An HR operating model describes how the HR function is organised to deliver its strategy: who advises leaders (HR business partners), who runs transactions (shared services and HR technology), and who owns expertise (centres of excellence in reward, talent, learning, employee relations). The classic three-pillar “Ulrich” model still underpins most large organisations, but in 2026 it is being reshaped by self-service technology, AI agents handling tier-one queries, and product-style teams that cut across pillars. The right model depends on scale, geography, business complexity and how much the organisation is willing to invest in HR technology.
An HR operating model is the answer to a deceptively simple question: when a manager has a people problem, who do they call, and what happens next? Behind that question sit decisions about structure, roles, technology, service levels and cost that determine whether HR is experienced as a strategic partner or as a helpdesk. This guide explains the standard model, the reasons it disappoints, the alternatives that have emerged, and a method for redesigning your own function. It pairs with our guides on building an HR strategy and organisation design.
What are the three pillars of the classic HR operating model?
HR business partners embedded with leaders, shared services (often with a technology and self-service layer) for transactions and queries, and centres of excellence that design policies and programmes in specialist areas.
Why do HRBP models underperform?
Because business partners get pulled into transactional work when shared services and self-service are weak, and because centres of excellence design programmes without enough operational input. The model only works when all three pillars are properly built.
What is changing in 2026?
AI agents and chat interfaces now resolve a large share of tier-one queries, shrinking shared-service headcount; product-based HR teams organise around employee journeys (hire, onboard, pay, grow, leave) rather than functions; and people analytics has become a pillar in its own right.
What is an HR operating model?
An HR operating model is the design of the HR function: its structure, roles, processes, technology, governance and service levels, arranged to deliver the people strategy at an agreed cost. It sits between the HR strategy (what HR must achieve) and the HR organisation chart (who reports to whom), and it explains how work flows from an employee or manager request to a resolved outcome.
Most operating model descriptions cover six elements: service catalogue (what HR provides), delivery channels (self-service, service centre, business partner, specialist), structure and roles, technology and data, governance and decision rights, and performance measures. When one of these is missing, the model fails predictably: a service catalogue without channels produces confusion about where to go; channels without governance produce inconsistent answers by country.
How does the three-pillar (Ulrich) model work?
The three-pillar model, popularised by Dave Ulrich in the late 1990s, separates strategic advice (HR business partners), specialist design (centres of excellence) and transactional delivery (shared services), with a technology layer enabling employee and manager self-service. Each pillar has a distinct customer, skill set and cost structure.
| Pillar | Customer | Core work | Typical ratio / cost driver |
|---|---|---|---|
| HR business partners (HRBP) | Executives and senior managers | Workforce planning, organisation design, talent decisions, change, senior employee relations | 1 HRBP per 500–1,500 employees; senior generalist pay |
| Centres of excellence (CoE) | The HR function and the enterprise | Reward, talent acquisition strategy, learning, DEI, employee relations policy, people analytics | Small specialist teams; consultancy-grade pay |
| Shared services / operations | All employees and managers | Payroll, data changes, queries, onboarding administration, benefits admin, case management | 1 agent per 300–800 employees before automation; location arbitrage |
| HR technology & self-service | Everyone | HRIS, case management, knowledge base, chatbots, workflow | Licence cost per employee; the cheapest channel per transaction |
The model’s logic is economic: push every transaction to the cheapest channel that can resolve it (self-service, then service centre), so that expensive HRBPs and specialists spend their time on work only they can do. The HRIS and case-management tools compared in our HRIS comparison are the backbone of that channel strategy, and payroll platforms in the payroll software comparison determine how much of the transactional load can be automated.
Why do HR business partner models so often disappoint?
HRBP models disappoint when the transactional layer is weak, so business partners become the escalation point for every query; when HRBPs are generalists renamed rather than developed into consultants; and when the organisation measures HRBPs on activity rather than business outcomes. The result is a role that is expensive, frustrated and perceived as administrative.
Fixing this usually means three moves. First, invest in shared services and self-service until the query volume reaching HRBPs drops below 10–15% of their time. Second, redefine the HRBP role around a small number of outcomes (workforce plans delivered, critical roles filled, engagement and retention in the business unit, change programmes landed) and staff it with people who can read a P&L and run a difficult conversation with a managing director. Third, create a clear route for managers to reach the service centre directly, so that the HRBP is not the default door. The capability side is covered in aligning HR strategy with business goals.
What alternatives to the three-pillar model exist?
The main alternatives are the product-based (or journey-based) model, which organises HR around employee journeys such as hire, onboard, pay, develop and leave; the agile HR model, which uses cross-functional squads and a small stable core; and the AI-augmented model, in which digital agents handle tier-one and much tier-two work, and human teams concentrate on judgement, relationships and design.
Product-based HR borrows from software organisations: each journey has a product owner, a backlog, metrics (time-to-productivity for onboarding, payroll accuracy, internal mobility rate) and a team drawn from what used to be separate pillars. It suits organisations with mature technology and a culture of continuous improvement, and it tends to expose the seams between CoE design and operational reality quickly. Agile HR is a lighter version, useful for change-heavy periods, but it struggles with statutory and payroll work that cannot be re-prioritised sprint by sprint. The AI-augmented model is less a structure than a reallocation: as chat agents resolve routine queries and generate documents, the shared-service tier shrinks and analytics, employee relations and change capability grow; our guides on agentic people analytics and CHRO bets on agentic AI track how far this has gone.
How do you size and cost the HR function?
HR headcount is usually sized against two benchmarks: HR-to-employee ratio (commonly 1:60 to 1:100 in mid-sized organisations, wider in large automated ones and narrower in small or highly regulated ones) and HR cost per employee, which varies by country and sector but is typically 1–3% of total workforce cost. Neither benchmark says anything about quality, so pair them with service and outcome measures.
A more useful approach is activity-based: estimate transaction volumes (hires, leavers, data changes, queries, payroll runs, cases), assign each to a channel with a unit cost, and derive the staffing per channel. This exposes where cost really sits (often in manual payroll and benefits administration) and where technology investment pays back. Country mix matters: employers with staff in many jurisdictions carry a fixed compliance overhead per country, which is why employer-of-record and global payroll platforms, compared in Deel vs Rippling, appear in so many operating-model business cases. Reward design costs are discussed in designing pay structures.
What role does technology play in the operating model?
Technology defines which channel an interaction goes to and therefore what the function costs. A single HRIS with clean data, an integrated case-management and knowledge tool, and increasingly an AI assistant for employees and managers, allow 60–80% of interactions to be resolved without a human in HR. Without them, the same interactions land on generalists at ten times the cost.
The 2026 shift is that AI assistants are moving from answering policy questions to executing transactions: initiating a leave request, drafting a contract amendment, or preparing a performance-improvement plan for manager review. This raises new governance questions (who approves what an agent does, how decisions are logged, how bias is tested) that the operating model must answer, and it changes the skill mix in shared services from processing to exception handling and quality control. Employee-facing tools are compared in our employee engagement software comparison and onboarding software comparison.
How should a multinational structure HR across countries?
Multinationals usually combine global centres of excellence (reward frameworks, talent processes, HR technology), regional shared-service hubs (payroll, benefits, queries, in a few languages and time zones) and local HR presence sized to statutory complexity, works council obligations and headcount. The design question is how much to standardise globally versus adapt locally.
Standardise the things that do not vary by law (job architecture, performance cycle, HRIS data model, talent reviews) and localise the things that do (contracts, payroll, working time, terminations, consultation). Countries with mandatory works councils or strong collective bargaining (Germany, France, the Netherlands, Austria, the Nordics) need more local HR capability than their headcount alone suggests. The 28 country guides in our Expat HR & Global Mobility hub set out the employment-law and payroll obligations that drive local staffing, and our centralisation vs decentralisation guide discusses the governance trade-offs.
How do you run an HR operating model redesign?
A redesign runs in five phases: diagnose (current cost, volumes, satisfaction, pain points), design (pillars or journeys, channels, decision rights), size (ratios, service levels, budget), enable (technology, data, AI agents) and transition (role changes, skills, governance, change management). Expect 9–18 months for a mid-sized organisation and longer where payroll or HRIS replacement is on the critical path.
Two practical warnings. Do not redesign HR in isolation from the finance and IT operating models; shared services in particular is frequently a cross-functional design. And treat the HR team itself as the workforce being changed: role clarity, reskilling and honest communication about which roles shrink are exactly what HR asks of other functions, and the credibility of the new model depends on HR living it. Our guides to managing organisational change and team structures and reporting lines apply directly.
What skills does the new HR operating model require?
The shift from processing to advising, and from manual to AI-assisted work, changes the skill profile of the function: business and financial literacy for HRBPs, data and analytics fluency across all pillars, product and service-design thinking in shared services, and employee-relations judgement that cannot be automated. Many HR teams find that a third of their existing roles change materially during a redesign.
Invest in a capability framework for HR itself, assess the team against it honestly, and fund development before the new structure goes live. The organisations that skipped this step in earlier waves of HR transformation ended up with new titles on old skills, which is where the disappointment with business-partner models usually began. Our guide to employee development and training applies as much to HR as to any other function.
Frequently Asked Questions
What is the ideal HR-to-employee ratio?
There is no single answer; 1:60 to 1:100 is common for mid-sized organisations, but automation, country mix and sector regulation move it substantially. Size by activity and service level rather than by ratio alone.
Is the Ulrich model outdated?
The logic of separating strategic, specialist and transactional work still holds. What has changed is the technology layer, which now absorbs much of the transactional tier, and the emergence of journey-based teams that cut across the pillars.
Do small companies need an operating model?
Yes, in a simplified form: a clear statement of what HR does, who handles what, which tools employees use, and what happens when the company doubles in size. Most companies below 150 employees run a generalist model with outsourced payroll.
How do you measure the HR operating model?
Service levels met, cost per employee and per transaction, manager and employee satisfaction with HR, time HRBPs spend on strategic work, and business outcomes such as critical-role fill and regretted attrition.
Discover more from Kurums | Business Intelligence
Subscribe to get the latest posts sent to your email.