Redeploy Before You Replace: How to Move People Into New Roles

When a board asks what AI means for headcount, the fastest answer is a reduction, and many companies come to regret it. In a survey of 600 HR leaders who oversaw layoffs in the previous 12 months by Careerminds, a career transition provider, reported in July 2026, only 8.4% said their AI-driven restructure delivered on its promises. A third said they lost critical skills, and 35.6% had already rehired for more than half of the eliminated roles.

Redeployment means moving people whose work is shrinking into roles where their skills still count, and most large employers already intend to do it. The World Economic Forum’s Future of Jobs Report 2025 found that 51% of employers plan to transition staff from declining to growing roles internally, and that for every 100 workers, employers expect 19 to be reskilled and redeployed within their organization by 2030. This playbook covers how to do it at the pace AI is changing the work.

Why redeploy people instead of hiring new ones?

The person you already employ knows your customers, systems and rules, and replacing that knowledge is slow and expensive. SHRM’s 2025 benchmarking puts the average cost per hire at $5,475 for a non-executive role and $35,879 for an executive, before any ramp-up time. Internal moves also keep people longer: LinkedIn reported that employees who made an internal move had a 75% chance of staying at their company after two years, compared with 56% for those who had not.

Executives expect the scale to be large. 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. Moving that many people needs the same planning discipline a company already applies to hiring.

How do you find adjacent roles by skill overlap?

An adjacent role shares most of the skills and work activities of the current one, so most of a person’s experience carries over. O*NET, the US Department of Labor’s occupational database, describes each occupation by its tasks and skills, which makes the overlap between any two roles measurable.

JobRoute’s analysis of public O*NET, BLS and Anthropic Economic Index data gives each occupation an AI exposure score from 0 to 100, an importance-weighted average of how much of each task AI can do. It also lists each occupation’s closest neighbors, with a skill-overlap score and an estimated transition time in months. Three examples show why overlap alone is not enough.

Current role	AI exposure	Closest moves (skill overlap, estimated transition)	Destination exposure
Insurance claims and policy processing clerks	77.9	Claims adjusters, examiners and investigators (74%, about 9 months)	68.5
Customer service representatives	44.1	Sales representatives of services (71%, about 10 months); patient representatives (61%, about 14 months)	60.5; 42.3
Paralegals and legal assistants	66.0	Legal secretaries and administrative assistants (78%, about 8 months)	66.3

Source: JobRoute’s analysis of public O*NET 30.3, BLS OEWS May 2025 and Anthropic Economic Index (June 2026) data. Exposure runs from 0 to 100; higher means more of the work is AI-doable.

Insurance claims clerks: a short step up

Claims and policy processing clerks are among the most exposed roles in the data, with 23 of their 25 tasks scoring 60 or higher and an overall exposure of 77.9. Their closest neighbor is claims adjusters, examiners and investigators, with a 74% skill overlap and an estimated transition of about nine months. The move lowers exposure to 68.5 and comes with a pay step: the BLS May 2025 wage estimates in JobRoute’s data put the median at $49,230 for clerks and $78,000 for adjusters.

This move will not shelter anyone for good. BLS projects adjuster employment to fall 5.5% between 2025 and 2035, and 26 of the adjuster role’s 29 tasks are also highly exposed. Treat it as the next step on a route, and build skills in the parts of the job that score lower, such as negotiating claim settlements, which scores 40 based on observed Anthropic Economic Index usage.

Customer service representatives: the average hides the core task

Customer service representatives score a moderate 44.1 overall, with only 4 of their 13 tasks at 60 or above. That average understates the change. The task O*NET rates most important for the role, conferring with customers by phone or in person to give information, take orders or handle complaints, scores 84.5, based on observed Anthropic Economic Index usage. A team that plans on the average will underestimate how much of the day changes.

The closest neighbor by skills, sales representatives of services (71% overlap, about 10 months), is more exposed at 60.5, although BLS projects it to grow 2.7% from 2025 to 2035 while customer service roles shrink 5.3%. Patient representatives are a lower-overlap move (61%, about 14 months), but more than half of their tasks, such as interviewing patients about problems with their care, score 40 or lower, and the BLS median wage for the broader occupation group that includes them is higher ($50,290 against $44,770). That is a realistic internal route for a health insurer or hospital system, though not for a retailer.

Paralegals: when the nearest role is a trap

Paralegals score 66.0, with 9 of their 12 tasks at 60 or higher. Their closest skill match is legal secretaries and administrative assistants, at 78% overlap and about eight months. In practice that move would put people into a role that is just as exposed (66.3), pays less (a median of $55,570 against $62,890) and is projected by BLS to shrink 5.2% by 2035.

The data also lists lawyers as a 64% skill match, which shows the limit of any overlap score, since no internal program replaces law school and a bar exam. For paralegals the better answer is often to redesign the role in place, shifting time toward coordinating case work and checking AI-drafted documents for accuracy. The WEF expects in-place change to be the larger group overall, with 29 of every 100 workers upskilled within their current roles compared with 19 redeployed.

The lesson from all three is to run two checks on every candidate move. Skill overlap tells you whether a person can get there, and the destination’s own exposure and demand tell you whether the move is worth making.

How long does retraining take, and how do you size it?

Treat transition estimates as planning ranges rather than promises. The months in JobRoute’s data estimate how long a typical person needs to close the gap between two roles, and your own numbers will depend on the person and the training design.

A simple way to size each move:

1. List the tasks in the destination role that the person does not do today, using the O*NET task list as a starting point.
2. Mark which can be learned on the job and which need formal training or a license.
3. Convert the gap into paid hours per week and a target date that the receiving manager signs off on.
4. Confirm that the destination has open positions or expected attrition in that window, so people finish training into a real job.

Fund the learning on paid time. BCG’s AI at Work research found that regular AI use is sharply higher for employees who receive at least five hours of training along with in-person coaching.

How do you pilot redeployment with a small cohort?

Start with one role pair where the data and the business agree, such as claims clerks moving into adjuster roles at an insurer that has adjuster openings. Recruit a small group of volunteers rather than assigning people, and name a receiving manager who wants them. Agree up front on what success means, such as the share who finish training and the share still in the role a year later.

Run the pilot long enough for the first group to finish training and work in the new role for at least a quarter. Talk to the participants throughout, because they will show you where the training was too thin and which managers quietly blocked moves, and fix those problems before you scale.

How do you make internal moves easier than hiring from outside?

In many companies it is easier to find a job somewhere else than down the hall. A Gartner survey reported by SHRM found that only 33% of job seekers had searched for a job at their current workplace first, only 51% were aware of internal openings, and only 17% said their manager helped them apply for internal jobs. A Gartner research director quoted in the same piece called talent hoarding by managers the No. 1 problem with internal mobility.

Practical fixes include:

• Post roles internally before externally for a set window, and tell exposed teams directly when relevant roles open.
• Drop the requirement for a current manager’s approval to apply, and make releasing people to new roles part of every manager’s goals.
• Screen internal candidates on skills rather than an exact title match.
• Give recruiters a target for internal fill rate in the functions where exposure is highest.

Should you protect pay during the transition?

Where you can, yes. A move into a lower-paid role, such as paralegal to legal secretary, will feel like a demotion unless pay is protected for a defined period. Where the destination pays more, as with clerk to adjuster, tie the increase to training milestones so the budget stays predictable.

Either way, put the terms in writing before the move, including pay during training and what happens if the new role does not work out. Ask employment counsel how pay protection fits your pay equity commitments and local rules.

What should you measure?

Report a small set of measures to the board next to headcount:

• Internal fill rate for roles in growing functions.
• Time to full productivity, compared with external hires in the same role.
• Retention of redeployed people at 12 and 24 months.
• Cost per move, including training and pay protection, against a layoff plus a new hire.
• The share of highly exposed roles with a named route, either a move or a redesign in place.

What does a successful redeployment look like?

Public examples are still rare, but IKEA’s is well documented. In 2023 Ingka Group, the largest IKEA retailer, told Reuters that it had trained 8,500 call centre workers as interior design advisers since 2021, while an AI bot named Billie handled 47% of customer queries to its call centres over two years. Asked whether AI would reduce headcount, Ingka’s global people and culture manager said that was not what the company was seeing at the time.

The pattern is worth copying. IKEA trained people into remote interior design, a channel Ingka told Reuters it aimed to grow to 10% of sales by 2028, while the bot took over routine queries. When you announce moves like these, our guide on how to tell employees about AI covers what to say first.

If you want to see where your own redeployment routes are, JobRoute maps them role by role. Send us 10 job titles and we will return a free AI exposure sample within two business days. You can request the free sample, or book a 15-minute call to talk through a fixed-fee AI Workforce Impact Assessment across your full roster.