Every job you’ve ever run had a critical path. You’ve just been finding it the hard way, after the fact, working backward from a finish date that moved.
Think about the last two delays on a job. One of them slipped three days and nothing happened; you absorbed it and the client never knew. The other slipped two days and somehow cost you a week and a half. Same size slip. Wildly different outcome.
That difference is the critical path. One of those tasks was on it. The other wasn’t.
What the critical path actually is
Strip away the terminology and it’s this: on any job, there’s a chain of work where each task can’t start until the one before it finishes, and that chain is the longest one on the project. Add up its durations and you get your finish date.
Everything on that chain is critical, in a specific and narrow sense: if it moves, the finish date moves with it. One-for-one. Lose two days on a critical task and the job ends two days later.
Everything not on that chain has slack. The industry word is float. Float is how many days a task can slip before it starts pushing something else. A task with six days of float can slip five days and cost you nothing at all.
That’s the whole concept. Critical path method, CPM, the thing that sounds like it requires a certification, is just the arithmetic that identifies which tasks have zero float.
Walking it on a real job
Take a kitchen and bath remodel. Twelve weeks, a normal sequence:
Demo. Rough plumbing and rough electrical. Rough inspection. Insulation. Drywall. Paint. Tile. Cabinets. Countertop template. Countertop install. Fixtures and trim. Final inspection.
Now ask one question of every task: what has to be finished before this can start?
Not “what happens before it on the calendar.” What genuinely blocks it. That distinction is where most schedules go soft. Plenty of tasks sit next to each other on a Gantt chart with no real relationship at all, and plenty of others have a hard dependency nobody ever wrote down.
Working through it:
- Rough plumbing and electrical both need demo done. They don’t need each other, so they can run at the same time, and usually should.
- Rough inspection needs both roughs complete. It’s a merge point.
- Insulation, drywall, and paint form a straight chain off the inspection.
- Tile needs paint done, or at least the ceilings and first coats.
- Cabinets need floors in and walls painted.
- The countertop template needs cabinets set. This is the one that gets people.
- Countertop install is two to three weeks after template, and that gap is a fabricator’s calendar, not your crew’s.
- Fixtures and trim need counters in.
- Final inspection needs everything.
Now trace the longest chain from start to finish. On a job like this it runs: demo, roughs, rough inspection, insulation, drywall, paint, cabinets, template, fabrication, counters, fixtures, final.
That’s the critical path.
And notice what’s not on it. Tile. Tile is a big-ticket, visible, expensive scope that everybody stresses about, and on this job it has float. As long as it’s done before final, the tile guy’s exact week doesn’t move your finish date. Meanwhile the countertop template, which takes about twenty minutes and costs you nothing, is on the path. Slip the template by four days and your job ends four days later, because the fabricator’s two-week clock doesn’t start until somebody shows up with a laser.
That’s the value of doing this once, honestly, on a real job. It tells you which twenty-minute tasks deserve a phone call and which forty-thousand-dollar scopes can wait a week.
How to calculate it by hand, once
Do this once so you trust it. Then never do it again.
Forward pass. Start at day zero. For each task, its earliest start is the latest finish of everything it depends on. Walk left to right through the whole job. The finish date you land on is the project duration.
Backward pass. Start from that finish date and work right to left. For each task, its latest finish is the earliest start of whatever depends on it, minus any lag.
Float. For each task, subtract the earliest start from the latest start. That’s your float.
The path. Every task with zero float is on the critical path.
On a forty-activity remodel that’s a solid forty-five minutes with a pad of paper. Which is exactly why almost no GC running six jobs does it, and why CPM has a reputation as a big-commercial thing, the domain of a full-time scheduler with a Primavera license and a plotter.
But look at what you just did. You added and subtracted numbers. There was no judgment call anywhere in it. Given the tasks, the durations, and the dependencies, the answer falls out the same way every time, for everyone.
That is precisely the kind of work that should be happening in software while you drive between sites.
The mistake: treating the path as a label
Here’s where most tools get it wrong, and it’s worth understanding before you evaluate any of them.
A lot of scheduling software lets you flag a task as critical. A checkbox, a color, a priority field. It feels useful. It is actively harmful.
The moment a human marks a task as critical, that mark is a snapshot of one person’s opinion on one particular day. Then the job moves, as jobs do, and the flag doesn’t. Three weeks in you’re looking at a schedule where four tasks are marked critical and only one still is, and two tasks that are genuinely critical now aren’t marked at all. You now trust a field that’s lying to you, which is worse than having no field.
The critical path is not an attribute of a task. It’s a result of the network. It changes every time a date changes, and it has to be recomputed, not remembered. It isn’t something you assign. It’s a reading you take.
What this changes on a Tuesday
The practical payoff isn’t a prettier chart. It’s three decisions that get easier.
Which delays to fight. When your framer texts that he’s a day behind, the first question isn’t “how bad is this.” It’s “is framing on the path.” If yes, you make calls today. If it has five days of float, you say “no problem” and mean it, and you save your leverage with that sub for the day you actually need it.
Where to spend money to compress. Overtime, a second crew, expedited material. Those only buy you a shorter job if you spend them on the path. Money spent accelerating a task with float buys you nothing but a task that finishes early and then waits. This is the single most common way GCs waste money trying to recover a schedule.
What to tell the client, and when. A delay on a task with float is information. A delay on the critical path is a new finish date. Confusing the two is how you end up either crying wolf or delivering a two-week surprise on day nine.
That third one compounds. We’ve written before about what a schedule delay actually costs once you price the general conditions burn and the next job that couldn’t start, and nearly all of that damage traces back to not knowing, in the moment, whether the thing that just moved was load-bearing.
What Relay does about this
We built the checkbox version early in Relay and pulled it out for exactly this reason. Relay computes the critical path from the dependency network, live. Drag a bar and it recalculates, including whether the task you just moved was even on the path in the first place. That’s the difference between “framing slipped two days” and “framing slipped two days and it cost us nothing, we had four days of float there.” Two very different phone calls to the client, and you know which one you’re making before you dial.
Start with the dependencies, not the dates
If you take one thing from this: the critical path is downstream of your dependencies. You cannot compute it, and no software can compute it for you, if the schedule doesn’t record what depends on what.
That’s the actual work, and it’s a one-time job per project type. Sit down with your standard remodel sequence and answer that one question, what has to finish before this can start, for every line. Not what comes before it on the calendar. What blocks it.
Do that once and you have something that behaves like a model instead of a drawing. Move one bar and the rest tells you the truth. Skip it, and you have a very nice picture of a plan that stopped being accurate in week two.
If you’re building that sequence for the first time, our walkthrough on putting together a construction schedule covers the setup end of it.
Worth doing before Q4. With architecture billings under the growth line for 41 straight months according to the AIA and backlogs thinning at the small end of the market, the margin on the job you’re standing on is the margin you get. There isn’t a stack of work behind it to make it up on.
Relay computes your critical path from your dependencies, live, every time a date moves. Try it free for 30 days on the Complete plan at relayconstruct.com. Card up front, $199.99/month when the trial converts.