Run the Workforce Plan on HR: What the Labor-Market Data Says About Your Own Team
AIHR data shows 59% of HR payroll sits in role families with 100+ candidates per opening, while two of HR's tightest talent markets get about 1.9%. Four checks for planning your own function.
Contents
The workforce plans you have run started the same way. What the business needs, and what the gap costs to close. You built the inventory, priced the options, took it to the CFO. Then you filed it and moved on, and the department that produced the analysis was not the one it got pointed at.
AIHR published HR Trends 2027 on 22 September 2026, and one number in it is a workforce planning finding about HR itself. Around 59% of HR payroll sits in role families with more than 100 available professionals per opening. HR Technology and Digital and People Analytics together account for about 1.9% of payroll, and AIHR puts both among HR’s tightest talent markets. AIHR calls the investment mix misaligned with changing capability demand.
Misaligned is AIHR’s word for it. Concentration is the word HR uses about any other function, and it belongs in an HR operating model redesign before the org chart does.
A concentration you would flag in anyone else’s plan
The rest of the release describes the market moving underneath that allocation. AIHR’s labor-market work with Revelio Labs covers 54 HR roles, more than 162,000 active US HR job postings and 3.88 million HR professionals. Demand for Organization Effectiveness Specialists rose 65.4% and Change Management Specialists 21.6%. HR Administrators fell 29.6% and HR Service Desk Agents 38.3%. For HR Technologists, five candidates are available per opening.
Check the clock on those figures. AIHR published all five of them in its HR Career Outlook 2026 back in March, drawn from seven months of Revelio data running August 2025 to February 2026. AIHR labels the Change Management movement as six-month growth and states no window for the other four. The September release does not say whether any have been refreshed. Read them as the shape of a shift already underway, not as this quarter’s news.
Erik van Vulpen, Co-Founder and Dean of AIHR, put it this way in the release: “AI is changing where HR creates value. The market is placing greater value on judgment, specialized expertise, and the ability to lead change.” Your payroll, if you are near the 59%, sits mostly in roles the market can refill from a deep bench.
That is a portfolio problem, and HR has a method for portfolio problems. HR runs that method on sales and on engineering, and rarely on itself.
What a 38% drop in postings measures
Demand for HR Service Desk Agents fell 38.3% in AIHR’s Revelio analysis. Revelio counts that demand in job postings, and the methodology note on AIHR’s March page calls these figures directional signals from posted demand rather than a complete measure of all hiring activity. Nobody stopped asking why their paycheck changed or which policy applies to a transfer between countries. Those questions still arrive, through more channels than before.
The work went somewhere. Some into HRIS workflows and self-service, the intended destination, and some into agents. The remainder went to whoever was standing closest, which in a newly reorganized function is the business partner hired to work on capability who spends Tuesday morning on a payroll question.
Set that beside the operating-model figures in the release. AIHR reports that 64% of HR functions reorganized inside two years. Only 19% report clear responsibilities across business partners, centers of excellence and shared services. Among those that reorganized, 60% provided no formal training on the new structure.
Those two findings belong together. When work leaves a named role without arriving at another named role, someone absorbs it, and nobody sees that until a capacity check finds your strategic roles doing service work. Unassigned volume is one reading of what sits behind that 19%. AIHR measures the postings and the operating model on different populations, so it is a reading, not something in the data.
An HR operating model redesign needs four checks first
1. Build the supply and demand view for HR roles
Headcount tells you how many people you have. Replaceability is a different question, and the AIHR data is built on the ratio that answers it: available professionals per opening. That ratio is what turns 59% of payroll into a finding.
Run it on your own org. For each HR role family, how many people could you hire, and how long would it take? The spread in AIHR’s own numbers is wide: five available candidates per opening for HR Technologists, against 757 per opening for VP of HR in the March Career Outlook. Those are different businesses, and they call for different retention and pay positioning.
Few HR functions can produce this view for themselves in an afternoon, which is itself informative, because they can produce it for the business. The obstacle is the one the business hires HR to solve: an org chart lists titles, and titles do not tell you what your people can do. Mapping how skills connect to roles and roles to people is what our knowledge graph does across a workforce, and HR sits inside that workforce. The Academy’s primer on taxonomic versus inferential skills intelligence covers why a fixed list of skill labels misses what people can already do.
2. Look at spend concentration before headcount
Concentration of spend moves slowly and binds hardest. If HR Technology and Digital plus People Analytics account for 1.9% of your payroll, no strategy deck changes what your function can do next year. Reallocation is a budget decision made a year ahead, not a hiring decision made when a requisition opens.
3. Separate the work that ended from the work that moved
Take your last restructure and list what the removed roles were doing. For each item: automated, absorbed by a system, absorbed by a person, or stopped. The “absorbed by a person” column predicts your next capacity problem, and it runs longer than anyone expects.
Work routed into a system has a named destination and a resolution rate you can measure. Work absorbed by a colleague has neither, and it surfaces later as attrition in a role you were counting on.
4. Name owners before you draw boxes
The 19% figure is a reason to do this in the opposite order. Decide who owns each category of work under the new model, write it down, and only then draw the structure that supports it. Then train people on it, which 60% of reorganizing functions did not do. A restructure that never names an owner moves the work somewhere quieter.
Buy-in is running ahead of capability
One more set of numbers from the release, across 961 HR teams and 13,665 professionals: 30% of HR professionals have received job-specific AI training. In a separate 334-team sample in the same release, 65% show strong AI buy-in and advocacy, 30% have a clear purpose, expected value and relevant use cases for AI, and 29% are confident they have the data, tools and infrastructure required to support AI at scale.
The posting data points the same way, though AIHR reads it differently. In an analysis of 540,277 skill-role entries run in November 2025 and published in March, AI skill requirements appeared in fewer than 1% of them. AIHR’s own conclusion is that not every HR professional has to become an AI expert. Set it beside the 30% training figure and a second reading opens up: HR has spent two years telling the business to plan for what AI does to work, and has written almost none of it into its own job specs. The redesign is landing in roles that still advertise the skills they advertised before.
Start with the money and the market
An HR operating model redesign usually starts with the structure question. The choices are well known: the three-legged stool that splits business partners, centers of excellence and shared services; a functional model organized around recruiting, rewards and learning; a matrixed model reporting into both the business and the function; and the hybrid variants that borrow from all three. Picking among them is the easy half. Two functions can run the same hybrid model and get opposite results because one of them has people it can hire and the other does not.
Start instead with the two questions you would ask about any other function. Where is the money, and how replaceable are the people it pays for? AIHR’s companion report, HR Priorities 2027, lands in the same place: two of its five priorities are redesigning HR’s operating model for human-AI delivery and refocusing HR investment on the capabilities with the highest return.
None of that needs a new methodology, only an existing one pointed inward, which is hard for reasons that have nothing to do with analytics. The planner and the planned are the same people, and the conclusions carry names.
Run it on your own function while the next reorganization is still a draft.
