How do companies decide who to lay off?

Companies decide who to lay off by eliminating positions first, then ranking the people in those positions against documented criteria such as tenure, performance, skills, or a weighted score.

Legal review follows, testing the list for disproportionate impact on protected groups before anyone hears the decision.

The decision runs top down in five stages, and the order matters more than the criteria.

Leadership sets a financial or structural target and finance and HR translate that target into roles.

Managers rank people within the affected roles, then legal and HR analytics test the resulting list.

1. Set the headcount or cost target, and name the business reason behind it
2. Identify the positions that disappear, by function, layer, or product line
3. Define the group each selection will be made from, which becomes your decisional unit
4. Rank people inside that group against written criteria
5. Test the list for disproportionate impact, then revise the criteria rather than swapping individuals

The failure mode sits at stage four with managers who receive a ranking request without the rationale behind it, so they rank on impression rather than evidence.

Careerminds coaches see the consequence downstream, when participants arrive unable to explain why they were selected and the employer cannot explain it either; a list that nobody can articulate is a list that will not survive a challenge.

Companies lay off employees for business reasons rather than individual fault, most often restructuring, cost pressure, or the loss of a product line.

Reorganization is the single most common trigger as 50% of HR professionals whose organization conducted a reduction in force cited reorganization as the reason, with business strategy change at 37% and poor business performance at 36% (SHRM, March 2025 Current Events Pulse, 2025).

The reasons cluster into five groups:

• Structural change, including reorganization, merger overlap, and closures
• Cost reduction, where labor is the largest controllable expense
• Demand decline, from lost contracts, falling sales, or market contraction
• Technology substitution, where automation or AI absorbs the work
• Correction after overhiring against growth that did not arrive

Technology has moved up that list fast with 57% of HR leaders say their organization is likely to conduct layoffs in the next 12 months, and those describing layoffs as very likely name increased AI use as the top reason (Careerminds, Responsible Workforce Change in the Age of AI, 2026).

Volume gives us the context with US employers announcing 1,206,374 job cuts in 2025, a 58% increase on 2024 and the highest annual total since 2020 (Challenger, Gray & Christmas, 2025 Year-End Job Cuts Report, 2026).

The stated reason now carries legal weight.

Several states require employers to disclose whether AI or technological change drove the cuts, so the public rationale and the internal selection file need to agree.

Decide which roles disappear before deciding which people leave, because position-level decisions are easier to justify than person-level ones.

A layoff removes a job for business reasons whereas a dismissal removes a person for conduct or performance.

Understanding the difference between a reduction in force and a layoff matters here, because the scope of the decision determines which rules apply.

Work through the remaining scope first and list the work that survives, then the roles required to deliver it and then the overlap.

Where two roles collapse into one, the question becomes which person can absorb the combined duties.

Before cutting, run the inventory you will wish you had run later.

32.9% of HR professionals say their organization lost critical skills when it laid off employees, and 52.1% say their organization rehired for eliminated roles within six months (Careerminds, AI-led layoffs: What HR leaders wish they knew, 2026).

Test alternatives to layoffs at this stage rather than after the list exists.

Hiring freezes, reduced hours, voluntary programs, and internal transfers all shrink the number of involuntary selections you have to defend.

Open a voluntary program to everyone and your strongest people, who have the most options elsewhere, volunteer first.

Five methods cover almost every layoff selection decision, and each trades business value against defensibility.

Method	Basis	Legal defensibility	Business value	Use when
Seniority	Length of service	High, objective and consistent	Low, ignores contribution	Union agreements apply, or performance records are weak
Employee status	Contingent and part-time first	Moderate, depends on classification	Low to moderate	Contingent labor is genuinely supplementary
Merit	Documented performance	Moderate, depends on record quality	High	Reviews are historical, consistent, and calibrated
Skills	Capability the business needs next	Moderate, needs written definitions	High	The business model is changing
Weighted ranking	Multiple scored factors	High when factors are objective	High	Large reductions span comparable roles

A written takeaway matters more than the grid.

Seniority is the easiest method to defend and the weakest at retaining the people you need, while merit and skills do the opposite unless your performance records predate any talk of a layoff

Last hired, first released.

The method is simple to apply and simple to explain, which is why collective bargaining agreements favor it.

It also tends to protect older workers, which lowers exposure to age discrimination claims.

That protection is not automatic and the EEOC illustrates the risk with a workforce that is 30% women producing a layoff list that is 85% women, purely because recent hiring skewed female (U.S. Equal Employment Opportunity Commission, Avoiding Discrimination in Layoffs or Reductions in Force, 2025).

A strategy lead hired four months ago scores identically to a weak hire from the same month.

Contingent and part-time workers go first.

Administration is lighter and employer brand damage is usually smaller, since the arrangement was understood to be temporary.

Misclassification and joint employment claims still follow contingent workers into a reduction, and contracts may carry notice or payment terms of their own.

The method also fails where contingent labor is not supplementary, because organizations running core delivery through contractors cut their own capacity first.

Performance records decide who stays.

This is the most commonly used approach and the one that generates the most litigation, because the record rarely supports the decision.

Ratings must be historical, contemporaneous, and calibrated across managers.

Ratings adjusted after the layoff decision are the clearest red flag an opposing lawyer can find.

Merit also ignores business weighting.

Cutting equally across a high-margin team and a cost center treats unequal contribution as equal.

Keep the capabilities the business needs next, rather than the ones it needed last year.

Define the skills in writing before scoring anyone, otherwise the method becomes preference with a spreadsheet attached.

Ask what breaks if this function halves, and ask the managers closest to the work rather than the executives furthest from it.

The risk sits in the age proxy, because scoring for recent technical skills can map closely onto recent graduates, and that correlation shows up in any later analysis.

Score each person across several factors, weight the factors, and cut upward from the lowest total.

This is the hardest method to build and the strongest to defend, because the reasoning is visible on paper.

Tenure, documented performance, certifications, and defined skills all survive scrutiny, because each one can be evidenced from records that already exist.

Potential for promotion and risk of being poached by a competitor do not survive it, because both are predictions about a person rather than facts about their work.

A selection matrix converts judgment into a score that someone else can audit.

Build it in four moves: define the group, choose the factors, set the weights, then score every person in the group on the same scale.

Tenure is the easiest factor to standardize:

• Under one year scores 0 points
• One to five years scores 1 point
• Six to ten years scores 2 points
• Over ten years scores 3 points

Weight the factors by what the business needs rather than by what is easy to measure, so tenure might carry double weight where institutional knowledge is the priority.

Three habits keep the matrix defensible once someone questions it:

• Record the organizational chart before and after, so the position-level logic survives the people-level questions.
• Score everyone in the decisional unit, not only the people you expect to cut.
• Keep the scoring sheet alongside the business case memo.

The matrix also doubles as a redeployment shortlist, because the same scores that identify who scores lowest against current criteria identify who scores highest against roles opening elsewhere.

Seniority holds up better under legal scrutiny, while performance produces better business outcomes.

Objective criteria win because they are consistent, and consistency is what a court examines.

Performance-based selection survives scrutiny only where the ratings predate the layoff decision, every manager applied the same standard, and someone calibrated the results across teams.

Most organizations fail the second of those, because rating inflation varies by manager and nobody notices until the scores sit side by side.

Use position elimination to do the heavy lifting, then apply a weighted matrix inside the surviving group, with objective factors carrying the most weight.

Compare each group’s share of the layoff list against its share of the workforce, then investigate any group that appears over-represented.

The EEOC recommends testing alternative criteria such as productivity, profitability, or expertise where a group is disproportionately affected.

The four-fifths rule gives you a screening threshold.

Where one group’s retention rate falls below 80% of the highest group’s retention rate, treat that as a signal to re-examine the criteria (U.S. Equal Employment Opportunity Commission, Uniform Guidelines on Employee Selection Procedures, 29 CFR 1607.4(D), 1978).

Passing the four-fifths screen is not a clearance, because small groups make the ratio unstable and courts also examine statistical significance.

Three checks sit alongside the headline analysis:

• Screen for anyone currently on or recently returned from protected leave
• Screen for anyone who recently raised a complaint, to avoid a retaliation inference
• Re-run the analysis after every change to the list, because each swap moves the numbers

Fix problems by changing the criteria, never by substituting individuals on the basis of their demographics.

Swapping names to balance the list creates the disparate treatment claim you were trying to avoid.

Two federal frameworks constrain layoff selection, and both are triggered by decisions you have already made by this point.

The federal WARN Act applies to employers with 100 or more employees and requires 60 days of notice for a plant closing affecting 50 or more people, or a mass layoff affecting 50 to 499 people who make up at least a third of a site, or 500 or more regardless of proportion.

Separate cuts within a 90-day window can be aggregated.

Sixteen states operate their own notice laws, several with lower employer thresholds or 90-day periods, so check the WARN Act notice requirements for every state you operate in.

The Older Workers Benefit Protection Act governs any release of age claims signed by an employee aged 40 or over.

In a group termination program, employees receive 45 days to consider the agreement and 7 days to revoke after signing.

The disclosure requirement reaches back into selection.

You must publish the decisional unit, the eligibility factors, and the job titles and ages of everyone selected and everyone in that unit who was not selected.

That document shows every older employee your selection pattern by age, laid out in a table.

Define the decisional unit carelessly and the pattern you did not notice becomes the pattern they see first.

Four developments since 2024 affect how layoff lists should be built and documented.

• Disparate impact enforcement narrowed: Executive Order 14281, signed on 23 April 2025, directed federal agencies to deprioritize disparate impact liability, and the EEOC moved to close pending charges resting solely on disparate impact. The statute did not change. Private plaintiffs and state agencies can still bring these claims, so adverse impact testing remains the right practice.
• Majority-group claims got easier: In Ames v. Ohio Department of Youth Services, decided unanimously in June 2025, the Supreme Court removed the extra evidential hurdle that majority-group plaintiffs previously faced. Any employee selected in a reduction now sues on the same footing.
• The threshold for Title VII harm dropped: Muldrow v. City of St. Louis, decided in April 2024, requires a plaintiff to show only some harm rather than significant harm. Transfers, demotions, and reassignments used as layoff alternatives became easier to challenge.
• AI disclosure arrived: New York added an AI question to its WARN filing in 2025, Connecticut requires AI-related disclosure from October 2026, and California requires technology displacement disclosures from January 2027.

Effective dates vary between sources for several of the newer state laws.

Have employment counsel confirm the position in your states before relying on any date here.

Recent hires, roles tied to paused growth, and redundant management layers tend to go first, with performance and cost deciding who goes inside those groups.

The tenure pattern is the strongest signal in the data.

An analysis of roughly 17,000 laid-off workers found their average length of service was about one year, far shorter than that of colleagues who stayed (Revelio Labs, Who Gets Laid Off, 2022).

Four patterns show up consistently:

• Recruiting and talent acquisition, because when hiring stops the function has less to do
• Management layers, where delayering removes a tier rather than individuals
• Acquired-company leadership, which shows the highest attrition after a merger
• Functions tied to a discontinued product or a closed site

Why those groups, rather than the weakest performers, often reflects how the selection was framed in the first place.

“The data shows that many organizations made restructuring decisions based on job titles rather than the full capabilities of their workforce. Employees often bring valuable skills that extend far beyond their current role, but if companies don’t have visibility into those capabilities, they risk eliminating talent they actually still need.”

Amanda Augustine, Careerminds’ resident career expert (CPCC)

Scale puts the pattern in context.

US employers recorded 21.2 million layoffs and discharges across 2025 (U.S. Bureau of Labor Statistics, Job Openings and Labor Turnover Survey, 2026).

Compensation band influences the decision but rarely appears as a documented criterion, because selecting on salary invites scrutiny as an age proxy.

Document the reasoning, prepare the mechanics, then plan the communication before anyone hears anything.

The selection work produces three artifacts worth keeping together: the business case, the scoring sheet, and the organizational charts from before and after.

Final pay deadlines, severance terms, benefits continuation dates, and notice obligations all vary by state, so confirm each one for every location the layoff touches.

A layoff plan template covers who tells whom, in what order, and what each person receives in writing on the day.

The sequence matters most to the people who stay, and research tracking employees across 146 companies found company confidence fell 16.9 points after a layoff (Harvard Business Review, Research: The Long-Term Costs of Layoffs, 2024).

A defensible list protects the organization legally.

How you execute it determines whether the remaining team stays productive.

• Decide which positions disappear before deciding which people leave, because position-level reasoning survives scrutiny that person-level reasoning does not.
• Objective criteria such as tenure and documented skills defend better than performance ratings, unless those ratings predate the layoff and were calibrated across managers.
• Define the decisional unit deliberately, since it determines both who gets scored and what your age disclosure will reveal.
• Test the list against workforce composition before it is final, and correct problems by revising the criteria rather than substituting people.
• Run a skills inventory first, because organizations that cut by job title routinely rehire for the same roles within months.

These three questions decide how a layoff is classified, what notice you owe, and whether performance belongs in the selection at all.

No. A layoff ends employment for business reasons such as a role being eliminated, while a firing ends employment for conduct or performance.

The distinction affects unemployment eligibility, what you say in the meeting, and how the separation reads if challenged later.

The federal WARN Act requires 60 days of notice for larger layoffs at a single site.

A number of states operate their own notice laws with lower employer thresholds or longer notice periods, so check both federal and state requirements for every location affected.

A layoff should rest on a business need rather than an individual’s performance.

Where performance forms part of a weighted selection, the ratings must predate the layoff decision and apply consistently across everyone in the same group.

This article provides general information and is not legal advice. Layoff rules vary by state and by situation. Consult employment counsel before acting.

Planning a reduction and want the selection criteria reviewed before the list is final? Speak with a Careerminds expert.