Project scheduling and look-ahead planning field guide

The schedule runs the job

Project scheduling is the plan for what gets built, in what order, and by when, with the logic that ties each activity to the ones before and after it. On a data center or a large MEP job the schedule is not paperwork that lives in the trailer. It is the instrument that decides whether twelve trades flow through the building in sequence or pile up in the same ceiling on the same Tuesday.

Three documents do the real work. The master schedule lays out the whole job from mobilization to commissioning on the critical path method, so everyone can see the chain that drives the finish date. The look-ahead, a rolling three to six week window, pulls that master into the now and tells the field what is coming and what has to be ready for it. The pull plan is where the trades themselves build the sequence backward from a milestone, so the order is theirs and not imposed. Lose any one of the three and the job drifts.

A job without a real schedule does not fail all at once. It drifts. The slab pour slips a week, nobody re-sequences behind it, the conduit crew shows up to a deck that is not ready, and three weeks later the trades are colliding in the overhead and tearing out each other's work. Two of the disciplines that keep that from happening have their own guides. Sequencing and coordinating the trades is covered in the subcontractor management guide. The open questions and gear approvals that constraints depend on live in the RFI and submittal guide. This guide is about the schedule itself: how it is built, how it is protected, and how it meets the field.

Why the schedule decides the job

The schedule controls four things the owner and the contractor both lose sleep over: the finish date, the trade sequence, the money tied to time, and cash flow.

The finish date is the one the contract is written around. On a data center, the owner has a customer waiting on power and cooling for a hall full of servers, and that date carries real money behind it. Miss it and the contract usually has liquidated damages, a fixed dollar amount per day of late completion that comes straight out of the contractor's pocket, set in advance because it is cheaper to agree on than to litigate the owner's actual lost revenue. Liquidated damages are not a threat the schedule helps you dodge. They are a large part of why the schedule exists.

The trade sequence is the daily version of the same problem. Twelve trades cannot occupy the same space at once, so the schedule decides who is in the overhead first and who follows. Get the order wrong and the second trade tears out the first to get its run in, and you pay twice for the same six feet of ceiling while you fall behind.

Cash flow rides on the schedule too. Progress billing is tied to work in place, so a schedule that slips is a payment that slips, and a contractor carrying payroll for crews that cannot work because the deck is not ready is bleeding money the budget never planned for. The schedule is what lets the PM see that collision coming far enough out to act. A job with no schedule finds out at the end. A job with a real one finds out with weeks to do something about it.

What is the master schedule?

The master schedule is the full plan for the job, every major activity from mobilization through commissioning and turnover, built on the critical path method so the dependencies are explicit and the finish date is calculated rather than guessed. It is the baseline the whole project is measured against.

CPM is built from three parts. Activities are the pieces of work, each with a scope you can put a crew on. Durations are how long each takes, ideally from production rates and crew sizes instead of wishful round numbers. Logic ties are the relationships between them: this activity cannot start until that one finishes, this one can start once the one ahead is partway done. Feed those in and the schedule computes the longest path, the early and late dates, and the float on everything off that path.

The discipline that separates a real master schedule from a bar chart someone drew is the logic. A chart with no ties is a wish list. The moment one activity slips, a chart with logic re-flows and shows what moves behind it, while a chart without logic just sits there lying. Build the schedule with the trades who will run it, not in an office in isolation, because the foreman who bends the pipe knows the real duration and the real predecessor. A schedule the field did not help build is a schedule the field will not follow.

What is the critical path?

The critical path is the longest chain of dependent activities through the schedule, the sequence with zero float, where any slip moves the finish date day for day. If an activity is on the critical path and runs a day late, the job finishes a day late, full stop. That is what makes the path critical and everything else secondary.

Find it before you manage it. The schedule calculates it, but the PM has to know it cold, because it tells you where to spend attention. On a data center the critical path usually runs through the long-lead gear and the systems that have to be energized to start commissioning: the utility service, the switchgear, the generators, the gear room, then power-on, then the commissioning sequence. Those are the activities to protect, expedite, and watch daily.

Protect the critical path above everything else. When two crews want the same Saturday and one is on the critical path, the critical-path crew wins. When a delivery can be expedited for a fee and it is on the critical path, you pay the fee. The path also moves: clear one bottleneck and the longest chain can shift to a different sequence, so re-check it at every update. Manage the float on the rest, but the critical path is the chain that finishes the job, and it gets managed first.

What is float in scheduling?

Float, also called slack, is how long an activity can slip without moving something else. Total float is how long it can slip without moving the project finish. Free float is how long it can slip without delaying the very next activity. Critical-path activities have zero total float by definition. Everything off the path has some, and that float is the room you have to absorb trouble.

Knowing the float tells you what you can let slide and what you cannot. An activity with fifteen days of float can wait while you throw the crew at critical-path work that has none. An activity with two days of float is nearly critical, and one bad week turns it critical. Watch the near-critical paths, because a schedule that only watches the single longest chain gets surprised when a second path with three days of float quietly eats it.

Who owns the float is a contract question, and it gets fought over. The older view was that the contractor owned it, because the contractor built the sequence. Most modern contracts treat float as a shared project resource that belongs to the job, available to whoever needs it first rather than banked by one party. The practical reality is that whoever consumes it first usually keeps it, so check what your contract actually says about float ownership before you assume you can spend it, and record who used it and why.

What is a look-ahead schedule?

A look-ahead schedule is a rolling short-interval plan, usually three to six weeks, that pulls the master schedule into the immediate work in front of the crews. The master tells you the job finishes in March. The look-ahead tells you what fifteen specific things have to happen in the next four weeks, who is doing each, and what each one needs to be ready. It is where the schedule meets the field.

The look-ahead does the work the master cannot. The master is too coarse to run a crew off of. The look-ahead breaks the next few weeks into activities a foreman can plan around, in enough detail to catch the constraint before the crew is standing at the work with nothing to do. Pull it from the master every week, roll the window forward, and make it the agenda of the weekly coordination meeting with the trades.

The look-ahead only earns its keep if the field can see it and update it where the work is. One that lives in a file on the scheduler's laptop is one nobody reads. FieldOS keeps the rolling look-ahead, the constraints behind each activity, and the field status in one shared place tied to the job, so the foreman sees what is coming and the office sees what actually moved. Its tradeos workflow holds the look-ahead and the field updates in the same record instead of scattered across email and a printout taped to the trailer wall, so the next three weeks have one version everyone is working from.

What is pull planning?

Pull planning, the core of the Last Planner System out of the lean construction community, is a session where the trades build the sequence backward from a milestone instead of having it handed to them. You put the milestone on the wall, say power-on or a hall ready for commissioning, and the trades work backward with sticky notes: for me to do my work, the trade before me has to finish theirs, and for that to happen, this has to be ready. The sequence assembles itself from the people who will run it.

Backward is the point. Planning forward from today, each trade optimizes its own work and the handoffs fall apart at the seams. Planning backward from the milestone forces every trade to state what it needs from the trade ahead, so the handoffs get negotiated in the room instead of fought over in the field. The output is a sequence the trades built, which means they own it, which means they actually follow it.

Pull planning and trade coordination are the same muscle. The session only works if the right trades are in the room and their scopes are clean enough that the handoffs are real, which is the work covered in the subcontractor management guide. Run the pull plan to set the sequence, then feed the agreed sequence into the look-ahead so the commitments the trades made on the wall become the activities the field is held to.

Clearing constraints and make-ready

A constraint is anything that has to be cleared before an activity can start, and clearing it before the activity is due is called make-ready. The four that stop work most often are an unanswered RFI, an unapproved submittal, material not on site, and the prior trade not finished. An activity with an open constraint is not ready, no matter what the schedule bar says.

The look-ahead exists to surface constraints early enough to clear them. Every activity in the window gets screened: is the information there, is the gear approved and delivered, is the crew and equipment lined up, is the area released by the trade ahead. The Last Planner approach screens against the flows of work, including information, design, materials, labor, equipment, space, prior work, and outside conditions, and anything not cleared becomes a logged constraint with an owner and a date it has to be cleared by. Only activities with every constraint cleared get released into the weekly work plan.

Make-ready is where the schedule connects to the paper that feeds it. The RFI that has to be answered and the submittal that has to be approved before an activity can start are tracked in the RFI and submittal guide, and the long-lead gear that has to arrive is the procurement problem below. The job of the look-ahead is to catch those constraints three to six weeks out, while there is still time to chase the RFI or expedite the delivery, instead of the morning the crew shows up. Clear the constraint before the task is due, or move the task.

Milestones and the gates that organize the schedule

Milestones are the fixed dates the schedule is organized around, the points where a major state is reached and the next phase can begin. On a data center the ones that matter are building dry-in or water-tight, permanent power available, equipment energized at power-on, and the start of commissioning. They are not activities with duration. They are gates.

Milestones give the schedule its checkpoints and the contract its interim dates. A contract often ties money or liquidated damages to milestone dates, not just the final completion, so hitting building water-tight on time can matter as much as the finish. They also force the trades to converge: every activity feeding a milestone has to land before the gate, which is exactly what the pull plan is built backward from.

Watch the milestone that releases the most work behind it. On a data center, power-on is the one, because almost nothing in commissioning can start until the gear is energized, and the whole back end of the job stacks up behind it. A milestone that slips does not slip alone. It drags everything sequenced behind it, which is why the activities feeding a major milestone get the same protection as the critical path.

Sequencing the trades

Sequencing is the order the trades run in, and most of it comes down to one rule: rough-in before cover. Anything that goes inside a wall, above a ceiling, or under a slab has to be in and inspected before the thing that closes it up. Get that backward and you are cutting open finished work to install what should have gone in first.

Overhead is where sequencing is won or lost on a data center. Duct, large pipe, conduit, cable tray, and sprinkler all want the same plenum, and there is an order that lets them all fit: the biggest and least flexible usually goes first, gravity systems that need slope go early, and the trades that can route around get sequenced behind them. A coordinated overhead model settles most of this before anyone is in the air, but the schedule has to hold the order the model set.

Sequencing the trades and coordinating them are the same job from two angles, and the field coordination side is covered in the subcontractor management guide. The schedule sets the order. The coordination meeting holds the trades to it. When a trade wants to jump its slot because its crew happens to be available, the answer is the sequence, not the convenience, because letting one trade work out of order is how the overhead turns into a tear-out.

Long-lead procurement drives the date

Long-lead items are the equipment with order-to-delivery times measured in months, and on a data center they usually sit on the critical path. Switchgear, generators, transformers, UPS systems, and large chillers can run many months from approved submittal to delivery, and the lead times have stretched in recent years. The date that gear arrives often drives the finish more than any field activity does.

Procurement drives the schedule, so the schedule has to drive procurement. Work backward from the date the gear has to be installed, to the date it has to ship, to the date the submittal has to be approved, to the date it has to be submitted, and put those dates in the schedule as activities with the same weight as field work. A switchgear delivery that slips two months is two months on the finish unless you re-sequence around it, and on the gear that feeds power-on you usually cannot.

The front of that chain is the submittal, the discipline covered in the RFI and submittal guide. The gear cannot be ordered until the submittal is approved, so a long-lead submittal that sits in review is a long-lead delivery that slips, which is a finish date that slips. Track the long-lead items on the schedule by their procurement dates, not just their install dates, and treat a late approval on critical gear as the schedule emergency it is.

Update the schedule, or it lies

Update the schedule on a fixed cycle, weekly on most large jobs, by recording actual progress against the plan: what started, what finished, what percent complete each open activity is, and what the remaining duration is. The update sets a new data date, the line that divides what actually happened from what is still forecast, and re-runs the critical path from there.

A stale schedule is a lie, and everyone on the job knows it. A schedule that has not been updated in a month shows activities complete that are not, shows a finish date that is no longer real, and gives the team false comfort while the job drifts behind it. The update is what keeps the schedule honest. It is the difference between a tool that warns you and a wall decoration that reassures you.

Percent complete has to be real, not the office guess. An activity is not fifty percent done because half the planned time elapsed. It is fifty percent done because half the work is in place, and only the field knows which. This is where the update most often goes wrong: the office marks progress from the plan, the field knows the real status, and the two never meet. The fix is to capture progress from the people doing the work, which is the next section.

Capture progress from the field, not the office guess

The schedule is only as honest as the progress feeding it, and the real status lives in the field, not the office. The superintendent walking the deck knows the conduit rough is eighty percent done and stalled on a missing fitting, that the area the ceiling crew needs is still held by the mechanical trade, that the pour the schedule shows complete is actually short one bay. None of that reaches the office unless someone captures it.

Capture progress from the field, with the detail the update needs, or the update is a guess dressed up as data. The most reliable way is to have the field record what moved as it moves, tied to the area and the activity, with a photo where the status is arguable, so the weekly update is assembled from real observations instead of a phone call the night before the meeting.

This is the work FieldOS is built for. The field captures actual progress, percent complete, and the constraint holding an activity, with a photo and tied to the job and the area, so the office updates the schedule from what the field actually saw instead of guessing. Its tradeos workflow keeps the field updates and the look-ahead in one record, so the status the foreman logs on the deck is the status the scheduler uses to re-run the critical path, and the office and the field are finally working from the same picture of the job.

What do you do when the schedule slips?

When the schedule slips, you recover it by re-sequencing, crashing, or fast-tracking, and most real recovery plans use a combination. First find the slip on the critical path, because time recovered anywhere else does not move the finish. Then pick the cheapest tool that buys back the days you need without creating a worse problem behind it.

Re-sequencing is the first move and the cheapest. Reorder the work so a delayed activity is no longer holding others, or shift float-rich work to free up crews for the critical path. It costs nothing but planning. Crashing adds resources to critical activities, more crews, longer shifts, overtime, weekend work, to compress duration, and it costs money, with diminishing returns as a space gets too crowded to add bodies. Fast-tracking runs activities in parallel that were planned in sequence, which buys time but adds risk, because the overlap means starting work before the activity ahead is fully done.

Build the recovery as a real schedule, not a promise. A recovery schedule shows the specific re-sequenced logic, the added resources, and the new dates, so the team and the owner can see how the days come back and whether the plan is credible. A recovery plan that is just a compressed bar chart with no added resources and no changed logic is a wish, and the owner will read it as one. Crash and fast-track the critical path, protect the crews doing the crashed work from burning out, and re-baseline only when the contract allows it.

Document the delay to protect the claim

Document every delay as it happens, because a delay you cannot prove is a delay you absorb. The record has to show what the delay was, who caused it, when it occurred, and how much time it cost the project, captured contemporaneously, while the facts are fresh and the schedule shows the impact. A claim assembled months later from memory loses to a claim built from a clean daily record.

Delays sort into categories that decide who pays, and the contract defines them. An excusable delay is one outside the contractor's control, which may earn a time extension. A compensable delay is an excusable delay the owner is responsible for, which may earn time and money. A non-excusable delay is the contractor's own, which earns neither and may trigger liquidated damages. Concurrent delays, where two causes overlap, are where most disputes live. Which category a given delay falls into is a contract and often a legal question, so the job in the field is to document the facts cleanly and let the contract and counsel sort the entitlement.

The schedule is the evidence. A time impact analysis inserts the delay into the schedule that was current when it happened and shows how it moved the critical path and the finish, which is far more defensible than an after-the-fact argument. That only works if the schedule was being updated all along, which is one more reason a stale schedule is a liability and a current one is protection. Tie each delay to the daily record, the RFI or the change that caused it, and the schedule update that shows the impact.

Weather days and contingency float

Weather days are the lost days you can reasonably expect for the season and the location, and a real schedule carries an allowance for them instead of pretending every day is workable. Build the anticipated weather days into the schedule up front, by season, so a normal run of rain or cold is already accounted for and only abnormal weather becomes a claim.

Contingency float is the schedule's version of money in reserve, time built in to absorb the unknowns that always show up. Where it sits and who controls it is a contract question, and burying hidden contingency in inflated activity durations is the version that backfires, because it hides the real plan and gets spent invisibly. Whether a weather delay is excusable usually turns on whether it exceeded the normal expectation for the period, so track actual weather days against the allowance. That comparison is what the contract and any claim will turn on.

Share the look-ahead with the trades

A schedule nobody sees is a schedule nobody follows. The look-ahead has to get in front of the trades every week, in the coordination meeting and in their hands, so every foreman knows what is coming, what their crew is committed to, and what has to be ready for them. A schedule that lives only with the scheduler controls nothing.

Share the look-ahead as the agenda, not as a report after the fact. The weekly coordination meeting is where the trades commit to the next few weeks of work, raise the constraints they see, and agree the sequence, which is the same meeting the subcontractor management guide builds around. The look-ahead is what that meeting runs on. Send it out before the meeting, work it in the room, and capture the commitments and the new constraints back into it.

Getting the same look-ahead in front of a dozen trades is exactly where a shared field tool pays for itself. FieldOS keeps the current look-ahead, the constraints, and the commitments in one place every foreman can see from the field, so the trades are working from one version instead of a printout that is two weeks stale. When the office re-sequences after a slip, the field sees the new plan the same day, not at the next meeting.

Scheduling tools: P6, MS Project, and the field look-ahead

The scheduling tools split into two layers, and a big job runs both. The master schedule and the critical path live in CPM software, Primavera P6 on most large data center jobs and Microsoft Project on smaller ones, where the logic, the float, and the calculated finish date are maintained. That is the engine. It is not where the field works.

The look-ahead and the pull plan live closer to the crew. Pull planning traditionally runs on a wall of sticky notes, and many teams now run it and the look-ahead on lean planning tools built for short-interval work. The point of the second layer is that the field will not open P6 every morning, so the plan they actually run off has to be in a form and a place they will use.

FieldOS sits at the field layer, where the look-ahead meets the work. The CPM schedule stays the source of the critical path and the finish date, and the look-ahead pulled from it, the constraints, and the field progress live in FieldOS where the foremen are. Its tradeos workflow keeps the field plan and the field updates in one record tied to the job, so the master schedule is fed by real progress and the field is run off a plan it can actually see, instead of the two layers drifting into different versions of the truth.

Metrics that tell you if the plan is real

Measure the schedule with a few numbers that tell you whether the plan is real and whether the job is gaining or losing. The three that earn their keep are percent plan complete, schedule variance, and milestone hit rate.

Percent plan complete, PPC, is the Last Planner metric: of the activities the look-ahead committed to this week, what fraction actually finished. Divide completed commitments by total commitments. A PPC in the fifties means half the plan is fiction and the constraints are not being cleared. Good teams run in the eighties or higher, and the trend matters more than any single week. PPC does not measure how much work got done. It measures whether the plan can be trusted, which is the thing that lets the trades commit to each other.

Schedule variance is where the job sits against the baseline: how many days ahead or behind the data date shows, and on which activities. Milestone hit rate is the count of milestones met on their planned date against those missed, which is the owner's scorecard for whether the interim dates are holding. Track all three over time. A single update is a snapshot. The trend across updates is what tells you the job is recovering or sliding, early enough to act.

Even the small job needs a look-ahead

Even a small job needs a look-ahead, scaled to fit. A two-week tenant fit-out does not need a thousand-activity P6 model, but it needs someone to look three weeks out, confirm the gear is ordered, confirm the inspections are scheduled, and confirm the trade ahead will be done before the next one shows up. The discipline is the same. Only the size of the tool changes.

Scale the schedule to the job, not the other way around. A small job's schedule might be a one-page look-ahead and a short list of long-lead items and milestones, run off a shared board the crew can see. The failure on small jobs is assuming they are too small to drift, then watching them drift anyway because nobody looked ahead far enough to catch the missing material or the inspection that was never called. A look-ahead is cheap. A crew standing around because the work was not ready is not.

Field checklist

What to document

The schedule record is what answers the question later: what was the plan, what actually happened, and why did the finish move. Keep it current and keep it in one place, because a schedule reconstructed after the fact convinces no one, least of all an owner or an arbitrator. The table below is the set of elements worth holding, and the note that makes each one defensible.

Common mistakes

• Running the job with no real schedule, so it drifts and the trades collide.
• Not knowing or protecting the critical path, so attention and resources go to work that has float.
• No short-interval look-ahead, so constraints surface the morning the crew shows up.
• Releasing tasks before the constraints are cleared, so crews mobilize to work that is not ready.
• Letting the schedule go stale, so it shows a finish date and a progress that are no longer true.
• No contemporaneous delay documentation, so the time extension or claim cannot be proven.
• Tracking long-lead gear by install date instead of procurement date, so the order slips unnoticed.
• Building the master schedule in the office without the trades, so the field never follows it.

Standards and references

Project scheduling sits on a method and a contract, not a single code, so cite both and let the contract control. The critical path method is the established practice for building and analyzing a schedule, defining activities, durations, and logic ties, and computing the critical path and the float. Total float and free float are defined terms in that practice, and the contract usually states how float is owned and used.

The look-ahead, pull planning, and the make-ready process come from the Last Planner System, developed in the lean construction community and published by the Lean Construction Institute, with percent plan complete as its reliability metric. The flows used to screen constraints, including information, design, materials, labor, equipment, space, prior work, and outside conditions, come from the same body of practice. These are methods, not mandates, so apply them to fit the job.

The dates, the milestones, the float ownership, the delay and time-extension provisions, and any liquidated damages live in the contract, often on an AIA or ConsensusDocs form, with the project scheduling specification setting the software, the update cycle, and the submission requirements. For data center work, the commissioning sequence and the standards behind it, such as the relevant ASHRAE thermal guidelines and Uptime Institute references, shape the back end of the schedule. Verify the actual requirements against the contract documents and the project specification, because the contract controls the schedule and the method only informs it.

Units, terms, and conversions

Scheduling carries its own vocabulary, and the same idea reads differently across a CPM report, a pull plan, and a contract.

The master schedule is sometimes called the baseline or the CPM schedule. The look-ahead is also called the short-interval schedule or the three-week or six-week look-ahead. Pull planning and the Last Planner System are used almost interchangeably on site. Durations are usually counted in working days, not calendar days, so a ten-day activity can span two weeks once weekends and holidays are out, which is exactly where schedules and expectations part ways.

Critical path

The longest chain of dependent activities; any slip on it moves the finish date

Total float / free float

Slip an activity can take without moving the finish (total) or the next activity (free)

Look-ahead

Rolling 3 to 6 week short-interval schedule pulled from the master into the field

Make-ready / constraint

Clearing what an activity needs, such as an RFI, submittal, material, or prior trade, before it is due

Pull planning / Last Planner

Trades build the sequence backward from a milestone so they own it

PPC

Percent plan complete: committed activities finished divided by total committed

Data date

The line in a schedule update dividing actual progress from remaining forecast

Liquidated damages

A fixed daily amount the contractor owes for late completion under the contract

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is subcontractor management?
• What is an RFI in construction?
• What is data center power density?
• What is lean construction?
• Cast-in vs post-installed anchors: which should I use?