A Staffing Plan Built on Real Numbers

A staffing plan answers three questions for a fixed period, usually a fiscal year: what work has to get done, how many people with which skills it takes, and how the gap between that and the current team gets closed. Most published methods are variants of the same model. This guide follows the version used by the U.S. federal government, because it is documented in full, free, and built for organisations that have to justify every headcount.

The model: six steps, two analyses

The U.S. Office of Personnel Management’s Workforce Planning Guide (November 2022) structures the work as strategic direction, workforce analysis (supply, demand, gap and risk), an action plan, then implementation and monitoring. The National Institutes of Health’s Office of Human Resources breaks the same process into six labelled steps: strategic direction, supply analysis, demand analysis, gap analysis, solution implementation, monitoring progress.

Two of those steps do the real work. Supply analysis is “how it is projected to change over time due to accessions, attrition, and trends”, in OPM’s phrasing. Demand analysis is what the business will need. Everything else is setup or follow-through.

Step 1: Pin the plan to business goals

Start from documents that already exist: the annual budget, the product or service roadmap, and any revenue or volume targets. Each one implies work. A plan to open a second distribution site implies a site manager, supervisors and a shift roster; a plan to cut ticket backlog implies support headcount or automation, not both.

OPM’s guide asks planners to identify mission-critical positions first, then plan for near- and long-term issues and risk factors. In a company, “mission-critical” means the roles where a vacancy stops revenue or breaks compliance. List those explicitly; they get priority in every later step.

Step 2: Supply analysis — what you actually have

Build a roster with, at minimum: role, location, employment type, start date, skills, and known departure risk (retirement eligibility, visa expiry, open transfer requests). Then project it forward using your own turnover history, not an industry average.

If you have no history, the national rate gives a rough ceiling. In the Bureau of Labor Statistics’ June 2026 Job Openings and Labor Turnover Survey, total separations were 5.4 million for the month against about 7.4 million open positions; quits were 3.2 million and layoffs and discharges 1.8 million. Separations at that level — roughly 3.4% of employment per month — compound to roughly a third of a workforce per year, so a 100-person team that plans on zero attrition is planning for something that has never happened.

OPM’s data-source list for this step includes competency assessments, workload ratios, exit and stay surveys, and engagement data, alongside external labour-market data from BLS and O*NET. A skills inventory matters most when the gap is a skill rather than a body: the World Economic Forum’s Future of Jobs Report 2025 reports employers expect 39% of workers’ core skills to change by 2030, which means part of next year’s demand will be for skills current staff do not hold yet.

Step 3: Demand analysis — what the work requires

Translate goals into units of work, then units of work into people. Methods in rough order of rigour:

• Ratio-based. Staff per unit of volume (tickets per agent, accounts per manager, revenue per salesperson), applied to forecast volume. Fast, and fine when the work is stable.
• Driver-based. Separate drivers per role — e.g. support headcount tracks active customers, implementation headcount tracks new deals. Better when different parts of the business grow at different rates.
• Scenario-based. Three volume cases (low, base, high) with a staffing number for each. This is what OPM means by risk analysis: weigh each risk by impact and likelihood rather than planning on one number.

AI forecasting tools in HRIS and workforce-planning suites automate the arithmetic. They do not fix a bad driver. If the model is told support headcount tracks revenue, and revenue grows because of price increases rather than customers, the forecast over-hires. Check the drivers by hand once a year.

Step 4: Gap analysis — and which gaps matter

Subtract projected supply from demand per role and quarter. The result is a list of surpluses and shortfalls, and it is almost always longer than the hiring budget. Rank it by the criticality from Step 1: a three-month gap in a mission-critical role outranks a year-long gap in a nice-to-have.

Record the gap type, because the fix depends on it:

• Headcount gap — enough skill, not enough people. Fill by hiring or temporary capacity.
• Skills gap — enough people, wrong skills. Fill by training, redeployment, or a targeted hire.
• Timing gap — the right people, but not when needed. Fill by schedule changes, overtime, or a short contract.

Step 5: Choose the fill and cost it

Each fill option has a different cost structure and a different legal footprint.

Option	Use when	Cost and compliance notes
Full-time hire	Need is ongoing and nobody internal can absorb it	Highest fixed cost; carries the full benefit load below
Part-time hire	Ongoing but limited to specific hours	Lower benefit cost; check eligibility thresholds for benefits and leave
Temporary or agency staff	Seasonal spike, absence cover, a project with an end date	Agency markup replaces recruiting and benefit cost; see employer of record vs staffing agency for the contract structures
Independent contractor	Defined deliverable, worker controls how it is done	Classification risk; see below
Redesign the work	Gap is small or timing-only	Overtime, cross-training, automation; no new headcount

Loaded cost. Salary is about 70% of the bill. In the BLS Employer Costs for Employee Compensation release for March 2026, private-industry compensation averaged $46.60 per hour worked: $32.60 in wages and salaries (69.9%) and $14.01 in benefits (30.1%), the latter covering paid leave, supplemental pay, insurance, retirement and legally required contributions. Budget every new role at salary multiplied by 1.3–1.45, depending on your benefits package, before adding recruiting cost.

Recruiting cost. SHRM’s benchmarking put the average cost per hire at nearly $4,700; that figure is from 2022 and will be low for specialised or executive roles. Use your own applicant-tracking data if you have a year of it.

Contractor classification. The federal test is in motion. In February 2026 the Department of Labor proposed rescinding the 2024 independent-contractor rule in favour of a test that weighs control over the work and the worker’s opportunity for profit or loss more heavily; the comment period closed in April 2026 and a final rule had not been published at the time of writing. State tests (California’s ABC test, for example) apply regardless. A staffing plan that leans on contractors should name who owns the classification review.

Step 6: Put AI-hiring compliance in the plan, not after it

Any plan that closes gaps through hiring will probably run candidates through automated screening. That creates obligations that are easier to budget for up front.

• If candidates are in New York City, Local Law 144 bars using an automated employment decision tool unless it had an independent bias audit within the prior year, the audit summary is public, and candidates received notice. The audit is an annual recurring cost; put it in the plan. Details are in our plain-English guide to Local Law 144.
• Federal disparate-impact liability under Title VII still applies to algorithmic selection even though the EEOC withdrew its AI technical-assistance documents in 2025. How AI hiring bias arises and how to audit for it covers the selection-rate maths.
• If the plan forecasts high volume for a role, the screening tool for that role is the one to audit first; that is where an adverse-impact problem does the most damage.

Step 7: Monitor with a short metric set

OPM’s final step is to monitor against milestones and revise. Four numbers cover most of it, reviewed quarterly:

• Forecast accuracy — planned headcount per role versus actual, with the reason for each variance.
• Time to fill for mission-critical roles — the only roles where the number drives a decision.
• Regretted attrition — departures you would have kept, as a share of total departures.
• Backfill lag — days between a departure and the replacement’s start date. What backfilling a position means covers how to track it.

Revisit the full plan when a driver changes (a product launch slips, a site closes), not on a fixed calendar. A plan reviewed once a year is a budget document; one tied to its drivers is a staffing plan.

What the evidence does and does not support

The process above is well documented by government HR bodies, and the cost and turnover figures come from BLS surveys with published methodology. What is thin is evidence that any particular forecasting technique outperforms another across organisations; vendor claims about AI workforce-planning accuracy are rarely published with comparison data. Treat forecast tooling as a way to run more scenarios faster, and keep the drivers, the criticality ranking and the classification review as human decisions that are written down.