Could you model a 5% labor-cost shift, by business unit, this quarter, without building anything new?
Over the last four years I’ve had the privilege of sitting with customers ranging from the Fortune 100 to new and fast-growing startups. Whether it was framed as People Analytics or Strategic Workforce Planning, the only consistent thing from one organization to the next was the inconsistency: the work was disconnected, running on bad or incomplete data, or worse, not happening at all.
Ask a supply chain team what a 5% change in materials cost does to margin, and they’ll have a model for it. Now ask what a 5% shift in that same labor cost does to margin, by business unit, this quarter, and somebody opens a new spreadsheet.
Workforce is the largest controllable cost line in almost every company and often the most precisely measured. It’s also the least modeled. Not because the tools don’t exist, but because the data underneath it (the jobs, the skills, the workers) was never governed or connected to the model.
Most workforce planning is really just position math. Headcount by cost center, times loaded cost, reconciled against Finance’s version once a quarter. That tells you what you spent. It doesn’t tell you what happens if demand shifts, a site scales up, or attrition spikes in one job family. Or whether the skills you keep hiring for could be built in-house.
Every other major cost line gets a model while the people line gets arithmetic.
There’s a second cost hiding under the first, and most leaders have stopped noticing it. When workforce data isn’t self-service, every question routes through HR or FP&A. A manager wants to know how their team’s loaded cost compares to plan, and the answer takes four days and two email chains.
We worked with a global consumer products company that had been live on Workday for more than a decade. Every month, FP&A analysts collected manager spreadsheets by hand and stitched them together just to show performance against plan. Managers who wanted their team’s numbers emailed HR and waited. We replaced the ritual with self-service dashboards and put HR and Finance planning in one model, so both sides now work from the same numbers. Nobody talked about the technology afterward. They talked about the hours they got back and the decisions that stopped waiting.
Now another cost line has the same problem. AI consumption is a real operating expense: metered to the unit, budgeted, and growing. Every agent you turn on, every AI-assisted decision, draws against an allocation you purchased, and most organizations will hit their first renewal with real consumption data on the table within the year.
Here’s the part most people miss: governing that spend is a people-data problem. Deciding which capabilities deserve the allocation, and knowing whether what you consumed came back as value, both depend on the data the AI runs against. AI on clean, connected workforce data returns value on every unit of spend. AI on stale job and skills data burns the same spend and gives you answers nobody trusts. The same foundation gap that keeps you from modeling your labor line keeps you from governing your AI bill.
And the spend is only half the AI story. AI is starting to decide where work goes: which tasks shift to agents, which roles change shape, where the hours land. That reallocation is where the real productivity gains live, and it runs straight through your labor line. If you can’t model that line today, you can’t plan the shift, and you can’t fund it with confidence.
STABILIZE: If the underlying data isn’t governed, start there. A model built on numbers nobody trusts is worse than no model. It just restarts the argument about whose numbers are right.
OPTIMIZE: If the data is sound, activate the self-service analytics and dashboards you already own. Managers get answers without a trip through HR and FP&A, and you’ll feel it in weeks, not quarters.
MODERNIZE: And when you’re ready to modernize, connect it all. Planning, analytics, and AI working on data both HR and Finance trust. One model, scenario-tested, maintained continuously, so the 5% question gets an answer instead of a spreadsheet.
Could you model a 5% labor-cost shift, by business unit, this quarter, without building anything new? Be honest. If the answer is no, you can pay your workforce to the penny, but you can’t plan it. And the AI spend you just turned on runs on the same ungoverned people data.