How Much of Your Payroll Could AI Actually Automate?
Less than the headlines say, and the difference matters. Most of the widely-quoted numbers ("AI could automate 30-40% of jobs") are measuring automatable tasks, then reporting them as if they were automatable headcount. Those are very different quantities, and conflating them is how leaders end up either panicking their workforce or budgeting for savings that were never really there.
Hours saved is not the same as jobs eliminated
If a role is made up of ten distinct tasks and four of them are automatable, that doesn't mean 40% of the people doing that role can be cut. It usually means each person doing that role gets roughly 40% of their time back on those specific tasks, time that then gets absorbed by the remaining six tasks, redirected into higher-value work, or, in some cases, genuinely freed up to reduce headcount growth going forward.
Which of those outcomes actually happens depends entirely on what the freed-up time gets used for, and that's a management decision, not something the automation percentage tells you on its own.
Our own analysis of what percentage of HR tasks AI can actually automate across 100 roles found exactly this gap: the task-level number and the commonly quoted role-level number rarely match.
Why the bigger number gets repeated anyway
"40% of jobs at risk" is a much better headline than "40% of certain tasks within certain jobs, with actual headcount impact varying widely by role and management decision." The first one drives clicks. The second one is accurate. Most of what circulates publicly is the first kind, which means most leaders are calibrating expectations against a number that was never designed to represent payroll impact in the first place.
How to calculate the real number for your business
- Map tasks, not roles. Break down what people in each role actually spend time on, at the task level.
- Flag which tasks are genuinely automatable today, not "in theory," but with tools that exist and work reliably right now.
- Convert automatable tasks into hours, not people. How many hours per week, per role, does this free up?
- Multiply by loaded cost per hour to get a dollar figure. This is your real savings number, and it's almost always more modest and more defensible than a headcount-reduction claim.
- Decide deliberately what happens to those hours, redeployed, reduced, or reinvested, rather than assuming the automation percentage makes that decision for you.
Why the modest number is actually the useful one
A defensible, task-level payroll-impact number is something you can put in front of a board without it being picked apart, and something you can share with employees without it reading as a threat dressed up in cautious language. The inflated headline number does neither. It's dramatic enough to worry people and vague enough to be useless for actual planning.
GoFIGR's AI Impact Assessment calculates this at the task level across your workforce, so the number you take to your board, or use to plan your next twelve months of hiring, is one you can actually stand behind.
.jpg)
Helena Turpin spent 20 years in talent and HR innovation where she solved people-related problems using data and technology. She left corporate life to create GoFIGR where she helps mid-sized organizations to develop and retain their people by connecting employee skills and aspirations to internal opportunities like projects, mentorship and learning.
.jpg)
