More Doors per Employee Is Not an AI Strategy

More Doors per Employee Is Not an AI Strategy

A higher doors-per-employee ratio can be a useful staffing signal. It is not, by itself, proof of productivity, AI return, or better operations.

With the door count held constant, the ratio rises when the employee denominator falls. That may reflect real capacity: fewer paid hours produce the same or better operating outcomes. It may also mean maintenance moved to a vendor, remaining employees worked more overtime, central teams absorbed site work, or technology and review work sit outside site payroll.

For an owner, the analysis has two parts. Did defined output per paid labor hour improve? Did total labor-and-technology cost and owner outcomes improve while service held or improved?

Those questions should be answered before staffing changes are credited to AI.

Headcount Is Not the Productivity Definition

The U.S. Bureau of Labor Statistics release dated September 3, 2026, defines labor productivity as output per hour worked, not output per employee headcount. It calculates unit labor cost as hourly compensation divided by labor productivity.

These are economywide statistical concepts, not multifamily operating benchmarks. They still offer a sound operating discipline: define the output, count the hours required to produce it, and account for the cost of those hours.

A doors-per-employee metric cannot do that on its own. It does not show:

• whether employees are full-time, part-time, or shared across properties;
• regular and overtime hours worked;
• work performed by central teams rather than site staff;
• work moved to vendors, after-hours coverage, or call handling;
• benefits, payroll burden, software, implementation, and support costs; or
• whether residents, owners, and property teams received the same level of service.

The BLS Employer Costs for Employee Compensation release dated September 9, 2026, separates wages and salaries from benefit costs and reports employer costs per hour worked. The June 2026 figures cover broad worker groups and are not evidence of multifamily wage or benefit levels. Their useful lesson here is narrower: salary and headcount alone are incomplete labor-cost measures.

Four Ways a Staffing Ratio Can Improve Without a Proven Gain

A contractor replaces payroll labor

An operator may move turns, maintenance calls, or after-hours work to a contractor. Payroll headcount declines, and doors per employee rises. The work still exists, and contractor invoices may offset some or all of the payroll reduction.

The staffing ratio improved. Total controllable operating cost may not have.

Centralized staff absorb site work

A central leasing, resident-services, accounting, or maintenance-support team may take on work formerly handled at the property. A site-level ratio can look better if central allocations are excluded.

That may be a sound operating model. It is not a demonstrated site productivity gain unless the comparison includes the central labor and technology required to support it.

Overtime carries a smaller team

A smaller employee roster may handle the same workload by working longer hours. Doors per employee rises because employee count fell. Hours per door and overtime premiums may rise at the same time.

The better capacity test is whether completed qualified work per paid hour improved, not whether each named employee now covers more units.

AI reduces visible headcount but adds other inputs

An AI-enabled workflow may reduce manual routing, drafting, triage, or follow-up. If it also requires subscriptions, integration, training, exception review, quality control, or human escalation, those inputs belong in the evaluation.

The supplied sources do not quantify those AI-related costs or establish that every implementation creates them. The point is methodological: a headcount ratio cannot establish return when material inputs sit outside the denominator.

Rebuild the Denominator Before Comparing Portfolios

NAA staffing guidance dated June 8, 2020, treats operating coverage, property characteristics, temporary staffing, and choices between in-house and contracted functions as staffing considerations. Historical NAA operating materials have also distinguished units per full-time employee from units per total employee and presented results by property characteristics.

These materials are useful for their definitions and cautions. They are historical and portfolio-specific, not current universal staffing targets.

Use two related denominators because hours and dollars answer different questions.

Labor capacity denominator: paid hours

For an owner-designed reconciliation, track the material labor used to produce defined outcomes:

• regular employee hours;
• overtime hours, shown separately;
• allocated hours from central teams when material;
• contractor hours when available and reliable;
• after-hours, call-center, and overflow-coverage hours where identifiable; and
• material owner or asset-management time directly required by the operating model.

Contractor invoices are not a substitute for contractor hours when assessing labor capacity. They remain essential for assessing economics when contractor hours are unavailable. Report that measurement gap rather than assuming outsourced work disappeared.

Cost denominator: total labor and technology cost

Track the material controllable inputs associated with the workflow:

• wages, salaries, benefits, and payroll burden;
• overtime premiums;
• central-team allocations;
• contractor and temporary-labor expense;
• after-hours and call-center expense;
• AI and related technology subscriptions;
• implementation, integration, training, and support costs; and
• material owner or asset-management costs when they are part of the evaluated change.

NAA's December 22, 2025 discussion of 2024 operating data defines property payroll expense to include payroll taxes, benefits, overtime, and other employment costs. That supports a fully loaded view rather than a salary-only comparison. Operators should still confirm the classifications used in their own accounting systems.

Do not force hours and costs into one number prematurely. Paid hours help assess capacity. Total cost helps assess economics. Both help distinguish a genuine gain from a shift in who performs or pays for the work.

Normalize the Numerator, Too

Raw doors are a weak proxy for workload. NAA's historical staffing guidance identifies property age, size, location, and structure type as factors that can change staffing requirements.

For a practical owner comparison, hold relevant conditions constant or segment the results. Depending on the portfolio, those conditions may include:

• asset class and building age or condition;
• geography and site density or dispersion;
• occupancy and turnover;
• affordability, compliance, or reporting obligations;
• service scope, including leasing, maintenance, turns, and after-hours coverage;
• work-order volume and maintenance intensity; and
• the measurement period.

The reviewed sources do not prescribe that full segmentation list. It is an operator-designed framework that should be adapted to the portfolio and available data.

A portfolio with older buildings or a broader service scope should not be compared with a newer or narrower-service portfolio through doors per employee alone. That comparison may measure portfolio mix as much as management performance.

Use outputs that match the work under review. Depending on the workflow, that could mean completed work orders, completed turns, resolved resident contacts, qualified leasing follow-ups, or another defined outcome. Define completion before measurement begins. A task closed in software is not necessarily a problem resolved in the field.

A Practical AI Capacity Scorecard

The reviewed sources do not prescribe or validate a universal multifamily AI scorecard. The framework below is an operator-designed control that applies their central lessons: define output, count hours, include compensation and other material costs, and disclose portfolio mix.

Every metric should carry a defined period, cohort, service scope, calculation method, and accountable data owner.

| Decision area | What to measure | What it helps reveal | |---|---|---| | Total input cost | Total labor-and-technology cost per occupied unit and per completed qualified outcome | Whether payroll reductions were offset by vendors, overtime, central support, or technology expense | | Capacity | Completed qualified work per paid labor hour, with outsourced activity identified separately | Whether the operation produces more defined output from the labor it pays for | | Backlog health | Age of open work orders and share past the operator's defined service standard | Whether apparent capacity came from deferring work | | Repair quality | Repeat repairs, reopened work orders, and follow-up contacts | Whether faster closure coincided with more rework | | Response quality | First-response time, completion time, escalations, and resident follow-up | Whether service held while staffing changed | | Owner outcomes | Controllable expense, budget variance, vacancy-ready cycle time, and other workflow-linked operating results | Whether the operating change produced an owner-relevant result |

Do not attribute every owner outcome to AI. Collections, renewals, occupancy, and other portfolio results can move for multiple reasons. Use them as causal evidence only when the operator can document a plausible workflow connection and address competing factors. Otherwise, report them as context.

Test for Capacity Gains, Not Just Payroll Changes

A credible comparison starts before the staffing decision is made.

1. Freeze definitions. Specify the properties, dates, included work, employee categories, centralized allocations, vendor categories, and technology costs.
2. Use a comparable cohort. Compare the same properties before and after implementation, or comparable groups with similar portfolio mix and service scope.
3. Show the full cost bridge. Reconcile changes in payroll, benefits, overtime, contractor expense, central allocations, and technology cost.
4. Report regular and overtime hours separately. A reduction in headcount accompanied by higher overtime is an operating fact, not a footnote.
5. Identify transferred work. State what moved to vendors, central teams, resident self-service, owners, or asset management.
6. Review service alongside cost. A lower-cost model accompanied by an older backlog or more repeat repairs has not yet demonstrated a durable operating improvement.
7. Consider phased or holdout comparisons where operations allow. The reviewed evidence does not establish a universal experimental design. A comparison group may still make a pre/post claim more interpretable when the groups and operating conditions are documented.

If contractor hours, central allocations, or owner time cannot be measured reliably, say so. The operator can then report the measured payroll or staffing change, but the broader productivity or AI-return conclusion remains provisional.

Report Results Without Overclaiming

Avoid: AI reduced staffing costs.

Use evidence-led language instead: For a defined portfolio cohort and service scope, completed qualified outcomes per paid hour changed in the documented direction. The report separately reconciles contractor activity, central allocations, overtime, and technology cost and shows how backlog age, repeat repairs, response measures, and owner outcomes changed during the same period.

That formulation identifies the comparison conditions and leaves room for mixed results. A program may create measured capacity without immediate cost savings. It may reduce measured cost while exposing a service issue that needs correction. Both findings are actionable when the underlying measures are disclosed.

The reviewed evidence does not establish a universal doors-per-employee target, a universal AI savings rate, or a causal link between AI adoption and lower multifamily operating cost. It supports a more disciplined conclusion: employee headcount is too narrow to carry the claim alone.

Bottom Line

Doors per employee remains a useful diagnostic when the employee definition, service scope, and portfolio mix stay consistent. Treat it as evidence of AI productivity only as part of a comparison that reconciles material labor and technology inputs, measures defined outcomes per paid hour, and reviews service and owner outcomes.

Build the baseline scorecard before changing staffing or declaring savings. It gives owners a defensible way to recognize measured capacity gains and prevents a smaller payroll denominator from doing more work than the evidence can support.

Sources Reviewed

• U.S. Bureau of Labor Statistics, Productivity and Costs, Second Quarter 2026, Revised, September 3, 2026.
• U.S. Bureau of Labor Statistics, Employer Costs for Employee Compensation, June 2026, September 9, 2026.
• National Apartment Association, Best Practice for Staffing Considerations, June 8, 2020.
• National Apartment Association, 2020 Survey of Operating Income and Expenses in Rental Apartment Communities: Executive Summary, October 2020.
• National Apartment Association, From Momentum to Management: Navigating Elevated Costs in a Constrained Operating Environment, December 22, 2025.