The crew you actually have.

So this week we do the checking. And then we deal with the two things that always follow: not enough people, and not enough time.

Load it, then look at it

Take your schedule and, activity by activity, put the crew on it. Six carpenters here. Four steel fixers there. The crane, three days.

Then add it up by week and draw it.

Almost every first draft looks like the left-hand chart. Mountains and valleys. Week four wants twenty men. Week seven wants four. You have twelve, every week, all year.

That spike isn't a scheduling detail. It's the week the job quietly falls apart — because on the day it arrives, twenty men do not appear. Eight activities get half-manned instead, everything slips a bit, and nobody can point to the reason.

Leveling: spend the float, not the time

Here's where last week's maths starts paying rent.

Some of the work in that spike is on the critical path. You can't touch it. But some of it isn't — and non-critical work has float. Room to move without hurting the end date.

So you move it. Slide the backfill into the following week. Push the non-urgent work into its float window. The mountain flattens, the valley fills, and the finish date doesn't move a day.

That's resource leveling: solving a manpower problem by spending float instead of money. It's the cheapest fix in project controls, and it's sitting in your schedule right now, unused.

Which reframes what float actually is. It isn't spare time. It isn't a buffer to be proud of. Float is the room you have to make the plan buildable. Spend it deliberately — or the project will spend it for you, badly.

When the float runs out

Sometimes you slide everything you can and the histogram still breaks the line. Now the software has nothing left to offer you.

This is the moment planners either become professionals or become invisible.

There are exactly three honest answers, and every one of them belongs to someone above you.

Get more crew. Subcontract the work, bring in a second gang. It costs money, and someone has to approve that.

Cut the scope. Hand over in phases; defer what can genuinely wait. That's a conversation with the client.

Move the date. Not comfortable. Far more comfortable now, in writing, with numbers — than in eight months, with excuses.

There is no fourth option. And what most people do instead is the thing that isn't on the list: they quietly type a shorter duration and let the plan look fine. That isn't a solution. That's a delay you haven't reported yet.

Too long. Now what?

The other problem: the network calculates a finish, and it lands after the contract date.

You have two real tools, and we met them back in Week 7. Crash it — buy time with money: extra crews, overtime, a night shift. Or fast-track it — buy time with risk: overlap work that was meant to run in sequence.

But there's a rule about crashing that almost everyone breaks the first time.

You can only buy time on the critical path.

Pay overtime on the backfill — which has four days of float — and you will spend real money, work real weekends, and finish the project on precisely the same day. The activity was never controlling the end date. You just bought a receipt.

It sounds obvious written down. It happens constantly, because the backfill is the thing that looks late, and the critical path is boring foundation work that nobody is shouting about.

Manage the chain that controls the date. Not the one that's making the most noise.

And one more thing about crashing: it has a floor. Doubling the gang doesn't halve the pour, and it never will — the concrete still cures in seven days no matter how many people stand around it. Past a point, you're just adding cost to a duration that has stopped listening.

Some work responds to more people. Some work simply doesn't. Knowing the difference is what separates a planner from someone who moves bars around until the picture looks acceptable.

Practical insight

Load your current schedule with crews and draw the histogram. Just once. Most planners never have.

Then ask: which week is the worst? Is any of that work non-critical — can I slide it into float? And if I've spent all the float and it still doesn't fit, which of the three conversations am I having, and when?

The histogram tells you the truth about your plan long before the site does. It just does it while there's still time to act.

Key takeaways

✔ CPM assumes unlimited resources — every date is provisional until you load the crews.
✔ Draw the histogram: spikes are the weeks the job quietly falls apart.
✔ Leveling slides non-critical work into its float — same work, same finish date, no extra cost.
✔ Float isn't spare time; it's the room you have to make the plan buildable.
✔ When float runs out there are three honest answers: more crew, less scope, more time.
✔ Typing a shorter duration is not an answer — it's an unreported delay.
✔ Crashing only buys time on the critical path; money spent off it buys nothing.
✔ Some work ignores extra people entirely — concrete cures on its own schedule.

What's coming next

We now have a schedule that's logical, estimated, and actually buildable with the crews we have.

And it's still a single number pretending to be certain. Next week we admit what every experienced engineer already knows — that durations are guesses with error bars — and we do something rigorous about it: schedule risk analysis, Monte Carlo, and how to work out what your contingency should actually be instead of adding 10% because it feels about right.

One date is an opinion. A range is an answer.