I've stood in enough Monday morning coordination meetings to know the exact moment a look-ahead schedule stops being useful. It's when a foreman looks at a line that says "Electrical — Level 3" and asks, dead serious, "Okay, but what am I actually doing Tuesday?" Nobody at the table can answer, because the activity was written at a level that means nothing to the guy with the tools. That gap — between what's on the schedule and what a crew can actually pick up and run with — is almost always a question of detail. Too little and the plan is wallpaper. Too much and you spend more time maintaining the schedule than building the job.
Getting the grain size right is one of the least-talked-about skills in short-interval scheduling, and it's the difference between a weekly work plan that drives the field and one that gets ignored by Wednesday. Here's how I think about it after two decades of building these things.
The right size is a day's work you can point at
My rule of thumb is that an activity should be one to three days of work for the crew that owns it, and you should be able to walk the job and physically point at whether it's done. "Rock the south corridor" — I can stand in that corridor and tell you if the board is hung. "Continue drywall" — I can't tell you anything. It could be 10% done or 90% done and the schedule reads identically either way.
Why one to three days? Shorter than a day and you're generating tracking overhead that outruns the value. You don't need "install door 214A" as its own line; the crew that hangs doors will hang a dozen in a shift and nobody needs a status update on each one. Longer than about three days and you lose visibility into whether the work is actually tracking. A five-day activity that's "behind" tells you nothing until it's too late to react — which is the whole point of a look-ahead in the first place. Two to three days is the sweet spot where a slip shows up in this week's plan-percent-complete instead of next month's schedule update.
The tell that you've got it right: your foreman can look at the line, nod, and commit to it without hedging. If he has to say "well, depends what you mean," the activity is too vague to promise.
Detail should sharpen as the work gets closer
Here's the mistake I see most on new teams — they try to plan all six weeks at the same fine grain. That's a waste, because most of what you write for week five is going to change before you get there. Details should come into focus as work approaches the commitment week, not before.
Think of it in three bands:
- The commitment week (week one). This is where you earn your keep. Every activity here should be crew-specific, location-specific, and constraint-cleared. This is the work you're actually promising to complete, so it needs to be tight enough that a foreman can look at it and honestly say "yes, I can get that done."
- The make-ready weeks (two and three). Detail enough to see the constraints — the material that needs to land, the inspection that has to clear, the trade ahead of you that has to finish. You're not committing yet; you're clearing the path. Activities can be a notch coarser, but still specific enough that you know what has to be true for the work to go.
- The outer horizon (weeks four through six). Summary-level is fine here. "Frame Level 4" is a perfectly good outer-week line. It'll get broken into wings, then rooms, then specific walls as it rolls in. Writing room-by-room detail this far out is guessing dressed up as planning.
This is progressive elaboration, and a good rolling schedule is built to do it — you pull an activity in from the outer weeks and break it down as it lands in the make-ready band. Tools like LookAheadWall make that mechanical instead of a retyping exercise, but the discipline matters more than the software: don't invest detail in work that hasn't survived contact with the outer weeks yet.
One crew, one activity
Every activity should be ownable by exactly one crew. The second you write a line that two trades share, you've created an accountability hole nobody will admit falling into. If the "install and connect rooftop units" line covers both the mechanical crew setting the unit and the electricians making it live, then when it doesn't finish, both point at the other and you learn nothing.
Split it. "Set RTUs 1–4" belongs to mechanical. "Power and controls to RTUs 1–4" belongs to electrical, and it's a make-ready item that depends on the first one finishing. Now when something slips you know exactly whose promise broke and why — which is the only way variance analysis ever teaches you anything.
The same logic applies to scope. An activity should match what that crew can realistically knock out in the window. If your drywall crew is four guys and you've written a two-day activity that's really a week of hanging, you didn't make a plan, you made a wish. Size the work package to the crew you actually have on site, not the crew you'd like.
Tie detail to location, because that's how work flows
The single most useful thing you can do to sharpen an activity is anchor it to a place. "Frame the building" is useless. "Frame Level 2" is better. "Frame Level 2, west wing, rooms 201–208" is something a crew can start and finish and a super can inspect.
Location detail does more than clarify scope — it's what lets you see the flow. When your activities carry real areas, you can look at the plan and watch the work move through the building: framing leads, MEP rough-in follows two areas back, insulation and board chase that, finishes trail. That's the whole idea behind trade-flow sequencing, and it only works if activities are pinned to locations. If everything's just "Level 2," you can't see the collision coming when the drywall crew and the rough-in crew both want the same corridor on Thursday. Location-based activities surface that a week early, while you can still stagger them.
This is also where a schedule built around locations beats a flat task list. When you can see who's working where, the coordination conversation in the weekly meeting stops being "are we on track" and becomes "you're in 204, so I'll take 205 and follow you down." That's the conversation that actually prevents trade stacking.
Not every trade wants the same grain
Detail isn't one-size-fits-all across trades. MEP rough-in in a complex building — think a hospital or a lab — often earns room-by-room breakdown, because the coordination is tight and a single congested ceiling can stall three trades. That level of detail pays for itself.
Linear work is the opposite. Underground, site utilities, a long run of exterior wall — that progresses by the foot, and chopping it into tiny pieces just clutters the plan. "Storm line, manhole 4 to manhole 7" is plenty. Finishes swing back toward fine detail, because they're location-heavy and punch-sensitive; "paint and trim rooms 310–318" gives you something you can actually walk and verify. Read the trade and the phase before you decide how fine to cut.
Detail is what makes a constraint findable
Here's the part people miss. The reason you break work down isn't just tracking — it's that you can't see a constraint against a vague activity. "Phase 2 HVAC" has no findable constraints because you don't know what it touches. Break it to "hang ductwork, rooms 301–310" and suddenly the constraints are obvious: is the ductwork on site, is the framing above complete, has the overhead layout been coordinated, is the area weather-tight. Specific activities make specific constraints, and specific constraints are the ones you can actually clear before they stop the work.
This is the engine of make-ready planning. You look two and three weeks out, you read each activity, and you ask what has to be true for it to go. You can only ask that question well if the activity is concrete enough to answer it.
Signs you've got it wrong in either direction
Two failure modes, and they leave different fingerprints.
Too coarse shows up as chronically low plan-percent-complete for no clear reason. Activities keep "not finishing," but nobody can say why, because the activity was so broad that some of it always slips. If your PPC is in the toilet and the post-mortems are all shrugs, your activities are probably too big to tell the truth. Chop them down until a miss points at a real cause.
Too fine shows up as update fatigue. The schedule takes an hour to maintain, foremen stop reading it because it's a wall of noise, and the weekly meeting runs long grinding through fifty micro-lines. If people are quietly ignoring the plan, it's often not because they don't care — it's because you buried the signal. Roll the trivial stuff up.
A quick gut check I use: a good three-week plan should print on a couple of pages a foreman will actually pin up, and read cleanly on a phone in the field. If it doesn't fit on paper and it doesn't fit on a screen, it doesn't fit in anyone's head either.
What good and bad actually look like
Good activities — specific scope, clear location, an obvious "done":
- Install ductwork, rooms 301–310
- Frame and board the south wing corridor
- Pull wire for Level 2 receptacles, west half
Poor activities — and why:
- "Continue electrical" — no scope, no location, no way to call it complete.
- "Phase 2 HVAC" — too broad to commit to or to find a constraint against.
- "Install receptacle in room 301" — the opposite problem; that's a task, not an activity, and tracking it individually is pure overhead.
How to dial it in on your own job
You won't nail the right grain from a blog post, and you shouldn't try to. Start moderate — activities in that one-to-three-day, single-crew, one-location range — and then let the field tune it for you. Ask your foremen straight up whether the lines are clear enough to commit to and small enough to manage. Watch your plan-percent-complete: if it's low because activities are fuzzy, sharpen them; if your update burden is eating your Fridays, roll things up. The number itself will tell you which way to move.
The goal isn't maximum detail or minimum detail. It's the level where a foreman can read a line, commit to it honestly, spot what's in his way a week out, and tell you at a glance whether it's done. Hit that, and the schedule stops being a document you maintain and starts being a tool the field actually uses. That's the only test that matters.