitments ran between six and nine items a week and were scored in the same column as everyone else's, and Kestrel's own PPC was below the job average for the first month. Publishing that was more persuasive than anything Ray said.

3. Subcontractors who reasonably decline to expose their float. A subcontractor's schedule buffer is a priced asset. It is how they survive the other four jobs they are on, and how they absorb the general contractor who is behind and has not admitted it yet. Asking them to surrender it in a pull-planning session, for nothing, in exchange for a promise from a party that has not yet demonstrated reliability, is asking them to donate margin. They are right to refuse. They will trade float for reliability. They will not trade it for enthusiasm. You have to go first, and you have to go first for a while.

4. The benefits land on the party that did not pay for them. This is the structural one. The general contractor pays for the make-ready staff, the meeting hours, and the facilitation. A large share of the benefit is subcontractor labor productivity — crews not standing, not remobilizing, not being displaced. Under a fixed-price subcontract, that gain is entirely the subcontractor's, and they contributed nothing to producing it. This is not a moral failing; it is what the contract says. Unless the commercial terms change — shared savings, incentive pools, IPD-style risk pools, or simply a general contractor willing to invest because reliability protects its own general conditions and liquidated-damages exposure — the economics are lopsided and everyone can see it.

5. Cultural resistance that is a rational response to past bad faith. When a foreman says "we tried this on my last job and it was garbage," listen carefully, because they are usually describing failure mode 1 or 2 accurately. The industry has burned a lot of goodwill with initiatives that were meetings. Treating that resistance as ignorance guarantees you will reproduce the thing they are objecting to.

6. Staff turnover. A project runs for eighteen months. The superintendent who believed in it gets promoted in month nine. The field engineer who removed constraints gets moved to the next job. The subcontractor foreman who was the best planner in the room finishes his scope and leaves. This system lives in specific people's habits, and habits do not transfer with a handoff memo. Every change of a key person is a re-implementation.

What has to be true for it to work

Strip everything else away and there are four conditions. If all four are true, it works. If any one is false, it decays, and the decay is measurable within six months.

🪞 Learning check-in.

You are twenty-seven chapters into this book. Stop and take stock of something specific.

Think back to how you read Chapter 14 — the forward pass, the backward pass, total float, the critical path. When you finished it, what did you believe a schedule was? Be honest. Most readers finish that chapter believing a schedule is a model of the future that is correct if the logic and durations are correct.

Now, after this chapter, what do you believe a schedule is?

Sit with the gap between those two answers, because it is a real change in how you will see a job site, and it is worth naming:

• A CPM schedule is a model of what the work should do if nothing is missing. It is a necessary, rigorous, contractually essential document, and it is silent about missingness.
• A look-ahead with a constraint log is a model of what the work can do given what is actually missing right now. It is neither rigorous nor contractual, and it is the one that tells a foreman where to stand on Tuesday.

Neither is the truth. Both are models, built for different questions, and the skill you are developing is knowing which question you are asking.

One more piece of self-assessment, and this one is about you rather than the material. Of the six failure modes above, which one would you be most likely to cause? Most new project managers, honestly, would cause number two — demanding commitments while making none — not out of arrogance but because it does not occur to them that they are a trade with deliverables too. If you know which one is yours, you can watch for it.

Spaced Review

Answer these before you read the responses. Retrieval works; recognition does not.

1. From Chapter 26: What are the properties of an action item that make it real rather than decorative — and how many of those properties does a Last Planner commitment share?

2. From Chapter 14: The critical path is a calculated result, not a management opinion — and float is a shared, consumable project asset. What does the pull plan in §27.4 do to the float on the enclosure path, and who benefits?

3. Deep callback, Chapter 19: What is trade stacking, what does it do to productivity, and what is the connection between the constraint screen in §27.5 and stacking?

⚠️ Safety alert — the connection you must not miss.

Go back to week 34 of this job. North elevation, level 3, 7:20 on a Tuesday morning. A frame scaffold had been partly modified overnight by a trade that did not erect it, to run conduit. A plank was lifted and not re-secured. Milo Serrano, a mason tender, stepped onto it and it shifted. He caught himself on the top rail and was not injured. Bea Salgado stopped work on the elevation for the day.

The investigation found three failures. The tag was two days stale. A scaffold was modified by a trade that did not erect it and was not re-inspected. And the third one — the one nobody wanted to write down — was that a crew was running behind after the steel acceleration, under an unwritten pressure to make it up.

Here is the link, and it is not decorative. Variability produces firefighting, and firefighting is the condition in which people get hurt. When a crew's work was not ready this morning, someone improvises. Someone modifies a scaffold at night to stay ahead of a schedule they are behind on. Someone works in an area that is congested because they were displaced from the area they planned to be in. Someone skips the inspection tag because the alternative is standing still while a superintendent watches.

A planning system that surfaces constraints six weeks out is a system that generates less of that pressure. That is not a marketing claim about lean — it is the direct consequence of finding Chapter 24's central argument and taking it seriously: safety is a property of the production system, not a rulebook. An unreliable production system produces schedule pressure the way a leaking pipe produces water, and no amount of toolbox talks will dry the floor while the pipe is still leaking.

The most underrated safety intervention in this book is a constraint log that runs six weeks ahead and is actually worked.

Project Checkpoint: The Willow Street Look-Ahead, Pull Plan, and Constraint Log

In Chapter 26 you built the Willow Street meeting structure — agendas, five completed daily reports, and a photo protocol. You now have a way to record what happened. This checkpoint gives you a way to decide what should happen next, and to know in advance whether it can.

Recall the project: the Willow Street Community Center, City of Rivermont Parks and Recreation, $6.8M lump sum, 24,000 SF, two stories, wood-framed second floor over structural steel and CMU first floor, 425 calendar days, liquidated damages $1,200/CD. Full package in Appendix K.

Deliverable: a four-part production planning package. Add it to the Project Notebook behind the meeting structure.

Part 1 — A pull plan for the enclosure and interior start. Define your phase and its end milestone: building dried-in. Pick the date from your Chapter 14 CPM schedule. Then build the wall backward from it. Your trades are roofing, masonry, rough carpentry and sheathing, windows and storefront, exterior finishes, waterproofing and flashing, and whichever mechanical and electrical work has to be complete before the roof and the walls close. For each note write: trade, deliverable, duration in work days, crew size, and — the essential line — what I need from the trade before me. Then write down what "dried-in" means, in one paragraph, in enough detail that three trades could not disagree about it. If you cannot write that paragraph, you have found the first real product of this exercise.

Part 2 — A six-week look-ahead. Take your Chapter 14 CPM and cut the next six weeks out of it, then re-cut it. Break every activity longer than five work days into pieces small enough that one crew could complete one in a week, and break them by area — by room, by wing, by elevation — not by activity name. You should end with somewhere between 25 and 45 lines. This re-cutting is the work; the printing is not.

Part 3 — A constraint log covering all eight categories. For every line in the look-ahead, run the eight questions: design information, submittals and approvals, material on site, labor, equipment, prerequisite work, permits and inspections, space and access. Log every constraint you find with a category, a named human owner, and a need-by date back-calculated from the activity's start. Then do the part that matters: screen. Mark each activity ready or not ready. Expect to screen out a third of them. If you screen out none, you have not been honest with yourself.

Pay particular attention to Willow Street's specific traps: the commercial kitchen (health-department review, long-lead equipment, and a gas service you do not control), the 8-inch water main relocation (a permit, a utility, and a shutdown window that belongs to the City), and the gymnasium wood floor (which has a moisture-content constraint that is a condition, not a date, and which almost nobody logs).

Part 4 — The PPC and reason-code tracking sheet. Build the sheet you will actually use: one row per commitment, columns for trade, foreman, deliverable, complete Y/N, and reason code from your eight-category list. Below it, a weekly summary line: commitments made, commitments completed, PPC, and a running Pareto of reason codes. Include a row for your own commitments as general contractor, and score yourself in the same column.

Next checkpoint: Part VI opens with Chapter 28, where you build the cost-code structure, load the budget, and produce a month-6 cost report with a cost-to-complete forecast. The planning system you just built tells you what is coming; the cost system tells you what it is doing to your money.

Chapter Summary

The four-word system. SHOULD → CAN → WILL → DID. Master schedule and pull plan say what should happen. The look-ahead screen says what can. The weekly work plan says what will. PPC and reason codes say what did — and feed it back.

The five levels, and what each produces:

The eight constraint categories — recite these: design information · submittals and approvals · material on site · labor · equipment · prerequisite work · permits and inspections · space and access.

The rules that make it work:

1. An activity does not enter the weekly work plan until every constraint is removed. Not "probably." Removed.
2. PPC is binary. Ninety percent of a handoff is zero handoffs.
3. Track commitments made alongside PPC, or the number will be gamed.
4. The reason codes, categorized and trended, are where improvement comes from. PPC is the thermometer.
5. Somebody must have protected time to remove constraints. Otherwise you have a meeting.
6. The general contractor's own commitments go on the same board and take the same score.
7. A recorded miss must be safe to declare, or the data dies.

The arithmetic worth remembering:

The four conditions. Protected make-ready time · the GC commits and is scored too · a miss is safe to declare · the commercial terms let the benefit be shared, or the GC knowingly invests. All four, or it decays inside six months.

And the sentence to carry out of Part V: the schedule tells you what should happen; the constraint log tells you what can — and making work ready is a separate discipline from scheduling it.

What's Next

Part V closes here. You have the three information systems that run a project: the paper trail that gives it a memory (Chapter 25), the meetings and reports that give it a voice (Chapter 26), and the planning system that decides what work is actually available to do.

Part VI is project controls — cost, time, and money. Chapter 28 builds the cost-code structure and the cost report, and asks the only question that matters at month six: are you making money, or do you just have a lot of cash in the account? Then Chapter 29 takes the schedule variance you now know how to see and turns it into updates, recovery plans, and priced acceleration options — including the one Kestrel had to run on the steel.

You have spent twenty-seven chapters learning to plan the work. Now you learn to find out what the work is doing to you, in dollars and days, while there is still time to change it.