The Global Payroll Reckoning and Intermezzo’s Fix for it

Across every conversation we have with customers, design partners, and industry peers, one theme keeps coming up: people know global payroll processes are broken. Too fragmented, too rigid, too slow to adapt.

We hear it from multinational enterprises bogged down by local providers they can't see into and can't hold accountable. We hear it from payroll teams who spend more time reconciling data across disconnected systems than they do running payroll.

Under this frustration lies the primary culprit: the in-country provider (ICP) model. It's been the industry's default for decades. It persists because building true payroll infrastructure in multiple countries is genuinely hard. But hard to build doesn’t mean impossible, and we should strive to build true global payroll infrastructure that provides an excellent customer experience that pays people correctly and on time.

The Problem Isn't Local Complexity, it’s the Architecture.

Global payroll is local by nature. Every country sets its own tax rules, statutory contribution schedules, filing deadlines, and labor law interpretations. The UK calculates payroll differently from Ireland. Germany's social insurance reporting obligations are among the most complex in the world. Australia runs a real-time reporting mandate through Single Touch Payroll. These country-specific differences won’t go away.

But the ICP model's response to that complexity — stitching together independent local providers behind a thin aggregation layer — creates a second layer of problems that has nothing to do with compliance. It has to do with architecture.

When an enterprise runs payroll across a dozen countries using an ICP-based provider, they are not operating a global payroll system. They are operating twelve country payroll systems with a shared login screen. Data moves between them by emails and spreadsheets that were never designed to interoperate seamlessly. Each in-country provider has its own data model, its own error formats, its own schedules. The aggregator sits in the middle, translating, reconciling, and hoping nothing breaks between pay cycles.

The results unfortunately speak for themselves. A 2025 Global Payroll Week survey from PayrollOrg found that 57% of respondents ranked ensuring local compliance as their single biggest challenge: not a compliance failure in any one country, but the operational burden of maintaining compliance coherently across all of them. Fragmented in-country providers were explicitly cited as a source of inconsistent processes and limited visibility for central payroll teams. And the pressure will only continue: between 2025 and 2026, dozens of countries updated wage floors, contribution rates, or reporting formats, making the maintenance overhead of a multi-vendor patchwork unsustainable.

What the ICP Model Actually Costs You

The problems with the ICP model fall into three categories: speed, accuracy, and control.

Speed. Every country in an ICP network is operated by an independent payroll provider. When a regulation changes — updated National Insurance thresholds in the UK, an updated Central Provident Fund contribution schedule in Singapore, a revised care insurance calculation in Germany — each provider updates on its own timeline. Customers wait. In the meantime, they carry compliance risk that’s hard to see and quantify.

Accuracy. The ICP model is structurally optimistic about data integrity. Employee records, compensation data, time inputs, and benefit elections all originate from spreadsheets (or an HCM or ERP system if you’re lucky) and must be translated into each provider's format. Every translation is a potential point of failure. Errors can survive undetected through multiple handoffs before surfacing in a payslip or a tax filing. The aggregate model also makes explainability nearly impossible: when an employee's deduction changes unexpectedly, tracing the cause through multiple independent systems is an exercise in frustration for payroll teams and a black box for employees.

Control. Perhaps the deepest cost of the ICP model is what it does to reporting and governance. A multinational enterprise should be able to see its global payroll position — total labor cost, tax exposure, variance by country, filing status — in real time. With fragmented providers, that visibility requires manual aggregation. Payroll data sits in silos, each owned by a different party with different refresh cycles. Cross-country governance, auditability, and strategic workforce analytics become aspirational rather than operational.

A Different Answer: What We're Building at Intermezzo

We started Intermezzo because we believe local dominance in payroll can be solved with superior technology and attention to detail. Having spent our careers at EOR and payroll companies like Velocity Global (now Pebl) and Intuit, we've seen firsthand how enterprises navigate the ICP model's limitations and what it costs them.

Our answer is to rebuild payroll infrastructure from the ground up: build an intelligent platform that owns the payroll engine in each country rather than outsourcing it to a local partner. Then expose that platform through one consistent, enterprise-grade API.

One API. Every Country.

Our API is designed so that a customer integrates once and adds countries incrementally. The resource model, URL conventions, authentication patterns, and documentation structure are consistent across every market. Adding Germany does not require rebuilding the integration that already runs for the UK.

This matters because every new country in an ICP network means a new set of quirks and edge cases that the customer's payroll team has to absorb. Our single global API absorbs those differences internally, presenting a stable surface to the customer while handling country-specific compliance logic behind it.

A Compliance Engine That Proactively Updates

The ICP model's approach to compliance is fundamentally reactive: when a regulation changes, someone eventually updates the local provider's system, and the change eventually propagates to the customer. Our approach is different.

At the core of our platform is a Global Knowledge Engine: an AI system that continuously monitors regulatory changes across markets and translates them directly into updated payroll calculations. When a tax rate changes in Germany or a statutory contribution schedule shifts in Singapore, the system picks it up, validates it, and applies it. No waiting for a local partner to update their software. No manual intervention by the HR team. The platform stays current because it's built to stay current.

The output of the Knowledge Engine is not a PDF summary for a human to review. It is a direct update to the computational graph that calculates each employee's payslip. That means our compliance posture is not a function of how quickly a local partner gets around to updating their software. It is embedded into our platform’s architecture.

Computational Accuracy and Built-In Explainability

Our Gross-to-Net Engine performs payroll calculations using deterministic computational graphs: mathematical representations of the rule chains that govern what an employee earns and what gets deducted. For every payroll run, every employee, we maintain a complete log of the computational path that produced each output.

This has two important consequences. First, the calculations are verifiable. They can be audited against the regulatory rules that generated them, and tested against official government test cases, which is exactly how our UK engine achieves HMRC PAYE recognition, and how our German engine is being built to pass ITSG certification.

Second, the calculations are explainable. Because the system logs its reasoning at the per-person, per-run level, questions like "why did my care insurance deduction increase by €20 this month?" can be answered programmatically and self-service. You no longer need to escalate to a local provider's support queue and wait for hours for a response, but rather the system immediately retraces the computational graph and surfaces the specific rule change or data input that caused the delta. 

Tax Filing Without the Waiting

The other half of the payroll cycle — statutory filings and payments — is where the ICP model's fragmentation can create real friction and stress for finance teams. Filing deadlines vary by country, by tax type, and sometimes by employer size. Under fragmented models, keeping up with all of that often means a lot of manual checking and last-minute corrections.

Our Tax Filing Engine generates and executes contextually correct filing schedules based on the same regulatory knowledge that drives payroll calculation. The system knows when filings are due, what format they require, and what validations must pass before submission. In Germany, this means direct integration with ITSG's data transmission standards — one of the most demanding regulatory interfaces in European payroll. In the UK, it means Real Time Information (RTI) submissions, HMRC notices, P11D filings, and BACS files for bank transfers, all generated from the same unified data model.

Where We Are Today and Where We're Headed

We’re ready in the United Kingdom and Germany, our two launch markets, and we chose them deliberately. Both are technically demanding payroll environments — Germany especially, given its ITSG certification requirements and detailed social insurance reporting obligations. Getting these right gave us a strong foundation to build from.

In the UK, our engine covers the full payroll lifecycle: PAYE and National Insurance calculations, Benefits in Kind, statutory payments, Auto Enrolment pension management, RTI and annual filing, and all employee documents from P45 onboarding through P60 year-end. The engine is HMRC recognised.

In Germany, we're pursuing ITSG certification, the most rigorous payroll certification in Europe, with a target of completing the system audit in 2026. Our German module covers contribution calculations, social insurance reporting (DEÜV), all electronic data transmission procedures, absence and benefit certificates, pension and wage certificates, and supplemental procedures including cross-border worker handling and short-time work benefits.

On our near-term roadmap: UAE, Ireland, and Poland. Each of these markets presents its own regulatory character — the UAE's Wage Protection System and end-of-service gratuity calculations, Ireland's PAYE Modernisation real-time reporting to Revenue, and Poland's ZUS social security contributions alongside mandatory PPK employee capital plan enrollment — and each will be built on the same platform that powers our UK and German engines. That's the whole point. What we build for one country informs how we build the next, and customers who integrate once don't need to rebuild when we add a market.

What We Believe

The ICP model's defenders argue that local expertise cannot be replaced. We agree that local expertise is essential. We disagree that it must live in a third-party provider that the customer never sees and cannot hold accountable.

Our architecture internalizes that expertise — encoding it in knowledge graphs, validating it against official regulatory sources, and surfacing it in an API that enterprise customers can build on. The goal is not to paper over local complexity. It is to make that complexity invisible to the customer: to make global payroll, finally, programmable. Enterprises are actively looking for platforms that can replace fragmented local systems with something they can integrate once and extend globally.

That's what we're building. Global payroll doesn't need to be painful. It needs to be re-engineered, and we’ve done it.

Learn more about what we’re building by booking a demo with us below or reaching out to us directly at sid@intermezzo.ai or kumar@intermezzo.ai.