Construction project timelines: phases, dependencies, and the dates that slip

A construction schedule is not a list of work. It is a claim about the order things can physically happen in, and about how long each of those things blocks the next one. Framing cannot start until the slab has cured. Drywall cannot start until the rough-in inspection passes. The inspector comes when the inspector comes.

That is what separates a construction project timeline from most other project timelines. In software or marketing, two tasks that overlap on paper usually can overlap in practice if someone works harder. On a site, they cannot. A timeline that does not model those constraints looks reasonable in week one and is wrong by week five, and the people holding it will not know which of the twenty remaining dates moved.

This article covers how to phase the work so the schedule reads at a glance, how to represent dependencies and the waiting time between them, which dates actually slip and why, and how to run the short-horizon schedule that keeps the long one honest.

Phase first, tasks second

The most common mistake is building the schedule bottom up from a task list. Someone lists two hundred activities, sorts them by date, and produces a chart nobody reads. The chart is technically correct and operationally useless, because the questions people ask are about phases.

Start with six to ten phases covering the whole job. A small commercial fit-out or a residential build usually resolves into something like this:

1. Pre-construction: design finalization, permits, procurement of long lead items
2. Site preparation: demolition, clearing, temporary services
3. Foundation and substructure
4. Structure and envelope: framing, roofing, windows, weatherproofing
5. Rough-in: mechanical, electrical, plumbing, followed by inspection
6. Interior finishes: insulation, drywall, flooring, paint, fixtures
7. Exterior and site work
8. Commissioning, punch list, handover

Each phase becomes one bar on the summary chart with a start date, an end date, and an owner. Tasks live underneath, and most people never need to look at them. The client reads the phase bars. The site supervisor reads the tasks inside the current phase and the next one.

Name phases by physical state rather than by trade. "Envelope closed" tells everyone on the job what it means. "Contractor B" does not, and it stops being accurate the moment scope moves between subcontractors.

Dependencies, and the waiting time between them

The phase chart is the summary. The dependencies are the part that makes the summary trustworthy, because they are what allows a date change at the front to propagate to the back instead of quietly breaking the plan.

Four relationships cover almost everything on a construction schedule.

Relationship	Meaning	Typical use
Finish to start	B cannot start until A finishes	Framing after foundation cures
Start to start	B cannot start until A starts	Site cleanup alongside demolition
Finish to finish	B cannot finish until A finishes	Final cleaning finishing with punch list
Start to finish	B cannot finish until A starts	Rare, mostly in handover of temporary services

Finish to start covers the large majority. Where schedules go wrong is not in choosing the relationship. It is in forgetting the gap.

Lag is the waiting time between two linked activities, and it is where construction schedules are won or lost. Concrete curing, paint drying, permit review, material delivery, and inspection scheduling are all lag. None of them is anyone's task, so none of them appears on a task list, and a schedule built purely from tasks compresses them to zero.

Model lag explicitly. If framing follows the pour with a seven day cure, that seven days belongs on the chart as a gap or as a named zero-effort activity, not as slack inside the framing bar. The difference matters when someone asks whether framing can start three days early. With the lag visible, the answer is a fact. With the lag buried, the answer is an opinion.

The same applies to inspections. An inspection is not a one hour event. It is a request, a queue, a visit, and possibly a re-inspection. Scheduling it as a single day and discovering the queue is eight days long is one of the most reliable ways to lose a week.

The dates that actually slip

Every construction schedule has a small number of dates that carry most of the risk. Finding them early is more useful than tightening estimates everywhere.

Long lead items. Custom windows, switchgear, elevators, specialty finishes. These are ordered in pre-construction and arrive months later, and the order date is often months before anyone feels urgency about it. The failure mode is subtle: the item is ordered on time, the lead time changes, and nobody is watching because the delivery is twelve weeks out. Put the order date, the confirmed lead time, and the required-on-site date on the chart as separate markers, and check the confirmation rather than the original quote.

Permits and approvals. The submission date is controllable. The approval date is not. Schedules that show a fixed two week review are showing a hope. Show the submission as a milestone, show the review period as lag with a named assumption, and track the actual duration on every job so the next estimate is based on the local reality rather than the code.

Weather-exposed phases. Excavation, foundation, roofing, and exterior work carry seasonal risk that interior work does not. A schedule that puts roofing in the wettest month of the year is not wrong, but it should say so in a line under the chart so the risk is a decision rather than a surprise.

Trade handoffs. The single most common source of quiet delay is a subcontractor who finishes on time, and the next one who cannot mobilize for four days. The schedule shows the phases touching. Reality has a gap. Handoffs between separate companies deserve explicit lag based on mobilization notice, not zero.

Changes and variations. A change order is not just cost. It consumes design time, possibly re-approval, and possibly a re-order of a long lead item. Attach every approved change to the schedule the same week it is approved, and show what it moved. Changes absorbed silently are the reason a job is four weeks late with no single event to point at.

Float, and why the critical path is worth identifying

Total float is the amount of time an activity can slip without moving the end date. The critical path is the chain of activities with no float, where any delay moves handover by the same amount.

Identifying it changes daily decisions. When two trades both want the same access on the same morning, the one on the critical path wins, and that is an answerable question rather than an argument. When a delay lands on an activity with ten days of float, the correct response is to note it and continue, not to escalate.

Two habits keep float honest.

Do not pad individual activities. Padding hides where the real slack is, and everyone consumes their private buffer because it is theirs. If a buffer is needed, put it at the end of a phase as a named contingency activity, visible to everyone.

Keep the baseline. Save the schedule as it stood at contract signing and never overwrite it. When a date moves, update the forecast and leave the baseline alone. The gap between the two is the only honest answer to how far the job has drifted, and it is the record that matters if the delay is ever disputed.

The look-ahead is the schedule people actually use

A master schedule covering nine months is a planning document. Nobody on site works from it. What crews work from is a three week look-ahead: the activities starting in the next fifteen working days, who is doing each one, what has to be on site for it, and what is blocking it.

The look-ahead is where the master schedule is tested. Every week, the activities that were supposed to start either started or did not, and the reason is concrete: material not delivered, inspection not booked, prior trade not finished, access not available. Those reasons are the data that makes the next master schedule update real instead of aspirational.

A workable rhythm looks like this.

• Weekly: review the look-ahead with the trades on site. Confirm what starts next week and what each start needs. Record what did not happen and why.
• Every two to four weeks: roll the reasons into the master schedule. Move the forecast dates. Leave the baseline untouched.
• At each phase boundary: check that the remaining long lead items still have confirmed dates, and that the next set of approvals has been submitted.

The critical point is that these are two views of the same information, not two documents. When the look-ahead is maintained separately from the master schedule, they diverge within a month and the master becomes fiction. When both are views of the same underlying activities with the same dates, keeping the look-ahead current keeps the master current as a side effect.

Choosing where the schedule lives

Most construction teams below the size of a large general contractor end up with a spreadsheet, a dedicated scheduling package, or a general project management tool with a Gantt view. All three are used successfully. The right answer depends on who has to touch the dates.

A spreadsheet works when one person owns the schedule and updates it from site reports. It costs nothing, and it is the correct choice for a small job with a short duration. It stops working when several people need to update their own dates, because then edits collide and the comments explaining a change live in email instead of next to the date.

A dedicated scheduling package models dependencies, lag, float, and baselines properly, and produces the reports a client or a lender expects. The cost is that it is a specialist tool, so the people using it are the ones trained on it, and the rest of the team receives PDF exports that are stale on arrival.

A general project management tool with a Gantt board sits between the two. The schedule lives with the task discussion, subcontractors can be given access to their own portion, and the same activities can be looked at as a board or a calendar by people who do not want to read a chart. When evaluating one, check that cards carry both a start and an end date, that dependency links are supported rather than simulated with dates, that a baseline can be kept, and that people outside the company can view without a full seat.

The features page shows how Gantt, calendar, and kanban boards work against the same activities, and the getting started guide covers setting up a first board with dates.

What to change first

Take the current schedule and mark the lag explicitly: curing, drying, permit review, delivery windows, and mobilization between trades. On most schedules that alone moves the end date, and it moves it now rather than in month four. Then start a three week look-ahead from the same activities, and record every week why a planned start did not happen. If the schedule is currently a file that one person updates, moving it somewhere the trades can update their own dates, such as Pinateca, removes the transcription step that makes it stale.

Q1. How many phases should a construction project timeline have?

Six to ten on the summary chart. Fewer than six and the phases are too coarse to show where the job is. More than ten and the page stops being readable at a glance. Tasks live inside the phases, where the site supervisor works, rather than on the summary the client reads.

Q2. What is lag in a construction schedule and why does it matter?

Lag is the waiting time between two linked activities, such as concrete curing, paint drying, permit review, or material delivery. It is nobody's task, so it disappears from schedules built purely from task lists. Leaving it out compresses the plan by days or weeks at every handoff, which is the most common reason a schedule is optimistic from day one.

Q3. Should the schedule dates be updated when the job slips?

Update the forecast dates and keep the original baseline untouched. The gap between baseline and forecast is the record of how far the job has drifted, and it is what matters if a delay is ever disputed. Overwriting the baseline erases the history and makes every schedule look on time.

Q4. What is a three week look-ahead and do small jobs need one?

It is the list of activities starting in roughly the next fifteen working days, with owners, required materials, and blockers. Small jobs benefit as much as large ones, because it is the document that catches a missing delivery or an unbooked inspection while there is still time to react. It should be a view of the same activities as the master schedule, not a separate file.

Q5. Is a spreadsheet enough for a construction project timeline?

For a short job where one person owns the schedule and updates it from site reports, yes. It becomes difficult when multiple trades need to update their own dates, because edits collide and the reasons behind date changes end up in email rather than next to the activity. At that point a tool with a shared Gantt view is easier to keep accurate.

Q6. Which dates carry the most schedule risk?

Long lead item deliveries, permit and inspection approvals, weather-exposed phases, and handoffs between separate subcontractors. Those four account for most slippage on small and mid-sized jobs. Marking them on the chart with their assumptions written down turns them from surprises into decisions.