How to build a 2027 mobility budget that survives Q2

It is budget season, so somebody in finance has asked you for a number.

You will give them one. The question is where it comes from.

For most mobility teams the answer is last year's actuals, adjusted. Add a percentage for inflation. Add a bit more if the business is growing. Take a bit off if someone has told you to find savings. Submit, defend it in a meeting, move on.

That number is usually wrong, and it is usually wrong in the same direction. It holds until about April, when the first assignment lands somewhere nobody modelled and the variance conversation starts.

Last year is a poor guide to next year's mix

The budget is not wrong because the arithmetic is bad. It is wrong because last year's spend answers a question nobody asked.

What you spent in 2026 reflects the moves you actually made: those destinations, those assignment types, those policy tiers, those people. Your 2027 spend will reflect a different set. A long-term assignment to Switzerland and a six-month deployment to Vietnam are not the same cost. No inflation percentage turns one into the other.

Three things change the mix faster than any of them change the rate:

You cannot budget what you cannot see

There is a harder problem underneath the mix problem.

Source: AI Inflection Point report.

If you cannot say where people worked last year, your actuals are not actuals. They are the portion of spend that happened to pass through systems you can query. Business travel booked outside the program. Remote work approved by a line manager. Short deployments signed off by project delivery. None of it shows up as mobility spend. All of it creates obligations somebody pays for.

So the budget inherits a gap. You forecast the visible programme and absorb the invisible one as variance.

Build from scenarios, not from history

The fix is not a better spreadsheet. It is changing the input.

Instead of starting with what you spent, start with what the business is planning to do, and model it. Ask the regions and functions what moves they expect. You will not get a precise list. You do not need one. You need the shape: roughly how many moves, roughly where, roughly what type.

Then model that shape properly. Not a per-diem estimate. A full employer cost for each scenario: gross compensation, employer tax and social security in both jurisdictions, allowances, housing, mobilisation, and any gross-up. Build three or four representative profiles rather than fifty individual ones. A long-term assignment to a high-cost location. A short-term project deployment. A permanent transfer. A commuter arrangement.

Multiply the profiles by the expected volumes. That is your budget. It is built on what each kind of move actually costs, not on what you happened to spend last year.

Source: AI Inflection Point report.

The appetite for working this way is not the obstacle. The obstacle is usually that modelling one scenario takes long enough that modelling four never happens.

Budget the exposure, not just the package

The second thing most mobility budgets miss is everything that is not the package.

Tax equalisation settlements that land a year after the assignment started. Shadow payroll corrections. Withholding in jurisdictions nobody registered in. Permanent establishment exposure from a project team that stayed longer than planned. Work authorisation remediation when somebody travelled on the wrong status.

None of these are line items when the budget is written. All of them are costs when they arrive. They usually arrive in a different year to the decision that caused them.

You will not forecast these precisely. If you can see how many days each person spent where, you can see which thresholds you are close to. Then you put a number against the risk instead of discovering it.

Five questions to answer before you submit

Run these against the number you are about to hand finance.

1. Does the budget reflect the moves the business is planning, or the moves you made last year?If it is the second, you are forecasting the past.
2. Can you show the assumptions behind each figure?Finance rarely loses confidence because a cost varied. They lose confidence when nobody can explain why. Defensible beats precise.
3. Which destinations in the plan have you not deployed to recently?Those figures are the most likely to be wrong. They deserve a proper model, not an analogue from a neighbouring country.
4. What is in the budget for compliance exposure, not just packages?If the answer is nothing, the answer is that it will come out of contingency or out of next year.
5. If the business adds ten moves in March, do you know what they cost?If answering that takes two weeks, you do not have a budget model. You have a budget document.

Where to start

You do not need a new process to improve next year's number. You need a better input.

Start with three or four representative scenarios and model them properly. Once you know what each kind of move actually costs, the budget builds itself, and so does the answer when finance asks how you got there.