Before a spreadsheet ranking becomes a termination list, HR needs to verify the data behind each score
A layoff spreadsheet can make a difficult decision look settled. Employees appear in rows, performance factors appear in columns, and a formula produces a ranking. Before that ranking becomes a list of people who will lose their jobs, HR needs to understand what the numbers represent and how they were produced.
Consider a hypothetical California employer reducing a customer support team. Its selection model combines attendance, tickets resolved and manager ratings. One employee ranks near the bottom. The spreadsheet does not show that part of the attendance period involved protected leave, that the employee handled unusually complex tickets, or that two managers used different rating standards. Those facts do not determine the outcome. They show why the total needs an explanation.
Start with the work that will remain
Define the business change before choosing the people. Which services will continue, which roles will disappear, and what capabilities will the remaining team need? A selection process is harder to explain when the organization identifies people first and constructs criteria around them afterward.
Define the comparison group as carefully as the criteria. Employees with similar titles may perform different work. Conversely, different titles may conceal substantially overlapping responsibilities. Record why particular roles are compared and why others are excluded. A credible explanation connects the group to the restructuring plan rather than treating the existing reporting chart as the only possible answer.
Make each input understandable
For every factor, identify its purpose, source, measurement period and limitations. A productivity figure should say whether it measures cases closed, revenue generated or something else. It should also explain how the measure relates to the work the business expects to retain.
Check whether employees had comparable opportunities to produce the result. A smaller sales territory, changed assignment or prolonged system outage may affect a number without showing weaker performance. Establish how incomplete data will be handled. Treating a blank cell as zero is a substantive decision, even when it happens through a default setting.
Apply the same care to manager ratings. Ask what observable behavior supports each rating, whether reviewers used the same scale and whether a second reviewer can follow the explanation. Calibration should test evidence and consistency. It should not become an exercise in adjusting ratings until a preferred list emerges.
Check leave and accommodation issues before scoring
Attendance and availability measures need particular attention. Review whether the data includes leave or circumstances that require different treatment under applicable law. Keep sensitive medical information within authorized channels; the selection worksheet generally needs the correct treatment of the input, not an employee’s diagnosis.
The U.S. Department of Labor explains that taking FMLA leave cannot be used as a negative factor in covered employment actions. That protection depends on the law’s coverage and the employee’s circumstances. An employee who uses protected leave is not automatically exempt from a genuine reduction in force, but the employer cannot simply penalize the leave through an attendance score. State leave and disability protections may require separate analysis.
In the hypothetical support team, HR should determine what the attendance field actually counts before the ranking is approved. If a protected absence was counted against the employee, address that input and any resulting score through a consistent review. Do not solve the problem with an unexplained exception that leaves the same defect in everyone else’s data.
Understand what the technology contributes
A weighted spreadsheet, a predictive model and a manager’s recommendation are different tools. Identify which parts of the selection process each one influences. Who chose the weights? Does a vendor’s score reflect the employer’s actual criteria? Can the reviewer explain what a change in the score means?
For a hypothetical California workplace, this also calls for checking whether the system falls within applicable automated decision system rules. California’s Civil Rights Department says its regulations clarify how existing employment discrimination protections apply to these systems. Using a tool does not transfer responsibility for a discriminatory decision to the software. An ordinary spreadsheet should not automatically be described as AI; assess its functions and the applicable definitions.
Where a tool cannot provide a meaningful explanation, record that limitation and obtain appropriate technical and legal review before relying on its output. A manager clicking approve does not demonstrate that anyone tested the inputs. Human review becomes useful when the reviewer has access to evidence, understands the limits and has authority to question the recommendation.
Review the proposed results before final approval
After applying the criteria, examine who would be affected. The EEOC recommends reviewing a proposed reduction in force for disproportionate effects on groups protected by federal employment discrimination laws and considering whether alternative criteria can meet the business need with less impact.
This review needs context. A demographic difference is a reason for careful analysis, not a complete legal conclusion. Work with qualified advisers on the appropriate comparison group, available data and applicable standards. Test legitimate alternatives against the same business objectives. Do not change individual scores because of protected characteristics or assume that a favorable overall percentage resolves every concern.
Return to individual records as well. A consistent formula can still apply inaccurate information consistently. Check disputed entries and unusual results before the process closes. If a correction changes the ranking, preserve both the original and corrected versions and explain the reason. Avoid quietly replacing a worksheet that later reviewers may need to understand.
Keep a record that explains the decision
A useful record connects the business objective, comparison group, criteria, evidence, review and final approval. It identifies who resolved disputed inputs and why. Separate the business reason for reducing roles from the reasons particular employees were selected. Both matter, and one does not automatically explain the other.
Plan how employees can raise a factual error through an appropriate channel, and who will assess it. This is a practical process recommendation, not a claim that every employee has the same legal appeal rights. Observe applicable notice, record retention, confidentiality and consultation requirements with the help of advisers.
The final question is straightforward: could someone who did not build the spreadsheet understand why this employee was selected? If the answer depends entirely on a total, the review is unfinished. HR needs an explanation grounded in the work, the evidence and the rules that apply before a score becomes a decision.
--------
George S. Azadian is a California employment lawyer and President of Azadian Law Group, PC. His practice focuses on representing employees in wrongful termination, discrimination, retaliation, and wage-and-hour disputes.
