Rolling-wave planning in construction: a field guide
What rolling-wave planning is, how it differs from a look-ahead and a CPM baseline, how big the detail window should be, plus a worked example to copy.
By Maximilian Rozenberg
Rolling-wave planning is a way of scheduling a construction project in which the whole job is planned as linked work packages, and only the next few weeks are broken down into detailed, crew-level tasks. As the job moves, the detail window rolls forward: packages that are about to start get detailed from what actually happened on site, while the critical path and the end date stay in view the whole time.
It sits between two habits that both fail on real jobs: the fully detailed schedule built in week one that nobody trusts by week ten, and the look-ahead spreadsheet that has quietly come loose from the master schedule. Below: how it works, how it fits with look-aheads, CPM baselines and the Last Planner System, how big the window should be, a worked example, and the mistakes that sink it.
What is rolling-wave planning in construction?
The term comes from project controls. The U.S. Government Accountability Office’s Schedule Assessment Guide puts it precisely: far-term work sits in the schedule as planning packages, and “as time passes and future elements of the program become better defined, planning packages are broken into detailed work packages. This incremental conversion of work from planning packages to work packages is commonly known as ‘rolling wave’ planning.”
On a construction job, that comes down to three moving parts:
- The outline. The whole project as work packages (sitework, foundations, framing, dry-in, rough-ins, inspections, finishes, closeout), each with a duration, a responsible trade, and logic ties to the packages around it. The outline carries the critical path and the end date from day one.
- The detail window. The next few weeks, broken down into tasks a foreman can run: which crew, which floor or area, how many days, and what has to be finished first.
- The roll. On a fixed cadence, usually weekly, actuals go in, the window advances, and the packages that have just come into range get detailed with what the team knows now, not what someone guessed months ago.
A package doesn’t have to be precise to be useful. GAO notes that a single planning package “may represent several months of effort, generic work to be accomplished by a trade or resource group, or even a future contract or phase.” It does need an honest duration and real logic.
A note on terms: in earned-value language, the far-term placeholders are planning packages and the detailed units are work packages. On most building sites, “work package” just means a chunk of scope for a trade, and that’s how this guide uses it.
Why detailing the whole schedule up front backfires
A schedule detailed to the task for the full duration on day one has two problems. Most of the far-out detail is fiction: finishes haven’t been selected, MEP coordination isn’t done, half the subs aren’t bought out, and the RFIs that will reshape the work haven’t been written yet. And maintaining that fiction is expensive, so nobody does it. The schedule stops matching the site, the field stops looking at it, and the real plan moves to a whiteboard in the trailer.
GAO says the same in more careful language: “care should be taken not to detail ill-defined far-term effort so soon as to require constant revision as time progresses. More detail does not necessarily mean greater accuracy, and pursuing too much detail too early may be detrimental to the schedule’s quality.”
The opposite failure is just as common: a short-range plan that has drifted away from the master schedule. In a study of projects using the Last Planner System, Hamzeh, Ballard and Tommelein found stretches where crews completed 100% of their weekly commitments while the project schedule kept slipping. On that project, the correlation between weekly plan completion and schedule gain from one update to the next was just 0.13. Their diagnosis: “weekly work plans are not properly founded in the master schedule.”
Rolling-wave planning goes after both problems. Detail gets produced only once it can be accurate, and every detailed task hangs off a package that sits on the whole-job logic.
Rolling-wave planning vs. look-ahead vs. CPM baseline vs. Last Planner
People confuse these because they all deal with “the next few weeks.” They are layers of one planning system, not competing methods.
| Layer | Horizon | Level of detail | Question it answers |
|---|---|---|---|
| CPM baseline | Whole job | Logic-tied activities at contract level | When do we finish, and what’s critical? |
| Rolling-wave plan | Whole job, plus a window of a few weeks | Packages beyond the window, crew tasks inside it | What’s the plan at the right resolution for each horizon? |
| Phase pull plan | One phase or milestone | Handoffs between trades | How do the trades sequence to hit this milestone? |
| Look-ahead (make-ready) | 3–12 weeks, 6 most common | Tasks broken into operations, with constraints | Can this work actually start when scheduled? |
| Weekly work plan | One week | Commitments by the people doing the work | What will we get done this week? |
Glenn Ballard and Iris Tommelein’s 2020 benchmark for the Last Planner System describes the same hierarchy: “The master schedule is expressed in phases. The phase schedule is expressed in processes. The lookahead schedule is initially expressed in processes, but after task breakdown consists of operations.” Rolling-wave planning is how the schedule itself gains that detail over time. The look-ahead is the make-ready process you run on the leading edge of the detail. The weekly work plan is where crews commit.
The benchmark also notes that traditional look-ahead planning “has tended to be a dropout from a higher-level schedule,” a filter showing the next few weeks of the master schedule. In the Last Planner System, the look-ahead means “making scheduled tasks ready, and replanning when some scheduled tasks cannot be made ready.” For the mechanics of the look-ahead itself, see our guide to building a look-ahead schedule. This article is about the layer above it: how the schedule earns its detail.
How big should the detail window be?
There’s no universal number. The window has to be at least as long as it takes to make work ready on your job. Here’s what the evidence says:
- Last Planner guidance. The 2020 benchmark says “lookahead periods typically range from 3 to 12 weeks, depending on the lead time needed to remove constraints.”
- What practitioners use. In a survey of 112 Last Planner users reported by Hamzeh, Ballard and Tommelein (2012), 62% ran a six-week look-ahead, 18% three weeks, 11% four weeks and 8% two weeks.
- Large programs. GAO, writing for government programs rather than building sites, describes detailed activities “typically 4 to 8 weeks long” covering near-term work “typically within 6 months to a year of the current date.” The principle scales; the numbers don’t transfer to a site.
These are the starting points I’d use. They’re rules of thumb from running jobs, not research findings:
| Job profile | Detail window | Make-ready horizon |
|---|---|---|
| Remodel or small fit-out, local subs, short lead times | 2–3 weeks | 3–4 weeks |
| Typical commercial or multifamily building | 4 weeks | 6 weeks |
| Remote site, imported materials, heavy permitting | 6–8 weeks | 8–12 weeks |
| Design still in progress (design-build, fast-track) | To the next design package or milestone | One window beyond that |
Two refinements matter more than the exact number.
Long-lead procurement does not belong in the window. Switchgear, elevators, trusses, storefront and anything else with a long fabrication queue lives in the outline as its own activities from day one. The U.S. Army Corps of Engineers’ project schedule specification requires schedule activities for long-lead procurement, which it defines as “those with an anticipated procurement sequence of over 90 calendar days.” No look-ahead window is long enough to catch those in time.
Fixed window or milestone wave. GAO’s view is that “a best practice is to plan the rolling wave to a design review, test, or other major milestone rather than to an arbitrary period such as 6 months,” a variant it notes is sometimes called block planning. On a building job I use both: a fixed window that rolls weekly for production, and milestone waves (dry-in, rough-in inspections, drywall complete) for the bigger re-plans, where you pull-plan the next phase with the trades. Per the Last Planner benchmark, that phase pull planning “should occur at least one lookahead period ahead of scheduled start so tasks can be made ready.”
How to set up a rolling-wave schedule, step by step
- Build the outline as linked packages. Give each package a scope, a location, a responsible trade, a duration and its predecessors. GAO is explicit that far-term packages “should be logically linked within the schedule” so the critical path can be monitored, and that packages “on or near the critical path or that carry significant risk should be broken into smaller activities.” If you can’t name the trade responsible for a package, it’s too vague.
- Put long-lead items, approvals and hold points in the outline now. Procurement, permits, utility connections and required inspections are activities with their own durations, not notes on a task you’ll detail later.
- Pick a window and a cadence. Start from the table above. Weekly suits most jobs because it matches the coordination meeting.
- Detail the first window with the people doing the work. The Lean Construction Institute defines the last planner as “the person closest to the work with authority to make decisions regarding the schedule and to make reliable commitments to complete the work of their discipline.” That’s who breaks a package into tasks, not the scheduler alone.
- Run make-ready on the leading edge. The benchmark’s timing guide, assuming a six-week look-ahead: constraint analysis starts six weeks before a task’s scheduled start, information and material constraints go first because they’re the hardest to clear, and other constraints “can usually be removed within 2-3 weeks.” If a constraint can’t be removed, the task moves.
- Roll on a fixed day. Use the checklist below.
- Keep the baseline at package level. The contract baseline is the record you’ll need for time extensions and delay analysis. Detail inside the window is a forecast: update it freely, but don’t rewrite the baseline every time you roll.
Work package template
| Field | What goes in it |
|---|---|
| ID and name | ”WP-07 Insulation and drywall” |
| Scope and location | Which work, which floors or areas |
| Responsible trade | The sub or crew that will detail and run it |
| Duration | Working days, and how you got the number |
| Logic | Predecessors and successors, including hold points |
| Long-lead items | Anything with a lead time longer than your window |
| Known constraints | Submittals, inspections, permits, access, equipment |
| Definition of done | What has to be true before the next trade starts |
| Detail-by date | When it must enter the window |
Weekly roll checklist
- Record actual starts, actual finishes and remaining durations for everything in the window.
- Log every new constraint or blocker raised during the week, with an owner and a needed-by date.
- Advance the window by one week.
- Detail each package that has entered the window, with the trade that will do the work.
- Compare the detailed duration with the package duration. If the detail doesn’t fit, show the difference. Don’t squeeze it.
- Recalculate the schedule and check the critical path and the end date.
- Screen the make-ready horizon. Anything that can’t be cleared in time gets resequenced now, not on the morning it was due to start.
- Get the new window to the field before the weekly coordination meeting.
Worked example: a three-story, 24-unit apartment building
The numbers below are illustrative and simplified, but they’re the kind you’d see on a real job.
The project is a three-story, wood-frame apartment building with 24 units on a slab-on-grade and a 56-week contract duration. The team runs a four-week detail window and a six-week make-ready horizon, and rolls every Friday. Here’s the outline:
| # | Work package | Duration | Weeks |
|---|---|---|---|
| 1 | Sitework, utilities, building pad | 6 wk | 1–6 |
| 2 | Foundations and slab | 4 wk | 6–9 |
| 3 | Framing, floors 1–3, roof trusses | 9 wk | 10–18 |
| 4 | Dry-in: roofing, windows, weather barrier | 4 wk | 17–20 |
| 5 | MEP and sprinkler rough-in, floor by floor | 7 wk | 15–21 |
| 6 | Rough-in and frame inspections (hold point) | 1 wk | 22 |
| 7 | Insulation and drywall | 8 wk | 23–30 |
| 8 | Exterior siding and trim | 8 wk | 21–28 |
| 9 | Interior finishes | 14 wk | 31–44 |
| 10 | MEP trim, fixtures, appliances | 6 wk | 42–47 |
| 11 | Site improvements and paving | 6 wk | 38–43 |
| 12 | Final inspections, punch list, closeout | 9 wk | 48–56 |
The critical path runs 1 → 2 → 3 → 5 → 6 → 7 → 9 → 10 → 12. Twelve lines describe the whole job, and weeks 23 to 56 stay at this resolution until they come within range.
Status at the end of week 18. The detail window covers weeks 19–22 and holds 38 tasks. An excerpt:
| Task | Crew | Days | Needs first |
|---|---|---|---|
| Roof sheathing and underlayment | Framing crew | 4 | Trusses set (done) |
| Shingles and flashing | Roofing sub | 6 | Underlayment complete |
| Windows, floor 3 | Window installer | 3 | Delivery due Monday, wk 19 |
| Plumbing rough-in, floor 2 | Plumber | 5 | Tubs set, floor 2 |
| Sprinkler rough-in, floor 2 | Sprinkler sub | 4 | Shop drawings approved |
| Electrical rough-in, floor 2 | Electrician | 6 | Plumbing and HVAC runs in |
The make-ready horizon reaches week 24, so the constraint log already covers the inspection hold point and the start of drywall:
| Constraint | Blocks | Owner | Status |
|---|---|---|---|
| Sprinkler shop drawings approved | Sprinkler rough-in, floors 2–3 | PM / sprinkler sub | Returned for revision |
| Rough-in inspections booked, by floor | Frame inspection | Superintendent | Book when floor 1 is complete |
| Frame inspection passed | Insulation (package 7) | Superintendent | Depends on the above |
| Drywall delivery and stocking plan | Hanging, floor 1 | Drywall sub | Ordered, date not confirmed |
The roll at the end of week 19. Three things change.
First, the floor 3 windows showed up on Thursday instead of Monday. Dry-in now finishes midweek in week 21 instead of at the end of week 20. Dry-in isn’t on the critical path and insulation doesn’t start until week 23, so the slip uses three days of float and the end date holds. The roll records that, so nobody counts on float that’s already gone.
Second, the sprinkler shop drawings are still out for revision. If they aren’t approved by midweek in week 20, sprinkler rough-in on floors 2 and 3 can’t finish before the week-22 inspections, and the hold point slides with everything behind it. The item has been on the log since week 16, which is why there’s still a week to recover. Found on the Monday it was due, it would have been a delay.
Third, package 7 (insulation and drywall) enters the window and gets detailed with the drywall sub’s actual crew plan: insulation and its inspection floor by floor, hanging, three coats of finish, and the rated corridor assemblies. The detailed critical chain comes to 43 working days; the package allowed 40. The roll shows the three-day overrun instead of compressing the tasks to fit, so the PM gets a decision now rather than a surprise in week 30: accept a later finish, or add a second finishing crew on floor 3. The PM and the drywall sub agree on the second crew, and the decision is written down against the package.
None of this needed weeks 31 to 56 planned in detail. It needed the next four weeks detailed, the next six weeks screened, and the whole job linked.
Make-ready and hold points: where rolling-wave plans earn their keep
A window full of tasks that can’t start is worse than no detail at all, because it looks like a plan. The Last Planner benchmark splits constraints into physical ones (“rebar installation prior to concrete placement”) and information ones (“soils report before foundation design, engineering details before fabrication, permit before hazardous work”). On a building job, screen every task entering the make-ready horizon against this list:
- Design information: current drawings, RFIs answered, conflicts resolved.
- Submittals: approved, not just submitted.
- Materials: on site, or delivery confirmed for a date before the start.
- Prerequisite work: finished and handed off, not “mostly done.”
- Inspections: booked, or passed where they’re a hold point.
- Permits: including work permits such as hot work.
- Equipment, access and space: lifts, laydown, and the right to be in the area.
- Crew: confirmed, with the right skills.
Most teams know this list. The gap is timing. In the Hamzeh, Ballard and Tommelein survey, 48% of respondents said half or more of their constraints were identified and removed at the weekly work plan level, the last step before execution. The authors concluded that companies “do not give enough attention to identifying constraints and removing them prior to activity execution.”
Hold points deserve their own line in the outline because they guard work that is about to disappear: rough-ins behind insulation, insulation behind drywall, under-slab plumbing under concrete. Once the next trade covers the work, nobody can check it. Make the inspection or sign-off an activity with its own duration, and capture the evidence before the cover-up. Photos taken to a checklist before the next trade starts are the cheapest record you’ll ever have, and it’s the job Nodle’s photo checks are built for: each task gets a checklist timed before the work is hidden, and the manager signs it off.
Common rolling-wave planning mistakes
Packages with no logic
An outline of unlinked bars is a picture, not a schedule. Without logic ties there’s no critical path, so the roll can’t tell you what a slip does to the end date.
A window shorter than your lead times
If a submittal takes three weeks to turn around and your window is two weeks, that constraint will always show up too late. Hamzeh and colleagues put it plainly: “When constraints requiring a lead time beyond the weekly work plan window are identified late in the process, the chances of removing them prior to task execution are seriously diminished.”
A look-ahead that’s just a filter
If the tasks in your window are the master schedule’s bars cropped to three weeks, nothing was planned. In the same survey, 47% of respondents agreed that tasks keep the same level of detail throughout the look-ahead process, and the authors list “presenting merely a lookahead filter of activities shown on the master schedule” among their main concerns with current practice.
Squeezing the detail to fit the package
When the detailed tasks add up to more than the package allowed, the tempting move is to shave durations until they fit. Don’t. The difference is the most useful number the roll produces. Surface it, decide, and write the decision down.
Rewriting the baseline every roll
Detail is a forecast; the baseline is a record. If every roll overwrites the baseline, you lose the evidence you’ll need when you ask for a time extension.
Detailing without the trades
A scheduler can break framing into tasks. Only the framing foreman knows the crew size, the sequence they actually use, and what they need from the trade before them.
Can you use rolling-wave planning when the contract requires a CPM schedule?
Yes, but the contract wins, so read the scheduling specification first. Some specs build rolling-wave planning in. The current USACE/NAVFAC/AFCEC guide specification for project schedules (UFGS 01 32 01.00 10, August 2026) asks for a preliminary schedule within 15 calendar days after notice to proceed is acknowledged, detailed for the first 90 calendar days. It “may be summary in nature for the remaining performance period” but must be “logically tied.” On design-build work, the initial schedule is detailed through dry-in, and the remaining construction is fully detailed “concurrent with the monthly schedule updating process” as design packages are developed. The same spec caps non-procurement activity durations at 20 workdays or 30 calendar days.
Where a spec demands full detail at baseline, comply with it and run rolling-wave planning underneath as the production plan. Map each detailed task to the contract activity it belongs to, so the contract schedule and the field plan never tell two different stories.
Doing it every week
The method is simple; the discipline is the hard part. The roll is the first thing dropped in a bad week, and a bad week is exactly when a stale window costs you. Any CPM tool or spreadsheet can hold the outline and the window. What most teams lack is someone with the time to roll them every week.
That’s the chore Nodle’s rolling-wave planning takes over. It outlines the whole job as work packages with the critical path intact, and details the next few weeks (four by default, adjustable from one to twelve). Every night it proposes the next roll from what happened on site: blockers raised in task chats, decisions, check-ins, and work that finished early or late. If a package needs longer than it was given, the proposal says so instead of squeezing it in. Inspections, permits, deliveries and approvals named in your drawings and documents become make-ready tasks that quote the exact line they came from. Nothing changes until you approve it.
The takeaway
Plan the whole job, but only at the resolution you can defend. Keep the outline linked so the critical path and the end date are always real. Detail the next few weeks with the people doing the work, screen the weeks after that for constraints, and roll the window on a fixed day. When the detail doesn’t fit the package, say so. Done this way, the schedule stays close to the site for the whole job, not just the first month.
If you’d rather not run the roll by hand, see how Nodle runs rolling-wave planning and try it on your next job.
Frequently asked questions
Is rolling-wave planning the same as progressive elaboration?
Close, but not identical. Progressive elaboration is the general principle that a plan gains detail as information improves. Rolling-wave planning applies it to the schedule on a cadence: far-term work stays in packages and becomes detailed tasks as it enters a fixed window or reaches a milestone.
Is rolling-wave planning agile or predictive?
Predictive, with an adaptive edge. The outline, logic, critical path and contract end date are planned up front, as in any CPM schedule; only the level of detail is deferred. That suits construction, where the end date is a contract term but the details of week 30 can’t be known in week 1.
How far ahead should a construction schedule be detailed?
Far enough to make the work ready. Last Planner guidance puts look-ahead windows at 3 to 12 weeks depending on how long constraints take to clear, and six weeks was the most common choice among Last Planner users surveyed. For most building jobs, a four- to six-week detail window works, with long-lead procurement in the outline from day one.
What is the difference between a work package and a planning package?
In earned-value and federal scheduling terms, a planning package is a placeholder for far-term work: scoped, estimated and logic-tied, but not yet broken down. A work package is the detailed version, created as the work comes within range. GAO’s guidance is that planning packages must still be logically linked so the critical path holds while they wait.
Does rolling-wave planning replace the Last Planner System?
No. Rolling-wave planning decides when the schedule gains detail. The Last Planner System governs how that detail is made ready and committed: phase pull planning, look-ahead make-ready, weekly work plans and learning from missed commitments. A rolling-wave schedule gives Last Planner a master schedule that stays connected to the weekly plan.
Can you do rolling-wave planning in Primavera P6 or Microsoft Project?
Yes. Any CPM tool can hold a package as a single logic-tied activity and later replace it with detailed activities. The hard part is the discipline: rolling every week, keeping the baseline at package level, and refusing to squeeze detail into a duration it doesn’t fit.
About the author
Maximilian Rozenberg is the founder of Nodle. He spent 15 years in construction, working his way from site supervisor to project director. As Director of Project Office at Renova, he delivered 850,000 m² (about 9.1 million square feet) of construction, and he built a construction-management platform before starting Nodle.
Sources
- U.S. Government Accountability Office, GAO Schedule Assessment Guide: Best Practices for Project Schedules (GAO-16-89G), “Rolling Wave Planning”, December 2015.
- Glenn Ballard and Iris D. Tommelein, 2020 Current Process Benchmark for the Last Planner System of Project Planning and Control, Project Production Systems Laboratory, University of California, Berkeley, 2021.
- Farook Hamzeh, Glenn Ballard and Iris D. Tommelein, “Rethinking Lookahead Planning to Optimize Construction Workflow”, Lean Construction Journal, 2012, pp. 15–34.
- Lean Construction Institute, “Last Planner System”, accessed October 2026.
- USACE / NAVFAC / AFCEC, Unified Facilities Guide Specifications, Section 01 32 01.00 10: Project Schedule, August 2026, via the Whole Building Design Guide.