
How to prove the value of AI in global mobility
A K2 Thinking Paper, November 2026
It's no surprise, then, that relocation is fertile ground for AI. AI-supported systems can retrieve policy information, draft communications, triage enquiries, flag incomplete records and summarise case information. Agentic systems may soon co-ordinate whole workflows under human supervision. But how do you know whether any of this is genuinely improving relocation outcomes?

Download the full thinking paper
Our new thinking paper, How to prove the value of AI in global mobility, makes an important distinction between activity and value. A chatbot answering a question is an activity. When that answer is accurate, arrives faster than a human could manage and fully resolves the employee's query, that's an operational outcome. Only when the outcome leads to greater employee confidence, lower service costs or fewer delays does it start to count as value.
The paper also highlights the risks of skipping steps. A chatbot that provides half an answer is frustrating. An agentic system that automatically triggers the wrong relocation benefit is a much bigger problem. Relocation data is spread across HR systems, suppliers, immigration providers and payroll, so the accuracy of the data behind an AI tool matters as much as the tool itself. EY's Mobility Reimagined Survey found that while 72% of mobility teams are scaling generative and agentic AI, only 51% trust their data enough to move to the next phase.


Efficiency gains need careful scrutiny too. If an AI tool halves the time spent on a process, but another team now has to review every output, time hasn't been saved; it's simply been redistributed. And where extra capacity is created, has it been redeployed into more cases or more tailored support for complex moves?
The paper sets out:
- Four pillars for measuring AI value in mobility
- Why AI maturity is constrained by data maturity
- A practical six-stage framework, including how to use A/B testing to compare AI-supported and existing processes
- How to calculate net value, accounting for build costs, maintenance and the human cost of change
Most importantly, it explains why the "humans in the loop" remain essential. AI can speed up a process, but it can't reassure a family worried about schooling, or guide an employee whose lump sum turns out to be worth less after tax than expected.
Download the full thinking paper
