Construction Workforce Capacity Planning: A Six-State Crew Allocation Register

Why current workforce announcements need a project-level check

On September 14, 2026, Meta announced an expansion of its Eagle Mountain data center and estimated support for more than 1,200 construction workers at peak construction. KUTV independently reported the announcement and attributed the workforce estimate to Meta. That confirms a public project-demand signal; it does not independently verify the forecast or show which trades, employers, shifts, or dates have available crews.

Also on September 14, the Oklahoma House and FOX23 published accounts of a September 8 apprenticeship hearing. The discussion distinguished training pathways and described employer and training-capacity constraints. A training program, enrollment figure, or proposed policy is not a named allocation to a project.

These examples explain why capacity questions are timely. The durable job is to trace each project commitment to a scoped source, expose overlapping allocations, and reopen the decision when its inputs change. The six-state model is NEXUS editorial guidance, not a standard adopted or endorsed by the organizations cited here.

Six capacity states and the evidence needed to move between them

A response to a request is not a reservation, and arrival is not evidence of productive work. Define the transition criteria before using colors or percentages in a dashboard. Move a record only when its required evidence exists, and retain the previous state and the reason for changing it.

• 1. Requested: identify the package, trade, crew shape, location, shift, earliest/latest need dates, duration, requester, and request timestamp. This records demand; it does not establish available capacity or acceptance.
• 2. Acknowledged: retain the named responder, response time, understood scope, questions, exceptions, and capacity source to be checked. An acknowledgment does not reserve a named crew.
• 3. Allocated: identify the crew or permitted capacity-pool reference, responsible employer or Sub, committed window, competing-work check, allocation owner, source date, and review/expiry date. Allocation does not establish that mobilization prerequisites are complete.
• 4. Mobilization-ready: link authorized reviews of supervision, qualifications, travel, lodging, access, orientation, equipment, material, and predecessor work. Readiness does not establish presence, productivity, safety for every task, or acceptance.
• 5. Onsite: retain a dated, authorized presence record with site, shift, employer, supervisor, actual start, and any difference from the planned crew. Presence alone does not prove output or acceptable work.
• 6. Productive: link the actual package, defined output measure, time window, conditions, constraints, source records, and reviewer. Output does not by itself establish quality acceptance, compliance, payment entitlement, schedule recovery, or future performance.

Define the package, crew shape, and need window

Start with the controlling scope revision, project, site or zone, phase, predecessor and successor work, and consequence if the window changes. Separate the requested date, earliest feasible site start, latest acceptable arrival, planned duration, and date the supporting evidence expires.

A request for twenty electricians remains incomplete without classification, shift, supervision, critical roles, travel assumptions, and the work to be performed. Use terms supplied by the responsible employer and governing project requirements. Do not let an AI summary treat different trade classifications or qualifications as equivalent.

• Demand fields: record ID, project, package, scope/revision, trade/classification, location, phase, shift, crew composition, critical roles, need-window start/end, and duration.
• Dependency fields: predecessor release, material availability, equipment, access, supervision, travel/lodging, qualification review, and the owner and required state of each dependency.
• Authority fields: requester, allocation owner, reviewers, source organization, planning decision, limitations, and next review trigger.

Make source age, scope, and uncertainty visible

For each capacity claim, capture who supplied it, their role, the source reference, when it was supplied, which trades and dates it covers, and what remains conditional. Distinguish a planning allowance, manager acknowledgment, named allocation, and actual field evidence.

Freshness is one input to confidence. A current but noncommittal estimate can be weaker than an older named allocation that has been rechecked against competing work. Explain confidence with evidence and limitations rather than assigning an unexplained high/medium/low label.

• Keep the original source or an authorized reference, the date, scope, assumptions, exclusions, reviewer, and expiry/revalidation rule.
• Record missing or disputed evidence explicitly; a blank field must not become an AI-generated fact.
• Use a permitted crew/pool identifier. Restrict personal, employment, and competing-project data to people authorized to see it.

Reconcile competing projects before accepting an allocation

The same supervisor, specialty crew, or equipment operator can appear in several project schedules. Reconcile overlapping windows at the level the organization has authority to review. Include probable awards, delayed predecessors, travel time, rotations, and critical-role coverage.

Illustrative planning example: Package A requests a crew for October 5–16, while Package B has an existing allocation for October 12–23. The overlap is a conflict to review, even if both schedules show green. The responsible allocator must establish which commitment governs, what capacity remains, and which fallback is feasible. These dates are a hypothetical example, not a NEXUS customer case.

• Record the two allocations, overlap window, source dates, affected packages, responsible allocator, review deadline, and current disposition.
• If a predecessor slips, recompute the overlap and reopen the allocation; do not keep the original availability label.
• Preserve which project loses capacity under each scenario and who is authorized to accept the resulting plan.

Separate assumptions, constraints, and fallback decisions

An assumption is a condition used for planning that has not been established. A constraint is a known limit on the feasible plan. Keep both linked to a source, responsible reviewer, and revision trigger. Wage, overtime, shift, travel, training, and productivity allowances should remain stated planning inputs; the register does not set employment policy or contract entitlement.

Possible fallbacks include another qualified crew or Sub, staged mobilization, resequencing, training, prefabrication, or a revised package boundary where permitted. For each, record the trigger, activation lead time, evidence gaps, cost/schedule and other required reviews, decision authority, and latest decision date. Listing an alternative does not authorize its execution.

Revalidate on events and retain the earlier decision

Use a suitable scheduled review and reopen the record when material evidence changes. A weekly meeting alone can miss a new award, absence, date change, or expired qualification. Store the prior value, new value, source, reason, reviewer, authorized decision, and effective time.

• Trigger review after a new award or probable award, scope/shift/sequence change, predecessor delay, or changed required-on-site window.
• Recheck after absence, reassignment, supervisor change, qualification/access expiry, or travel, equipment, and material changes.
• Link daily-log actuals when onsite crew shape, start, conditions, or output differ from the plan. Preserve what was known at the earlier bid or schedule decision.

Pilot the register with seven practical review steps

Choose one representative package and two potentially competing project windows. Reconstruct the evidence without inventing missing records, then have the responsible people test the workflow before extending it across a portfolio.

• 1. Define demand: package, scope revision, crew shape, location, shift, need range, duration, and dependencies.
• 2. Request a scoped capacity response with a named source and exceptions.
• 3. Classify the state using the six transition criteria.
• 4. Reconcile competing commitments, critical roles, probable awards, travel, and sequence changes.
• 5. Test assumptions and fallbacks, including their activation lead time and review requirements.
• 6. Record the human planning decision, accepted limitations, authority, and expiry.
• 7. Revalidate on events and connect later presence/output evidence to the retained planning history.

Measure the quality of the process before claiming a result

Measure the share of claims with a named source/date/scope, allocations checked for competing work, time from a material event to review, stale allocations found before mobilization, and fallbacks with a decision owner and deadline. Compare planned crew composition with authorized onsite records and the defined output measure.

Track reviewer effort and correction rates when AI assists with extraction or comparison. These measures assess the register and review process; they do not establish worker quality, legal compliance, safety, productivity, savings, or recovered schedule.

Where NEXUS can assist and who retains the decision

NEXUS is a BETA construction platform. Its public product facts describe connected schedules, RFIs, submittals, change orders, daily logs, field records, photos, timesheets, and project documentation. These records can support human review of the proposed register. The six states and conflict controls in this guide are editorial recommendations, not claims of a dedicated NEXUS workforce module.

AI may help draft fields, compare source revisions, and flag possible gaps for review. NEXUS and AI do not establish actual labor availability, assign or dispatch crews, verify qualifications or legal/safety compliance, set employment policy, authorize contract changes, or guarantee staffing, productivity, cost, quality, or schedule performance. Employers, contractors, qualified professionals, and authorized project representatives retain those responsibilities.

AI assists. People decide.