A lookahead is not a zoomed-in master schedule
Walk into most project trailers and the three-week lookahead on the wall is a filtered export of the CPM schedule, printed Monday and ignored by Tuesday. It restates what the baseline always said should happen, which is exactly the information the crews do not need. They already know the sequence. What they need to know is which of next week's activities can actually start, and what is standing in the way of the ones that cannot.
That is the difference in purpose. The master schedule is a contractual model of the whole job. The lookahead is a short-horizon production plan, and its real subject is constraints. When a lookahead is built from the constraint list rather than from the Gantt chart, it stops being a reporting artifact and becomes the one document the foremen argue over, which is the only reliable sign that a schedule is being used.
Every activity is either ready, constrained, or committed
Give each activity in the window one of three states and the document immediately becomes useful. Ready means every prerequisite is satisfied: the area is available, the preceding trade is done, the material is on site, the submittal is approved, the answer to the RFI is in hand, the inspection has been signed off. Constrained means at least one of those is missing, and the lookahead names which one and who owns it. Committed means a specific crew has agreed to perform specific work in a specific week.
Only ready work should ever be committed. Committing constrained work is how projects train their people to disbelieve the schedule: the crew shows up, the area is not available, and everyone learns that the plan is fiction. Protecting that rule is the discipline the whole method rests on, and it is usually the part that needs management support rather than software.
The constraint log is the real deliverable
A constraint is a named, dated, owned obstacle. Not a category like submittals, but a specific item: the mechanical submittal on package 15 is in review and the ductwork cannot be fabricated until it returns. A week three activity is constrained by a material release with a six-week lead time, which means the decision is already late. Each entry needs a description, the activity it blocks, an owner, and a need-by date that is earlier than the start date it protects.
This log is where the lookahead meeting should spend most of its time, and it is also the part that maps cleanly onto systems you already have. Open RFIs, submittals in review, purchase orders without confirmed delivery dates, and outstanding inspection requests are all structured records somewhere in your stack. Pulling them into one list against the activities they block is mechanical work, which makes it a reasonable thing to automate.
What automation should and should not do here
A system can assemble the draft: take next three weeks of activities, attach every open RFI, submittal, procurement item, and inspection that touches the same area or scope, flag activities whose need-by dates have already passed, and hand the superintendent a marked-up starting point half an hour before the meeting. That single step removes most of the clerical work that causes lookaheads to be skipped when the job gets busy, which is precisely when they matter most.
What it should not do is decide what is ready. Readiness depends on things the records do not contain: whether the slab is actually clean, whether the other trade's crew is really finishing Friday, whether the owner is about to change the room layout again. The superintendent holds that knowledge. A good system presents the evidence and lets the person who walks the building make the call, then records the call so the reasoning survives.
Run the meeting in a fixed order
The meeting should be short, weekly, and always the same. Start by scoring last week: which committed activities finished as planned, and for each miss, the single reason. Then clear constraints: walk the log, confirm owners and need-by dates, escalate anything already late. Then plan the next window: move newly unconstrained work into ready, and take commitments from each foreman in their own words. Ending with commitments rather than assignments is what makes them stick.
Keep a written record of who committed to what, because the value compounds. After six weeks you can see which trades hit their commitments and which quietly do not, and which constraint types cause most of the misses. That conversation is far more productive than a generic complaint about the schedule slipping, because it points at a specific repeated cause.
Measure percent plan complete, honestly
Percent plan complete is the simplest useful metric in short-interval planning: the number of committed activities completed divided by the number committed, with no partial credit. Teams new to it typically start in the fifties and are often unpleasantly surprised. A number in that range is not a sign the method is failing; it is the first honest measurement of how much of the plan was ever real.
Resist the urge to improve the number by committing to less or by scoring partial completion. Both make the metric comfortable and useless. The legitimate way to raise it is to stop committing constrained work and to clear constraints earlier, and when that happens, the improvement is real production capacity rather than better bookkeeping. Track the reasons for misses alongside the percentage, because the reasons are what you actually manage.
Start with one project and one superintendent
Do not roll this out across a portfolio. Pick one project with a superintendent who is willing, run the cycle for six weeks, and keep the artifacts to three: the three-week grid, the constraint log, and the weekly percent plan complete with reasons for misses. If it works there, the foremen on that job will ask for it on the next one, and that is a far better adoption mechanism than a directive from the office.
The test of whether it is working is not whether the documents look good. It is whether the constraint log shrinks from the front, meaning problems are being cleared before they reach the week they would have blocked. A lookahead that surfaces the same constraint for four consecutive weeks is still a useful diagnosis, but it means the bottleneck is decision-making upstream, not scheduling in the field.
