Why three weeks became the number crews actually use
Walk any decent-sized jobsite on a Friday afternoon and watch what the foremen do. They're not studying the master schedule. That 2,000-line Gantt lives on the trailer wall and mostly gets pointed at during owner meetings. What the foremen are looking at is a short window — this week, next week, and the week after. Three weeks. Ask them why, and most can't give you a textbook answer. They just know it's the range where the schedule stops being a wish and starts being a plan they can bet their crew's week on.
I've run enough jobs to watch this play out over and over. Push a foreman to plan in fine detail out to six or eight weeks and his eyes glaze over. Give him a one-week list and he's flying blind on anything that needs a lead time. Three weeks lands in the pocket. It's worth understanding why, because if you build your weekly planning and your coordination meetings around that window, the field actually engages instead of nodding and walking off.
Three weeks matches the lead times that actually bite you
The real reason three weeks works isn't psychology — it's procurement and inspection lead time. Think about what has to line up before a crew can put hands on work:
- Material that's stocked or already on site: same-day to a couple days.
- A special-order item, a submittal that needs a shop drawing back, or a long-lead fixture: one to three weeks, sometimes far more, but three weeks is where you can still react.
- An inspection or a sign-off from another trade: usually needs 48 hours of notice and a clean predecessor.
- A crew you have to pull from another area or a sub you have to call in: a week of notice keeps you honest with them.
Three weeks is roughly the horizon where you can still do something about a problem. If you spot on Monday that Week 3 needs an anchor bolt template that isn't on site, you've got time to expedite it. Spot the same gap at one week and you're eating a delay. That's the whole game with a look-ahead — it's not about seeing far, it's about seeing far enough to still have moves. Six weeks out, you'll flag a hundred things that resolve themselves before they matter. One week out, everything you flag is already a fire.
The execute / prepare / warn structure
The cleanest way to run a three-week window is to give each week a different job. They're not three copies of the same list at different distances — they carry different weight.
Week one is a commitment, not a forecast
This week's work should be locked. Material staged, crew assigned, the predecessor trade actually finished — not "should be finished." If an activity is in Week 1 and any of that is still soft, it doesn't belong there yet. This is the discipline the Last Planner System calls a reliable promise: you only commit to work that's genuinely ready. The metric people track for it is Percent Plan Complete — of the tasks you committed to Monday, how many actually got done by Friday. A crew that habitually runs 50% PPC isn't lazy; it's committing to work that was never ready, and Week 1 is where that lie shows up.
Week two is where the real work of planning happens
Week 2 is active prep. This is where you confirm the delivery, book the inspection, and make the phone calls to the trades you're handing off to or receiving from. Most of a good foreman's planning time should live here. If Week 1 is already locked and Week 2 is where you're removing constraints, then by the time Week 2 rolls into Week 1, it walks in clean. Skip the Week 2 prep and every week arrives as a scramble.
Week three is a warning shot
Week 3 is coarse on purpose. You're not sequencing it hour by hour — you're scanning it for the long-lead landmines. What needs to be ordered now to be here in three weeks? What inspection or engineering answer has a queue? What area is going to have two trades trying to occupy it at once? You flag it, you start the clock, and you let the detail fill in as it moves forward. Trying to plan Week 3 to the same resolution as Week 1 is wasted effort — half of it will change.
Execute, prepare, warn. That rhythm is why the window works, and it's why a good look-ahead schedule visually separates the weeks instead of showing one flat bar chart. A foreman should be able to glance at it and know which mode he's in for each column.
A quick example of the window doing its job
Here's a coordination failure I've watched more than once, and how a three-week look-ahead catches it. Drywall is scheduled to close up a wall in Week 1. Looks fine. But the electrician hasn't had his rough-in inspected, and the low-voltage sub still has cable to pull in that same wall. On a master schedule those three activities are separate lines that all technically land in the same week — nobody's looking at them together.
On a three-week look-ahead built around location, they're stacked in the same wall, in the same week, and the conflict is obvious. You catch it in the Monday meeting: "You can't close that wall Thursday — Mike still needs the megger reading on those runs and Dave's got tel/data to pull." Now the fix is a two-minute conversation instead of a demo-and-repair. That's the entire value proposition. The schedule didn't do anything magic; it just put the trades that share a space in the same view, at a range where somebody could still act.
Why not six weeks? Or one?
People push longer horizons for a reason that sounds good — more warning is better, right? In practice a detailed six-week look-ahead quietly lies to the crew. It implies you know things about Week 5 that you flatly don't: the design might still be in RFI, the predecessor trade might slip, the owner might change the finish. When the field sees a schedule that pretends to certainty it doesn't have, they stop trusting the whole thing. And trust is the only currency a schedule runs on. Break it a few times — activities that keep sliding, work that shows up "ready" but isn't — and the foremen go back to running the job out of their heads.
One week is the opposite failure. It's honest but it's blind. You can't expedite a three-week-lead item in a one-week window. You can't give a sub fair notice. You're perpetually reacting. Three weeks is the shortest horizon that still gives you room to manage lead times, and the longest one where the field genuinely believes what it's looking at. That's not a compromise — it's the actual sweet spot.
It also fits how people live
There's a human piece to this that's easy to dismiss but shouldn't be. Crews think in weeks — pay periods, weekends, who's got a kid's thing on Thursday, who's out for a doctor's appointment. A three-week view spans about a pay-and-a-half period, which is a range people already plan their lives around. Ask a foreman on Friday "what are you doing next week and what do you need to get ready for the week after," and three weeks answers it exactly. Ask him about Week 6 and you're asking him to care about something that isn't real to him yet.
The same goes for the screen it lives on. A foreman is checking the plan on his phone, standing in a stairwell, one bar of signal, gloves half off. Three weeks fits that screen. Months of activities don't — they turn into pinch-zooming and scrolling, and a schedule you have to fight to read is a schedule that doesn't get read. This is a big part of why field tools like LookAheadWall default to the near-term window and keep the longer horizon a tap away rather than the front page. The default should serve the guy in the stairwell.
How to actually run it
If you want the three-week window to earn its keep, a few habits matter more than the software you pick:
- Run the weekly plan the same way every week. Confirm Week 1 commitments, work the Week 2 constraints, scan Week 3 for lead-time items. Same agenda, same order. Predictability is what gets the field to show up prepared.
- Only commit to ready work. An activity moves into Week 1 when its constraints are gone — material, prior trade, inspection, access, information. Not before. Guard that line and your PPC climbs on its own.
- Make the foremen write the plan, not receive it. A look-ahead someone hands down gets ignored. One the crew leaders build together in the room, they defend. This is where newer foremen learn to think ahead, and three weeks is a small enough canvas that they can actually learn it.
- Track what you promised versus what happened. Not to punish anybody — to find the pattern. If the same trade keeps missing its Week 1 commitments, the problem is usually upstream: it was never ready and got committed anyway.
None of this requires fancy tools. You can run a real three-week look-ahead with sticky notes on a whiteboard, and plenty of good superintendents still do. What software buys you is the location view, the trade-flow links that show a slip cascading downstream, and a version the subs can see from their own phones without you photographing a whiteboard. The point of a tool here isn't to replace the judgment — it's to make the conflicts visible at the range where the judgment can still do something.
The three-week window stuck around because it's the honest one. It's short enough that the field believes it and long enough that you can still steer. Build your planning rhythm around that, and the schedule stops being the thing crews tolerate and starts being the thing they run the job with.