Construction Scheduling Services and CPM Schedules That Get Accepted
We build the CPM schedule your contract actually asks for: baseline, monthly updates with a narrative, look-aheads your superintendent will use, and fragnets when the job slips. Native P6, Microsoft Project or Asta files come back with a readable PDF Gantt.
What ships with every schedule
- The native file, XER, XML, MPP or Asta, yours to update
- A readable PDF Gantt, and a plot-size layout on request
- Basis of schedule narrative: crews, rates, calendars
- Two-step check: a self-check, then an independent second pass
- A clause by clause spec matrix with every baseline
The submittal ladder, from the proposal schedule to baseline schedule development to as-built
Contractors talk about the schedule as if there is one. Your contract names five or six, each with its own due date and acceptance rule, and CPM scheduling services that ignore the difference collect comments on all of them.
Before award, and the first three weeks after NTP
Two documents that are not the baseline, and both of which get treated like it.
The proposal or bid schedule
PRE-AWARDA logic sketch, twenty to sixty activities, showing a selection committee you understand the sequence and can hold the duration. Nobody accepts it as a baseline and no update runs against it.
Outsourced bid management servicesContract time determination
PUBLIC WORKOn most public work the owner fixed the duration before the solicitation went out, from a production rate analysis on the controlling items. If it is short, say so in the pre-bid question period, not in an update eighteen months later.
The preliminary schedule
NTP TO NTP+21Due within days of notice to proceed, covering the first 60 to 90 days in detail plus the milestone spine, so your first pay application is not hostage to a baseline still in review.
Baseline schedule development, and the update chain behind it
The contract document, and the only two events allowed to replace it.
The initial or baseline schedule
THE CONTRACT DOCUMENTFull logic, every activity coded, calendars set, durations defended in the narrative, typically due 30 to 60 days after NTP. Most specifications say the owner accepts it without adopting your means and methods and without waiving anything: acceptance is not approval.
What we will not promise you
That your owner accepts it first pass. Nobody honest promises that. What we commit to: comments on scope we scheduled get a resubmittal at no charge, and you get in writing which are scope changes and which are ours.
Periodic updates
MONTHLY, USUALLYActual starts, actual finishes, remaining durations and a data date, on the cycle the specification names. Each is a dated snapshot, and the snapshots are the evidence.
See the monthly cycleThe revised baseline
WHEN THE PLAN CHANGESNot a monthly event. It belongs to a change in the work, an accepted time extension, or a sequence that is dead. Re-baselining to make slippage disappear erases the variance record you will want later.
As-built schedule preparation, and the ones outside the contract
Two documents rarely written into the specification and eventually asked for anyway.
The as-built schedule
AT CLOSEOUTEvery activity carrying its real start and finish. Assembled from the update chain if it was kept, reconstructed from daily reports and pay applications if it was not. The second route is the expensive one, and it is the one an owner's consultant gets to pick apart.
A schedule for a permit or a lender draw
OUTSIDE THE CONTRACTA phasing plan for a right of way permit, a shutdown window for a hospital infection control submission, or a schedule a lender wants tied to real durations instead of twelve equal draws.
Why owners reject baseline schedules, item by item
Owners and their schedule consultants send baselines back for a short list of reasons, and it is the same list every time. Here it is, in the order a reviewer works through it.
Baseline rejected by your owner? Send the file before you resubmit
Send the schedule and the comment log. You get the metric counts, which comments are defects and which are scope, and a straight answer on what a resubmittal takes.
What we need from you before a single activity is typed
A scheduler who starts typing before this list is answered produces a bar chart, not a schedule. Here is what we ask for, what happens when a piece does not exist yet, and the jobs where the answer is that you should not hire us.
Start with the paper. The schedule specification section, whether it is 01 32 16 or an agency number, sets the software, the submission dates, the coding, the level of detail, the narrative content and the float ownership clause. Add the contract time, the NTP date, the milestones and the liquidated damages exposure. Half the rejection list disappears when somebody reads that section first.
Then the work. The current drawing set with its issue date, the specifications, and the quantities if you hold them. If you do not, we measure the controlling items first, because a duration derived from a quantity survives an owner challenge and one copied off the last job does not. Add the schedule of values if payment will run off this file.
Then the person who is going to build it. We book a working call with your superintendent, an hour and usually closer to two, taking sequencing, crew sizes, planned shifts, laydown and crane constraints, and the shutdown windows nobody wrote down. What they tell us goes into the basis narrative attributed to them, so when a reviewer asks where the masonry production rate came from, the answer has a name against it.
Some jobs should not come to us at all. A ten week fit-out with fourteen activities and an owner who wants a bar chart does not need a CPM desk, and a contractor already paying a scheduler who can absorb one more job is buying the same file twice. If your specification names a credentialed scheduler by title, read that clause before you buy: nobody here holds a PSP, a PMI-SP or a PE, and no deliverable of ours satisfies a clause that requires one.
Logic, lags and constraints, and the ones reviewers flag
Relationships and constraints are where a schedule is either honest or quietly rigged, and a reviewer tells which inside ten minutes because the tells are countable.
The relationship types, and the two that get abused
Finish to start is the default for a reason. Each of the others solves one problem and creates another.
Finish to start
THE DEFAULTOne thing ends, the next begins. Most specifications want 90% or more of relationships to be FS, because it is the only type where the driving logic is obvious to somebody reading the bar chart.
Start to start and finish to finish
USE WITH A REASONSS with a lag models crews moving through a building behind one another. FF ties punch work to substantial completion. An activity whose only successor is tied start to start has nothing hanging off its finish, so it can run three months late and push nothing. That is a dangling finish, and it is an open end even though the activity has logic on it.
Reviewers count the dangling ones
Missing successors are the first thing an automated screen reports. Start to finish, the fourth type, is legitimate about once in a thousand relationships: a file with forty was drawn by dragging arrows.
Lag, lead, and why negative lag reads as manipulation
A lag is elapsed time with no responsible party attached. Use it for cure, never for work.
Legitimate lag
CURE AND WAITConcrete cure before stripping, mortar set, paint dry, a 48 hour pressure test hold. Time passing with nobody working, where there is nothing else to track.
Lag hiding an activity
FLAGGED EVERY TIMEA 20 day lag between shop drawing submission and fabrication is not cure time. It is a review activity with a responsible party and a contract duration, and buried in a lag a slow review never reaches the driving path.
Negative lag, also called a lead
USUALLY PROHIBITEDA lead lets a successor start before its predecessor finishes without saying which part of the predecessor is complete. Most agency specifications ban it, and the fix is always the same: split the predecessor into the two pieces the lead pretended existed.
Constraints, and what a Mandatory Finish does to your float
Soft constraints inform the calculation. Hard constraints sit on top of it and overrule it.
Soft constraints
ACCEPTABLEStart On or After, for a date you cannot beat: a permit issue, an owner furnished delivery, a seasonal restriction. The calculation runs through it and float computes honestly on both sides.
Hard constraints
THEY BREAK THE MATHMandatory Start and Mandatory Finish override the network instead of living inside it. A Mandatory Finish on substantial completion pins that date whatever its predecessors do, so the calculated finish stops moving and the delay strands itself as negative float on the activities behind the constraint. Use a Finish On or Before milestone instead, which lets the completion date go negative where a reviewer can read it.
Out of sequence progress
THE SETTING NOBODY CHECKSWhen work starts before its predecessor finishes, retained logic makes the remaining portion wait and progress override ignores the relationship, so the same file calculates two completion dates. We declare the setting in every narrative and never switch it mid-job.
Who owns the float, in the schedule we build
Float is a project asset three parties each believe belongs to them, and your contract has a clause about it that nobody on your team has read.
Who owns the float in a construction schedule?
Whoever your contract says. Three positions exist: contractor owned, owner owned, and project owned on a first come first served basis. Where the contract is silent, US practice treats float as a shared project resource, so the party that needs it first is the party that gets to spend it. Find the clause before the baseline is built.
What is total float versus free float?
Total float is how many days an activity can slip before the project completion date moves. Free float is how many before its own successor is pushed. An activity with thirty days of total float and none of free float will not delay the job, and it delays the next trade on day one.
What does negative float actually mean?
That the calculated finish is later than the date you are contractually held to, by that many days. It is not an error. It is the file telling you the plan misses the date, and an update carrying honest negative float is the strongest support a time extension request can have.
Can I protect float without hiding it?
Yes, and the visible method is the only one that survives review. Put the buffer where it belongs, as a named, coded activity ahead of the milestone it protects. What does not survive is padded durations or dropped-in constraints, because a reviewer counts both and it costs you credibility on everything else.
Where a defensible duration comes from, quantities, crews and production rates
Ask three schedulers where a 15 day duration came from and two will answer experience. Here is the arithmetic that answers an owner challenge, and where each input comes from.
Take the quantity off the drawings, not off the last job
Name the crew you will actually field
Apply a production rate you can name the source of
Convert working days on the right calendar
Reconcile the total hours against the priced estimate
Resource and cost loading, tied to the schedule of values
A resource loaded schedule in Primavera P6 and a cost loaded schedule driven off your schedule of values are two jobs on one file. This is what each carries and what a reviewer does with it.
| Labor hours | Crew hours per activity, off the estimate | Manpower histogram and leveling | A 42 man week you cannot staff |
|---|---|---|---|
| Crew count | The superintendent's real crews | Checking the curve is buildable | Six crews planned, three exist |
| Activity cost | Your schedule of values, split down | The basis of the monthly pay app | SOV lines that map to no activity |
| Mobilization | Schedule of values, line one | Watched closely for front loading | Cash pulled forward into month one |
| Physical percent | Installed quantity over total quantity | Earned value and the S-curve | Percent read off the calendar instead |
| General conditions | Duration driven, level of effort | Extends when the job extends | Loaded as a fixed lump sum |
| Owner furnished items | The procurement log | Their long lead risk, tracked | Left out, then charged against you |
Procurement, submittals and long lead equipment
More jobs miss substantial completion over a switchgear lead time than over a slow trade. Procurement belongs inside the CPM as activities with durations, not in a spreadsheet beside it.
One piece of equipment is five activities, not one
Modeled as one bar reading procure switchgear, a two week review overrun and a twelve week fabrication overrun look identical, and neither tells you where to push.
The review duration is in your contract, so schedule that number
Most specifications give the architect 14 or 21 calendar days per submittal and a defined duration for a resubmittal. Schedule the contract number, not the one you are hoping for. When a review runs long, the update shows it against an agreed duration.
A long lead item tracking schedule your PM can actually run
A register whose need-by dates are driven backwards from the installation activity in the CPM, not typed independently. Every line carries the required-on-site date, the promised date, the float between them and whoever is chasing it.
Switchgear, air handlers, elevators and curtain wall
The equipment that decides commercial schedules: distribution gear, air handlers and chillers, elevators, curtain wall, generators, and anything with a custom fabrication drawing. On a core and shell these are the activities the sequence gets built around.
Commercial construction estimating servicesOwner furnished equipment and the date you never controlled
Owner furnished, contractor installed equipment belongs in the file as a delivery milestone with the owner named in the responsibility code. When their gear lands eleven weeks late, the update already shows whose activity it was.
The monthly update cycle, timed to your pay application
The update is not administration. It is the only contemporaneous record of what happened and when, and it is what a time extension request gets built out of eighteen months later.
Day 20 to 22: the data date is set and published
Day 22 to 24: field status comes back on one form
Day 24 to 26: the file is statused and recalculated
Day 26: the monthly schedule update and narrative report
Day 27 to 28: the independent check, then it ships
Ask a scheduler about your spec before you commit to anything
Float ownership, a submission date you cannot make, whether the spec really requires a cost loaded file. Put it to a scheduler and get a direct answer.
Three and six week look-aheads, pulled from the CPM, not drawn beside it
The commonest failure here is not a bad baseline. It is a field team running a whiteboard with no connection to the file the owner is sent every month.
Time impact and windows analysis, and what the updates have to carry
Time impact analysis is not a document written after the fact. It is a calculation run on a schedule already being maintained, which is why the update cycle matters more than the method.
The four methods, and when each one is the right one
AACE 29R-03 sorts these into observational and modeled. Which you can use is decided by what your records support.
As-planned versus as-built
OBSERVATIONALThe baseline laid against what actually happened, period by period. Cheap and readable by non-schedulers, weak wherever there is concurrency, because it shows that dates moved without showing what moved them.
Windows or contemporaneous period analysis
OBSERVATIONALThe job cut into periods at the monthly updates, with the driving path recalculated inside each one. The method most reviewers prefer, and it runs on your own updates: if nine were never produced, it is unavailable to you.
Time impact analysis by fragnet insertion
MODELED, PROSPECTIVEA fragnet, the small network representing the change, inserted into the accepted update immediately before the event and recalculated. The movement in the completion date is the time impact, which is why specifications want it attached to the request.
Collapsed as-built
MODELED, RETROSPECTIVEDelay activities stripped out of the as-built and the network recalculated to show what would have happened without them. Expensive, heavily contested, credible only on a well documented as-built.
Excusable, compensable, and the concurrency problem
Three categories, and the single argument that collapses most contractor claims.
The three categories
ENTITLEMENTExcusable and compensable gets time and money. Excusable and non-compensable, unusually severe weather being the example, gets time only. Non-excusable gets neither and leaves liquidated damages running. Cause decides, documented as it happened.
Concurrent delay
WHERE CLAIMS DIETwo delays, one yours and one theirs, driving the completion date across the same window. The usual outcome is time without money. Whether they were truly concurrent turns on both sitting on the driving path at once.
Pacing
THE DISTINCTIONSlowing your own work in response to a delay you did not cause is not concurrent delay, provided you can show a deliberate decision and the ability to resume. Undocumented, it looks identical to culpable delay.
Schedule delay analysis services, and the line we do not cross
What this desk does, plainly, including the part we will not be doing for you.
What we prepare
DOCUMENTATION SUPPORTThe fragnet and its insertion at the correct data date. The windows analysis off your update chain. The as-built reconstruction. The exhibits and narrative chronology, in a form your counsel can work from.
What we do not do
THE BOUNDARYWe do not testify. We do not issue sealed opinions, and nobody here holds a professional engineering licence to seal one with. We do not opine on entitlement, interpret your contract, or choose your means and methods.
Why the support is still worth buying
WHAT IT SAVESOnly while the job is running. If the update chain was kept, the analysis is arithmetic on documents you already own. If it was not, somebody rebuilds four years of history first at expert rates, and what they produce is still weaker evidence than the updates you did not run.
Recovery and acceleration, including the trap nobody mentions
You have had the letter. Fourteen days to submit a recovery schedule showing completion by the contract date. What you do next can cost you a time extension you were entitled to.
There are three ways to recover time and they are not interchangeable. Crashing adds crews, shifts or overtime to activities already on the driving path, buying days at a cost per day that climbs steeply. Fast-tracking overlaps activities that were sequential, buying days and rework risk. Re-sequencing changes the order of the work, which is frequently free and is the option nobody spent an afternoon on.
All three hit the same wall, which is trade stacking. Put four crews into a space sized for two and productivity falls for all four, so a recovery assuming full output from stacked crews is a plan to be late twice. A defensible one applies a productivity loss factor to the stacked windows and says so, because a reviewer reads its absence as optimism.
Now the part rarely said out loud. If you submit a recovery schedule showing completion by the original contract date while a time extension request is live for a delay that was not yours, you have produced a document arguing the date was achievable all along. Owners have used a contractor's own recovery schedule to defeat that contractor's claim. Refusing carries its own consequences, so state in the transmittal what the recovery assumes, what it costs, and that it goes in without prejudice to the pending request.
The other distinction worth putting in writing is directed acceleration against constructive acceleration. Directed is when the owner instructs you to accelerate, and it is normally compensable. Constructive is when the owner refuses a time extension you were entitled to and holds the date anyway, leaving you to accelerate at your own cost. It is arguable only where the request, the refusal and the cost were documented as they happened.
The schedule health check, run before anything reaches your owner
This is the metric list a reviewer or a third party schedule consultant runs against your file, modelled on the DCMA 14 point schedule assessment. We publish it, and we run it on your current schedule for nothing.
Hire a fractional construction scheduler instead of carrying a seat
A monthly partnership runs every job's update cycle on a priority queue, revisions included. A dedicated seat is eight hours a day, five days a week, on your projects only.
Agency schedule specifications, and the compliance matrix that answers them
Federal and DOT schedule specifications prescribe coding, level of detail, submission format and payment linkage, not a bar chart. Every baseline built to one of them ships with a clause by clause matrix mapping the spec to the deliverable.
| UFGS 01 32 01.00 10 | A P6 network, prescribed coding, narrative, monthly update | Native XER with the coding dictionary applied |
|---|---|---|
| SDEF export | A fixed format activity file the agency loads on its side | SDEF exported from P6 with the coded fields populated |
| Activity coding dictionary | Responsibility, feature of work, area, phase, bid item | Codes applied while building, not retrofitted after review |
| Level of detail | Activity durations capped, often at 20 working days | Durations sized to the update period, listed in narrative |
| Cost loading | Every activity carrying value, summing to the contract | Loaded off your schedule of values, reconciled to the total |
| Float ownership clause | Usually states float belongs to the project, not to you | Sequencing built knowing that, and flagged at intake |
| State DOT CPM provisions | Bid item numbers as activity codes, letting specific dates | Activity codes mapped to your own bid item numbers |
| Submission format | Native file, PDF, narrative, sometimes a printed plot | All four, on the submission date the spec fixes |
| P6 against MS Project | Agency specs name P6; most private owners accept MPP | Built in the tool your owner will accept, named at quote |
Scheduling questions, answered straight
What contractors ask a construction schedule consultant before they send anything over.
Cost and timing
Float and logic
When it goes wrong
Files and specifications
Hire a CPM scheduler
for one schedule or for every job
Send the schedule specification section, your contract dates and the current drawing set. A scheduler reads it, quotes it in writing with a delivery date, and nothing is invoiced until you approve that number.