muqawil · مقاول
FeaturesHow it worksPricingGuideBlogPartners
العربيةSign inGet started
  1. Home
  2. /
  3. Blog
  4. /
  5. Critical Path: Why Schedules Slip

Scheduling

The critical path, total float, and why a construction schedule slips unnoticed

Published: 10 August 202612 min read

Every project starts with a schedule and very few finish on one. The usual explanation points at the site — a late supplier, weather, a permit that took three weeks longer than promised. But the real damage shows up somewhere else: in a schedule that has no way to translate a two-week delay on one activity into a new handover date. The activities sit side by side with no logic between them, so the finish date on paper holds perfectly still while the reality underneath it moves.

Key takeaways

  • A schedule is a network of logic, not a list of dates: if moving one activity moves nothing else, what you have is a drawing of a schedule.
  • The critical path is a calculated result, not a manual label — a forward pass gives early dates, a backward pass gives late dates, and the gap between them is total float. Zero float is what makes an activity critical.
  • The number that runs your week is float, not the critical flag: an activity with 12 days of float can slip 11 days for free, and becomes the problem on day 13.
  • A schedule that calls every activity critical carries no information at all — criticality that includes everyone is identical to no criticality.
  • Without a saved baseline you cannot prove what slipped: the live schedule rewrites itself with every update, and an extension-of-time claim is built from the gap between an approved plan and a documented reality.

In this article

  1. 01A bar chart is not a schedule
  2. 02The four dependency types, and the lag everyone forgets
  3. 03The critical path is a calculation, not a label
  4. 04When everything is critical, nothing is
  5. 05The baseline is the only way to prove what slipped
  6. 06A schedule nobody updates is worse than no schedule
  7. 07How this works in muqawil

A bar chart is not a schedule

Most site schedules begin life as one bar per activity, with a start and a finish typed in by hand. It looks convincing — coloured bars, phases, a handover date at the end of the page. Then excavation runs two weeks late, its bar gets dragged, and nothing else moves: not the foundations, not the structure, not the handover date. The file has no idea the activities are connected, because nobody ever told it they were.

Note: The test: move one bar

Push the first activity on your project two weeks later and look at what changed. If the handover date sits exactly where it was, the schedule carries no internal logic — it is a record of an intention, not a planning tool.

What turns the drawing into a schedule is the logic: a statement that this activity cannot start until that one finishes, and that a third can run alongside both. Once that logic is recorded, the finish date is derived from the network rather than typed beside it — and that alone is enough to make the date move on its own when something early slips.

The four dependency types, and the lag everyone forgets

Finish-to-Start is the one everybody knows: painting does not start until plastering is done. Relying on it alone forces the planner to sequence everything end to end, while half the work on a real site overlaps. The four types exist because they describe what actually happens out there:

The four dependency types and what each means on site
TypeConstraintSite example
Finish-to-Start (FS)The successor cannot start until the predecessor finishesFoundations cannot pour until excavation is complete
Start-to-Start (SS)The successor cannot start until the predecessor startsPipe rough-in starts two days after blockwork starts
Finish-to-Finish (FF)The successor cannot finish until the predecessor finishesElectrical inspection cannot close before the wiring does
Start-to-Finish (SF)The successor cannot finish until the predecessor startsTemporary power cannot end before permanent power is energised

Then there is lag: two days, or seven, between two activities where nobody works but the wait is mandatory. Seven days of concrete curing is not an activity you assign to anyone, yet it is seven real days in the schedule — recorded as a lag on the relationship itself. A negative value is a lead: start the electrical rough-in three days before blockwork finishes.

Warning: A lead is a promise about overlap, not a shortcut

Pulling an activity three days ahead of its predecessor means two crews will share the same space for three days. If the site cannot actually absorb that, the lead did not save three days — it hid a clash that will reappear later as a delay with no recorded cause.

The critical path is a calculation, not a label

Once the relationships are recorded, the critical path stops being an opinion. It is computed with two passes over the network, and the output is a number per activity:

  1. 1

    Forward pass: the earliest anything can happen

    Working from the project start, each activity gets an early start (ES) — the largest constraint its predecessors impose — and an early finish (EF = ES + duration).

  2. 2

    Backward pass: the latest it may happen

    Working back from the project finish, each activity gets a late finish (LF) — the latest it can end without delaying the project — and a late start (LS = LF − duration).

  3. 3

    Total float: the gap between them

    Total float = LS − ES. It is the number of days the activity can slip before the handover date moves at all.

  4. 4

    Critical: whatever has zero float

    Activities with zero total float form the critical path — one day late on any of them is one day late on the project.

Example: four activities, separated by float rather than by size
ActivityDurationTotal floatWhat it means this week
Excavation10 days0Critical — a day lost is a day lost on handover
Foundations14 days0Critical — no margin whatsoever
Boundary wall6 days12 daysCan slip 12 days for free; on day 13 it becomes critical
Guard house4 days28 daysWide margin — safe to defer and release a crew onto the critical path

The number a project manager actually runs their week on is float, not the critical flag. The boundary wall with 12 days of float is not a problem today, and is a problem on day 13 — a schedule that shows float tells you exactly when, while a schedule that shows a binary label says "no" right up until it suddenly says "yes".

When everything is critical, nothing is

There is a failure mode in scheduling tools worth naming, because it passes unnoticed: flagging every activity that has a due date as being on the critical path. The screen looks entirely correct — red bars, warnings, a list of critical activities — and carries zero information. A manager looking at forty critical activities cannot prioritise between them, so they fall back on ranking by instinct, which is precisely what the schedule was supposed to replace.

Note: We shipped this bug ourselves

The first version of our own schedule view ran a forward pass only and counted every dated task as critical. It did not look broken — it looked strict. Replacing it with the full calculation (forward pass, backward pass, float) made criticality mean what it says: genuine membership of a zero-float path.

On a project whose logic is built properly, the critical path is a minority of the activities — commonly 10% to 20% of them. If your schedule says otherwise, either the network is strung end to end with no real parallel work in it, or the tool is not computing float at all.

The baseline is the only way to prove what slipped

A live schedule rewrites itself continuously: dates get updated, activities get added, durations get revised. Six months in you have exactly one schedule, perfectly consistent with reality, and no idea what it looked like on the day it was approved. That is where most extension-of-time claims die before they start — a claim is built on the gap between an approved plan and a documented reality, and if the plan has been rewritten there is no gap left to measure.

A baseline is a frozen snapshot of the schedule at a moment in time: every activity with the dates and percent complete it had then. The schedule keeps moving afterwards; the snapshot does not. Every later view can then put "planned" and "actual" on the same row.

Tip: Take the baseline at approval, not during the crisis

A baseline captured two months into trouble documents the trouble as the plan. The right moment is when the owner or the consultant approves the programme — before work starts, not once you need to prove something.

That difference — baseline date against current date, activity by activity — is the raw material for both an extension-of-time claim and the schedule performance index (SPI) in earned value analysis. The two read the same number from opposite ends.

A schedule nobody updates is worse than no schedule

An abandoned schedule does not stay neutral, it turns misleading. A manager reading a three-week-old percent complete makes a decision about a situation that no longer exists, which is worse than reading nothing and phoning the site. So the practical question in scheduling is never "which planning tool?" — it is "who enters progress, when, and at what cost to their day?".

  • Site engineer or foreman: percent complete on the activities they run, from a phone, at the end of the day.
  • The daily report: weather, crews, equipment and obstructions — the record that later explains why a specific activity slipped.
  • Project manager: the logic, the durations and the baseline. Those are planning decisions, not data entry.
  • Finance: actual cost against planned, because schedule slippage and cost overrun usually surface together.

One operational condition governs all of it: the entry has to work with no signal. An update that needs a stable connection does not happen on the third floor of a concrete frame — it gets deferred to the office, then to tomorrow, then written from memory on Thursday.

How this works in muqawil

Scheduling in muqawil is built on the network rather than the bars: tasks carry a start and finish or a duration in days, they get linked by dependencies, and the critical path is computed from the network itself every time the view is opened.

  • All four dependency types (FS, SS, FF, SF), each with a lag — or a negative lead — in days.
  • A full critical-path calculation: forward pass, backward pass, total float, and criticality meaning zero float rather than "has a date".
  • Network-computed dates shown alongside the planned ones, so adding a dependency genuinely moves the bar.
  • Named bilingual baselines, compared against the current schedule task by task in the same view.
  • Three guards on every link: no task depends on itself, none links across projects, and no cycle can close in the network.
  • Progress entered from the field and working offline, syncing itself once the connection comes back.

Tip: Why a cycle is refused at write time

Linking "A depends on B" while B already depends on A — even five steps around — makes the critical-path calculation run forever. The link is refused the moment it is created rather than accepted and then rendered as a schedule that cannot be computed. Even when two people wire up the network simultaneously, the second link is refused exactly as if it had arrived after the first.

Frequently asked questions

How is the critical path different from the longest list of activities?+

The critical path is the longest chain of logically linked activities from start to finish, not the longest list. A very long activity with 30 days of float is not critical, and a two-day activity with no float is.

Do I have to enter a start and a finish date for every task?+

No. A duration in days plus the dependencies is enough — the network derives the start from the predecessors and lays the task out for its duration. Entering both ends by hand is useful for tasks pinned by an external date, like a site handover or a permit.

Can a task depend on a task in another project?+

No. A dependency only has meaning inside one project; linking across two would pull dates from a foreign schedule into this project's critical path, so it is refused explicitly.

If we revise the schedule after taking a baseline, is the baseline lost?+

No. A baseline is an independent snapshot stored with its own name and date; editing the schedule afterwards does not touch it, which is the entire point of taking one.

How many baselines should a project have?+

One at the original approval, then one after each formally approved extension of time. Taking a fresh baseline every time the project slips erases the evidence of the slippage — which is exactly what removes the basis for a claim.

Does the critical path replace the schedule performance index (SPI)?+

No, they answer different questions. The critical path says which activities move the handover date if they slip; SPI says whether the project as a whole is ahead of or behind its plan in value terms. They are read together, not instead of one another.

Make your schedule move when the site moves

Link your project tasks with real dependencies in muqawil, save a baseline, and let the critical path be calculated instead of guessed.

Start a free trialExplore the features

Related reading

Cost control20 July 2026·12 min read

Earned Value Management for Contractors

You have spent 60% of the budget and used 60% of the programme — are you on track? Those two numbers cannot tell you. Earned value is the missing third.

Read the article
Programme30 July 2026·12 min read

Construction Delay Claims and Extension of Time

Claims are not won by argument, they are won by records. The contractor who keeps an accurate daily report has won months before the claim file is ever opened.

Read the article
Field operations24 July 2026·10 min read

Daily Site Reports That Hold Up in a Claim

Most daily reports are written to be filed, not read. The gap between a worthless archive and a record that survives scrutiny is five fields, filled in on the day itself.

Read the article
Cost control11 August 2026·13 min read

Why Construction Projects Go Over Budget

A budget overrun does not happen in month nine; it gets discovered in month nine. The money had been leaking since month two, and nobody was looking at the place it leaked from.

Read the article
All articles
muqawil · مقاول

Construction management for the Arab world, in Arabic and English.

Product

  • All features
  • How it works
  • Pricing

Resources

  • Blog
  • Complete guide
  • Glossary

Company

  • Create account
  • Sign in
  • Contact
  • Referral program

Legal

  • Terms of service
  • Privacy policy
© 2026 muqawil. All rights reserved.العربية