Before You Cut Headcount for AI: Seven Questions for CFOs and CHROs

A proposal to cut headcount because of AI usually arrives with a savings figure attached and little else. The Conference Board’s latest survey of U.S. CHROs suggests why. Of the CHROs surveyed, 82% have changed workforce planning because of AI, but half call the change slight, only 19% have built expected AI effects into enterprise-wide workforce and financial planning, and the top barrier, cited by 41%, is difficulty predicting AI’s workforce impact (Conference Board).

The cost of getting it wrong is now being measured. Careerminds, which sells outplacement services, surveyed 600 HR leaders who had overseen layoffs in the past 12 months, and 35.6% had rehired for more than half of the eliminated roles and 32.9% said they lost critical skills (CPA Practice Advisor). The seven questions below are meant for the meeting where a cut is first proposed. Finance and HR should answer them together and write the answers down.

1. Is the work automated today, or only exposed?

Exposure tells you AI could perform some of a role’s tasks. Automation means a tool is doing them now, in your systems, at an error rate you accept, with a named person accountable for checking the output. The gap between the two is wide. The Anthropic Economic Index found that 49% of jobs have seen AI used for at least a quarter of their tasks, and it cautions that high task coverage does not necessarily mean a large job impact once the time each task takes and the AI’s success rate are factored in.

Many cuts skip this step. In a survey of 300 U.S. HR managers by Orgvue, a workforce planning software company, 23% of those whose organizations had made layoffs said they relied on broad assumptions about AI’s capabilities rather than detailed, role-specific analysis (Orgvue). A workable test is to list the role’s main tasks, mark the ones a production tool handles today, and estimate the share of working hours they represent. A six-week vendor pilot does not count yet.

2. Can these people move to adjacent roles?

A role that shrinks does not have to mean a person who leaves. Mercer’s Global Talent Trends 2026 found that 65% of executives expect 11% to 30% of their workforce to be redeployed or reskilled because of AI within two years, and 63% of employees said they would trade a 10% raise for the chance to build AI and digital skills (Mercer). Firms expect movement in both directions. In the NBER survey of nearly 9,000 executives, more than 40% anticipated AI-driven employment gains or losses that roughly offset only in aggregate (NBER).

Before any notice goes out, match the skills of people in affected roles against open requisitions and the roles your plan says will grow. Standard Chartered, which plans to cut more than 15% of back-office roles by 2030, said affected employees would receive retraining support or packages to help them move into other opportunities (Private Banker International). Our guide Redeploy Before You Replace covers how to run that match.

3. What will replacement and rehiring really cost?

Savings models usually count the salaries removed and rarely count the cost of bringing work back. In the Careerminds survey, 52.1% said their organization rehired for eliminated roles within six months, and for 30.9% of U.S. organizations rehiring costs exceeded the layoff savings, while 42.4% broke even. Orgvue separately found that 32% of organizations that cut staff on the strength of AI’s promised savings had since rehired.

4. What do federal and state WARN rules require?

Federal WARN covers employers with 100 or more employees, not counting part-time staff. It requires 60 days’ written notice before a plant closing that causes job losses for 50 or more employees at a single site, or a mass layoff of 500 or more, or of 50 to 499 when they make up at least 33% of the site (29 U.S.C. 2101, 2102). Several states set lower thresholds or longer notice.

Rule	Who is covered	Notice	What else to know
Federal WARN	100 or more employees, excluding part-time	60 days	Notice goes to employees or their representatives, the state and the local chief elected official
New York	50 or more full-time employees in the state	90 days	Covers closings affecting 25 or more, and layoffs of 25 or more full-time staff making up at least 33% of a site, or 250 or more
California	Establishments with 75 or more employees	60 days	Covers layoffs of 50 or more within 30 days at any percentage, plus closures and relocations of 100 miles or more regardless of headcount; from 2026, notices must describe how you will support laid-off workers
New Jersey	100 or more employees	90 days	Covers layoffs of 50 or more across the employer’s New Jersey locations; mandatory severance of one week per full year of service, plus four weeks if notice is short

New York deserves extra attention for AI decisions. Since March 2025 its filing has asked whether “technological innovation or automation” contributed to a layoff, and a pending bill, A9533A, would require advance notice and a 90-day transition period with wages or retraining for automation-driven displacement at employers with 50 or more full-time staff (Hunton Andrews Kurth). Public companies should also note that securities disclosure duties can arise when planned terminations will create material charges (Harvard Law School Forum).

5. What will employees, customers and regulators read into it?

Employees will read any AI-linked cut as a signal about their own jobs. A record 27% of U.S. workers already worry technology could make their jobs obsolete (Gallup). Gallup also found that workers who come to expect their job will be eliminated score lower on engagement and satisfaction, higher on burnout, and are 4.1 points more likely to be actively looking for a job (Gallup Workplace).

Customers also notice when service changes. Klarna’s chief executive told Bloomberg in May 2025 that its AI chatbots produced lower quality output than human staff (Entrepreneur). In Australia, Commonwealth Bank reversed a plan to replace 45 customer service roles with an AI voice bot after its union disputed the bank’s claim that call volumes had fallen and took the matter to the Fair Work Commission. Affected workers were offered their jobs back or a voluntary exit (American Banker, Cyber Daily).

6. What does the board need to see?

Many boards are not yet getting this view. NACD’s 2026 survey of public company boards found that fewer than half have reviewed how AI and the related workforce changes affect talent strategy, and it calls for reporting that goes “beyond headcount and turnover” to show capability gaps, reskilling needs and leadership pipeline risks (NACD). Michael Peregrine of McDermott Will & Emery writes that the board’s core task is to confirm that AI-driven layoffs reflect a measured corporate strategy rather than a narrow focus on expense reduction, and suggests holding management accountable for “structured and equitable employee transition” (Harvard Law School Forum).

A useful board pack answers the first five questions on one page per role family. It shows the tasks automated today, the people affected and their redeployment matches, the full-cycle cost, the legal timeline and the communication plan, with sources and dates for each figure.

7. How will you know whether the AI delivered?

Decide the measures before the cut. Bain’s 2026 survey of 102 CFOs found that 48% ranked speed and cycle-time reduction as the biggest AI win, compared with 34% who chose headcount or cost savings, and Bain recommends treating time-to-insight and time-to-action as primary outcomes alongside cost (Bain). Executives in the NBER survey reported that AI raised annual labor productivity growth by 0.2% over the last three years, a modest figure that argues for measuring your own results rather than assuming them.

Set baselines for output, error rates, cycle time and customer outcomes, and review them at six and 12 months. Name in advance the results that would lead you to pause, rehire or expand. Only 8.4% of HR leaders in the Careerminds survey said their AI restructuring met its goals and that they would repeat it unchanged, so build a review point into the plan from the start.

How do the seven questions work together?

Answered honestly, the questions can change the decision. Some roles turn out to be ready for real change, and others are better served by redesign and redeployment. Either way, the written record shows employees and the board how the call was made. For the commitments that make the people side work, see What Good Employers Commit to During an AI Transition, and for recent examples, AI Layoffs in 2026.

If you want a head start on the first two questions, JobRoute will build a free AI exposure sample for 10 of your roles in two business days, showing which tasks in each role AI is already being used for, according to public AI usage data, and where people could move. You can request the sample, or book a 15-minute call if you would rather talk through a decision that is already on the table.