Knowledge · Scheduling
The critical path and task dependencies,
the chain that sets the handover date.
Every residential job has one chain of linked tasks where a one-day slip is a one-day-later handover. This node explains dependencies in plain terms, how the critical path is found and why it moves, what float is and why it disappears quietly, and where the path actually runs on an Australian residential job, through inspections, curing, selections and supervision more than through any single trade's speed.
01 / Overview
What a dependency is, and what the critical path is
A dependency is a plain statement about the physical world, this task cannot start until that one finishes. The frame cannot stand until the slab exists, the plasterer cannot sheet until the rough-ins are inspected, the tiler cannot tile until the waterproofing is signed off. Most links on a residential job are that finish-to-start shape. A few are different and equally real, some tasks genuinely start together (demolition and the site fencing that must be up from its first hour), and some must finish together (the last trade's touch-ups and the final clean that cannot complete until that trade leaves). This node is part of the residential construction scheduling reference, and it covers the mechanism the rest of that cluster stands on.
Link every task to what it genuinely waits on and the job stops being a list and becomes a network of chains running from site start to handover. The critical path is the longest of those chains. Its defining property is brutal and useful, a one-day slip anywhere on it is a one-day-later handover, because nothing behind it can absorb the delay. Every task off that chain has float, some number of days it can slip before it, too, moves the end date. None of this is proprietary or contested; the critical path method has been standard project-management practice for decades, and residential builders have run it informally, in their heads, for at least as long.
Why it matters
The critical path tells a builder where attention buys back time and where it does not. A day of effort spent accelerating a critical task shortens the job by a day; the same day spent on a floating task shortens nothing. Without the path, urgency gets allocated by noise, whichever trade rings most, whichever problem is most visible on site, and the quiet dependency that actually sets the handover date (a certifier's booking, a curing slab, an unmade selection) slips unwatched. The path is also what makes a delay explainable. A handover conversation that walks a client down a chain of linked dates is a different conversation from one that offers an apology and a guess.
02 / The lifecycle
Where this sits in the scheduling discipline
Dependencies and the critical path are the mechanics underneath everything else in the scheduling cluster. The order the trades and stages run in, and the material lead times that shape it, are covered in lead times and sequencing; whether the job is actually tracking against the programme it promised is covered in baselines and progress tracking. Both of those depend on this node's mechanism, because a sequence is a set of dependencies read forwards, and a baseline comparison only means something when the schedule being compared recalculates honestly. The practical end-to-end treatment, from first programme to handover, lives in the residential construction scheduling guide.
The path also reaches outside the scheduling cluster. Order dates for long-lead packages are derived backwards from the programme, so the buying is downstream of the dependency network (covered in aligning procurement with the schedule), and a schedule that moves without the orders moving with it delivers materials to the wrong week. Money is downstream of it too. Every day the critical path stretches is another day of preliminaries, supervision, site costs and facilities, burning against a fixed price, which is why the longest chain on the programme is also a line in the job's cost story.
03 / Process workflow
Building a schedule whose links are real
Eight steps. The first five build the dependency network and find the path; the last three are what keep it true once the job starts moving.
- 01
List the real tasks, including the waiting
A task is anything that consumes time, not just anything that consumes labour. Curing, approvals, inspections, engineering sign-offs and deliveries are tasks. A schedule that only lists trade work has already hidden most of its critical path.
- 02
Link each task to what it genuinely waits on
For every task, ask what must be true before it can start. Most links are finish-to-start (the frame waits on the slab); some tasks genuinely start together or must finish together. Link only what is real, because false links create false urgency and missing links hide real delay.
- 03
Add the hold points as tasks with owners
Mandatory inspections, certification stages and engineer sign-offs are tasks someone must book, evidence someone must supply, and time nobody can compress. Give each one an owner and a lead time, the same as a material order.
- 04
Find the longest chain
Follow the links from start to handover and add up each chain. The longest one is the critical path, the chain where a one-day slip is a one-day-later handover. It is date arithmetic through the links, not judgement, which is why software can do it and memory eventually cannot.
- 05
Read the float on everything else
Every task off the critical path can slip by some number of days before it moves the end date. That number is its float. Know it before the job starts, because it is the only spare time the job has and it does not announce when it runs out.
- 06
Protect the chain deliberately
Supervision attention, order dates and client selection deadlines all get derived from the critical path, not spread evenly across the job. A day of the supervisor’s time spent on a critical task buys back a day; the same day on a floating task buys back nothing.
- 07
Re-run the links when reality diverges
A variation, a failed inspection, a wet fortnight or a late selection changes the arithmetic. Re-running the links is what tells you whether the critical path still runs where it did yesterday, and it very often does not.
- 08
Resequence with the dependencies, not against them
When the job needs to catch up, overlap the tasks whose links allow it and leave alone the ones whose links do not. Compression that ignores a real dependency is not a faster schedule, it is a defect or a failed inspection booked in advance.
04 / Key mechanics
Where the residential critical path actually runs
On a residential job the longest chain rarely runs through any single trade's speed. It runs through the waits, the sign-offs and the decisions between the trades.
Inspection and certification hold points
Mandatory inspections and certification stages are dependencies nobody can negotiate with. The work after a hold point cannot start until the sign-off exists, and no amount of extra labour compresses the wait for a certifier’s diary.
Engineering and certifier evidence
Engineers and certifiers need site photos, records and results in the right format at critical stages. Evidence sent to a phone instead of the formal record gets asked for twice, and the second ask is pure critical-path time.
Wet trades and curing
Concrete cures, waterproofing dries and screeds set at the pace of chemistry and weather, not the programme. These are real tasks with real durations, and schedules that omit them are optimistic by exactly that much.
Supervision capacity
A supervisor running several jobs is a shared dependency none of the schedules show. When two jobs need the same person the same week, one of them slips, and it is usually the one whose critical path nobody was watching.
Client selections and decisions
A selection that is not locked holds every order and trade behind it. Selection deadlines belong on the schedule as dependencies with dates, because a client decision can sit on the critical path as surely as a slab.
Long-lead deliveries
Trusses, windows and joinery arrive a lead time after they are ordered, and the order waits on confirmed drawings and locked selections. The delivery is on site’s schedule; the dependency chain behind it starts months earlier.
Float, and why it quietly disappears
Float is the schedule's shock absorber, the days a task can slip before it drags the handover with it. It has two properties that catch builders out. It is shared along a chain, so when an early task consumes a week of float, every task behind it on that chain has a week less, whether or not anyone tells them. And it vanishes silently. There is no moment on site when a floating task visibly becomes a critical one; the arithmetic changes and nothing else does, until the next small slip on that task moves the end date and everyone is surprised. In practice, float spent early in a job on things that merely drifted is float that is not available later for the things that genuinely break.
Resequencing when reality diverges
Programmes diverge from reality on every job, and the honest response is to resequence with the dependencies rather than around them. Some overlaps are genuinely available, external works can often run alongside internal fit-out once the building is locked up, different zones of a house can carry different trades at once, and a floating package can be pulled forward to fill a gap. Other overlaps do not exist no matter how late the job is. The frame does not load a slab that has not cured, tiling does not cover waterproofing that has not been inspected, and no stage proceeds through a hold point that has not been signed. The dependency network is what tells you which kind of overlap you are looking at before the site finds out the hard way.
Dependency honesty
A schedule whose tasks are not really linked recalculates nothing when one of them moves. It looks like a programme, bars, dates, trade names, but each bar sits where a person dragged it, so every change in the real world has to be re-entered by hand on every bar it touches. After the second or third change nobody re-drags everything, the document and the site part company, and the builder is back to running the job from memory with a decorative chart behind them. The test of a schedule is not how it looks on day one but what it does when a date moves, and only real links give a useful answer.
05 / Best practice
How experienced builders carry the critical path
Experienced supervisors know the critical path without software, because they know which delay they cannot buy back. Ask one why the certifier gets chased at seven in the morning while a late paint delivery gets a shrug, and the answer is the path, whether or not they use the words. Paint has float; the certifier's sign-off holds the next stage. That knowledge is real and hard-won, and its weakness is that it lives in one head. It goes on holidays when the supervisor does, it stretches thin when one person carries several jobs, and it leaves the business when they do. The discipline software adds is not intelligence the supervisor lacked; it is making that knowledge survive the supervisor's absence, and re-running the arithmetic in seconds instead of an evening.
The classic residential trap is treating the critical path as fixed. The path calculated at contract is a snapshot, and a variation or a late selection reroutes it silently. A mid-build variation that needs an engineering re-check puts a new task in the middle of the chain; a client selection that runs past its deadline can drag a task that had weeks of float directly onto the path. Builders who have been caught once tend to build the reflex, every approved variation and every missed selection deadline triggers a re-run of the links, because the question is never only what does this change cost but what does this change now hold up. In pre-construction the same mechanism runs in miniature, jobs commonly cannot start while variations and final pricing are still being settled, so the critical path of the whole project can begin at a desk, not on site.
Where software fits the workflow
Traditionally the dependency knowledge lives in the supervisor's head and the programme lives in a spreadsheet or a wall chart that neither knows nor recalculates the links. In VIABUILD the schedule is task-based with real dependencies, so when a date slips the chain of dependent tasks moves with it, the new critical path is visible immediately, and the affected trades can be told from the task itself rather than from memory. The sequencing judgement stays with the builder, which overlaps are real, which delay to chase first; what changes is that the arithmetic is done in seconds, and it is still done when the person who carries the job in their head is on leave.
06 / Australian considerations
Hold points, curing and delay on Australian jobs
Critical path scheduling itself is working method, not a regulated act, but several Australian realities put fixed points in the network. The points below are labelled by evidence class and are general information; requirements differ by state and change over time, so confirm the current position with your certifier, regulator or adviser before relying on any of them.
- Legislation. Each state and territory's building legislation prescribes stages at which work must be inspected or certified before the job proceeds, and the stages, the inspecting party and the process differ by jurisdiction. On the programme these are hold points that cannot be compressed by adding labour, and the current requirements for your state need confirming with your certifier or building regulator.
- Common practice. Engineers and certifiers require site evidence (photos, records, test results) at critical stages, in a form they can rely on for the formal record. A recurring cause of hold-point delay is not missing work but evidence in the wrong place or format, a photo on a supervisor's phone instead of a record against the job, which gets asked for twice.
- Industry best practice. The critical path method is standard, uncontested project-management practice, documented in recognised bodies of knowledge such as the PMBOK Guide. Residential builders apply the same mechanism at smaller scale, and the value is in honest links and durations, not in the sophistication of the tool.
- Common practice. Curing and drying times for concrete, waterproofing and other wet trades are set by the relevant standards, product specifications and site conditions, not by the programme. Durations vary by product and weather, so schedule them from the specification for the actual system used rather than from habit.
- Common practice. Delays on the critical path connect to the contract. Residential building contracts commonly provide for extensions of time for qualifying causes, with notice requirements, and the rules differ by contract and jurisdiction. A dependency-honest schedule is also the evidence trail for that conversation; confirm the position under your contract and state legislation.
- Professional recommendation. Derive client selection deadlines from the dependency network and issue them early, in writing, with the downstream consequence stated. A selection is a dependency, and a client who can see the chain between their decision and their handover date decides sooner.
07 / Common mistakes
Where dependency thinking goes wrong
Each of these is a mechanism, not misfortune, and each one is invisible on the day it happens and expensive on the day it surfaces.
A task list wearing a schedule’s clothes
Bars placed on dates by feel, with no links between them, recalculate nothing when one moves. Every change means someone re-drags every bar by hand, and after the second change nobody bothers, so the schedule quietly stops being true.
Treating float as spare time to spend
Float absorbs slippage; it is not a budget for starting late. Spend a floating task’s slack casually and it joins the critical path without telling anyone, and the next small slip on it moves the handover.
Assuming the critical path is fixed
The path calculated at contract is one snapshot. A variation, a failed inspection or a late selection reroutes it mid-job, and a builder protecting last month’s path is guarding a door the delay no longer uses.
Leaving the waiting off the programme
Curing, approvals, inspection bookings and delivery lead times consume calendar days whether or not the schedule admits it. Omitting them makes the programme shorter on paper and identical on site.
Padding every task instead of managing the chain
Hidden contingency on every line makes every duration untrustworthy and the critical path invisible. Honest durations with float read openly beat padded durations nobody believes, because the first can be managed and the second can only be discovered.
Compressing through a real dependency
Tiling over waterproofing that has not been inspected, or loading a frame before the slab has cured, converts a schedule problem into a rectification problem. The dependency was never optional; it was just ignored until it was expensive.
08 / Practical example
A worked reroute of the critical path
Illustrative only, not a benchmark. A single-storey custom home is mid-frame, and the critical path runs the expected route, frame, frame inspection, lockup, rough-ins, wet areas, fit-off, handover. The tile selection is open but harmless, sitting on a floating chain with roughly three weeks of slack. Then the client approves a variation moving a bathroom wall. The engineering re-check and revised drawings insert a new task in front of the frame inspection, and the inspection date moves out a fortnight, taking every internal trade behind it along. Less visibly, the wet-area layout has changed, the waterproofing now waits on the revised drawings and the tile selection, and the selection's three weeks of float have been consumed by a decision made in a meeting, not by anything that happened on site.
The builder who re-runs the links the day the variation is approved sees both movements at once, books the certifier against the new date, puts a written selection deadline in front of the client, and resets the handover conversation while it is still a conversation. The builder who treats the variation as a pricing exercise discovers the same facts one at a time, from the certifier's diary, then from the waterproofer, then from the tiler asking which tiles. The site looked busy the whole time, because the frame crew kept working elsewhere on the house. Activity was never the question; the chain was.
09 / FAQ
Common questions.
No. It is the longest connected chain of dependent tasks from start to handover, and chain length is what matters, not the size of any single task on it. A two-day inspection booking that sits on the chain moves the handover if it slips; a three-week landscaping package that sits off the chain, with float behind it, does not. That inversion is why the busiest trade on site and the most schedule-critical task are frequently different things, and why watching activity is not the same as watching the path.
Yes, and on residential jobs it routinely does. The path calculated at contract reflects the durations and links known that day. A variation that triggers an engineering re-check, a selection that runs late, a failed inspection or a wet month all change the arithmetic, and the longest chain can reroute through tasks that previously had float. This is why the useful discipline is not calculating the path once but re-running the links whenever something real moves, because the path you are protecting is only ever the path as of the last recalculation.
Treat it as absorbing capacity rather than a budget. Float is the number of days a task can slip before it moves the end date, and it exists to soak up the ordinary friction of a building site. Two properties make it dangerous to spend casually. It is shared along a chain, so an early task that consumes it leaves nothing for the tasks behind it, and it disappears silently, with no alarm at the moment a floating task becomes a critical one. Many builders find the jobs that end late are not the ones with a single big delay but the ones whose float was spent a day at a time.
They already use it, mostly without the name. A supervisor deciding which subbie to chase first is running the calculation in their head, from experience of which delay cannot be bought back. What a small builder gains from making the dependencies explicit is not new knowledge so much as durability and arithmetic. The knowledge survives the supervisor being on leave or on another job, the recalculation happens in seconds instead of an evening, and the client conversation about a slip can point at a chain of linked dates rather than an opinion.
They put hard hold points in it. Building legislation in each state and territory prescribes stages at which work must be inspected or certified before the next stage proceeds, and the stages and processes differ by jurisdiction, so the current requirements need confirming with your certifier or regulator. On the schedule, each hold point behaves like a task with three parts, the evidence the engineer or certifier needs, the booking of the inspection itself, and the sign-off, and none of the three compresses by adding labour. Jobs that run smoothly through hold points tend to be the ones where the evidence was captured as the work happened rather than hunted for afterwards.
Finish-to-start covers most residential links, one task cannot start until another finishes. Two other relationships are worth knowing in plain terms. Some tasks genuinely start together, such as a demolition and the site security that must exist from its first hour, which schedulers call start-to-start. Some tasks must finish together, such as final trade touch-ups and the cleaning that cannot complete until the last trade leaves, which schedulers call finish-to-finish. The jargon matters less than the honesty; the question for every link is what this task really waits on, and at which end.
10 / Terms
Glossary for this topic
Dependency (a link stating what a task genuinely waits on), finish-to-start (the common link, one task starts after another finishes), start-together and finish-together links (start-to-start and finish-to-finish in scheduling terms), critical path (the longest chain of dependent tasks, where a slip moves the handover), float or slack (the days a task can slip before it moves the end date), hold point (a mandatory inspection or certification the next stage waits on), resequencing (reordering or overlapping tasks within their real dependencies), baseline (the saved programme that progress is measured against). The wider vocabulary lives in the construction glossary. From here, the natural next article is lead times and sequencing, how material lead times and trade availability shape the order the chain runs in.
11 / Keep reading
Related knowledge, guides and features
12 / Further reading
Primary sources
- Project Management Institute , publisher of the PMBOK Guide, the recognised body of knowledge documenting the critical path method.
- Your state or territory's building regulator, for the mandatory inspection and certification stages that put hold points in your programme.
- Your engineer's and certifier's documentation requirements, the primary record of what evidence each hold point needs and in what form.
- HIA and Master Builders Australia, for contract administration guidance on delays and extensions of time aligned to state legislation.
Know which delay you cannot buy back.
VIABUILD runs a task-based schedule with real dependencies, so when a date slips the chain moves with it, the new critical path is visible immediately, and the affected trades can be told from the task itself, whoever is holding the job that week.