How to Build a Three-Week Look-Ahead Schedule Crews Actually Follow

It's Monday morning and the three-week look-ahead is printed and taped to the trailer wall. By Tuesday the crew has moved off the first activity on it, because the rebar didn't ship and nobody knew until the truck didn't show. By Thursday nobody is looking at the sheet. Next Monday a new one goes up with the same activities shifted right by a week, and the same thing happens again.

That isn't a scheduling failure. The look-ahead was probably accurate the moment it printed. The problem is what it was: a filtered view of the master schedule showing the next twenty-one days of activities that were supposed to start. It listed what the plan wanted. It never asked whether any of it could actually happen.

A filtered master schedule isn't a look-ahead

Most look-aheads are produced the same way. Open the program, filter the date range, export to Excel, tidy the formatting, print. Ten minutes of work, and the output is a report, not a plan. Everything on it is inherited from logic and durations set months ago, before the utility crew slipped and before the shop drawings came back with comments.

That's why crews stop trusting the sheet. They can tell the difference between a plan somebody thought about and a printout somebody generated. If half the activities in week one are physically impossible on the day they're listed, the whole document is noise, and it gets treated as noise.

The genuinely useful version answers a different question. Not "what is supposed to start in the next three weeks" but "what has to be true before each of these can start, who is making it true, and by when."

The make-ready screen

This idea isn't new and it isn't ours. Look-ahead planning, make-ready work, and constraint analysis come out of the Last Planner System, developed by Glenn Ballard and Greg Howell and published through the Lean Construction Institute. You don't need to adopt lean construction wholesale to use the core move, which is this: take every activity in the window and screen it against a fixed list of things that can stop it.

On civil and marine work, that list is short and it doesn't change much job to job.

Preceding work. Is the activity in front of this one actually finished, or finished enough to hand over the area? Not per the schedule. Per the field.

Information. Are the drawings issued for construction, are the RFIs answered, are the shop drawings returned without comments that change the work?

Materials. Is it on site, or ordered with a confirmed delivery date inside the window? "Ordered" is not a cleared constraint. A confirmed date is.

Labour and equipment. Is the crew assigned by name, is the equipment on site or booked, and does the operator ticket match the machine?

Permits and approvals. Street use, traffic management, environmental approvals, utility locates. On BC work near water, the reduced-risk timing windows published by DFO and the province are effectively a permit constraint with a calendar attached, and they're region and species specific.

Access and space. Can the crew physically get to the work, and is the area free of somebody else's material?

Hold points. Does an inspection, a geotechnical sign-off, or an environmental monitor need to be there before the work proceeds or before it gets covered?

Run each activity through the list. Anything that isn't cleared becomes a constraint, and the activity stays out of the committed plan until it's cleared. That screening conversation, done properly, takes maybe forty minutes a week with the superintendent and whoever runs procurement. It is the entire value of the exercise.

The constraint log is the actual deliverable

The look-ahead sheet is what gets taped to the wall. The constraint log is what makes the sheet true. It needs four columns and no more: the constraint, the person who owns clearing it, the date it must be cleared by, and its status.

The need-by date is the column people get wrong. It is not the date the activity starts. It's the date the constraint has to be cleared so the activity can start on time, which for anything involving procurement or a third party is well before it. If the pipe has to be on site Monday the 14th, the confirmed delivery date is needed a week earlier so there's time to react when it slips. Set the need-by date at the activity start and you'll find out you have a problem on the morning you needed the solution.

One name in the owner column, always. Not "procurement," not "the sub." A person. Shared ownership of a constraint means nobody chases it.

Commit only to what's clear

This is the part that changes crew behaviour, and it's uncomfortable at first. Week one of the look-ahead is a commitment. Activities go into it only when their constraints are cleared. Everything still constrained sits in weeks two and three, visible, with an owner and a date against it.

The instinct is to leave constrained work in week one anyway, because taking it out looks like admitting slippage. But an honest week one is the only thing that earns the sheet any credibility. When the crew sees that what got committed last week is what actually happened, they start planning their own week around it. That mechanism doesn't work if week one is aspirational.

Track whether the plan held. The Last Planner metric is Percent Plan Complete: committed activities finished as planned, divided by activities committed. The percentage matters less than the reason column beside it. Five weeks of reasons tells you where your system actually leaks: late material, a sub who commits to work they can't staff, a design package that keeps coming back with questions.

Three weeks isn't long enough for some constraints

A twenty-one day window is the right horizon for committing crews. It's the wrong horizon for finding constraints. A twelve-week procurement lead on a specialty item, a fabrication slot, a permit with a statutory review period, an in-water work window fixed by the calendar: none of those can be cleared inside three weeks, and they won't appear on a three-week screen until it's already too late.

So run the constraint screen further out than you commit. Six to eight weeks is common and it costs almost nothing, because the screen at that distance is coarse. You're not sequencing crews, you're asking one question: is there anything in this activity that takes longer than three weeks to arrange? Log those the moment you see them and let them sit with a long need-by date. The three-week sheet stays a commitment tool. The longer screen is what keeps it from being ambushed.

What to automate here, and what to leave alone

The mechanical half of this is genuinely automatable. Pulling the activity list from the program into the look-ahead format, rolling the window forward each week, carrying open constraints across, flagging anything past its need-by date, drafting the Monday email with last week's plan-hit rate and the overdue list. That's transcription and chasing, and it's the part that quietly consumes a coordinator's Friday. A small script or a well-scoped AI-assisted workflow reading your existing schedule file and constraint log can do it on the tools you already own, without a platform purchase.

The screening conversation is not automatable and shouldn't be. "Is the area actually handed over" is a judgment call by someone who was on site this week. An AI tool that invents a cleared constraint is worse than no tool, because the sheet keeps its authority while losing its truth. Automate the paperwork around the meeting. Keep the meeting.

The vendors selling look-ahead modules aren't wrong that tooling helps, mostly with distribution and version control. But no tool decides your constraint categories, makes the superintendent run the screen, or holds week one honest. Those are process decisions and they're free.

Start with one project

Don't roll this out company-wide. Pick one active job. Add the constraint log next to the existing look-ahead, run the screen once a week for a month, and hold the rule that nothing enters week one with an open constraint. You'll know inside three weeks whether it's working, because the crews will start asking where the sheet is instead of ignoring it.

If your look-ahead is getting produced every Monday and overtaken by Wednesday, the gap is usually between the schedule and the constraints nobody logged. On a discovery call we walk through how planning information actually moves through your team, where the commitments break, and what the smallest fix looks like on the systems you already run.