Transfer pricing for digital services means setting a price for intercompany transactions, such as software licenses, cloud infrastructure sharing, platform access, or marketing support between related companies, that matches what unrelated parties would charge. Tax authorities test this using the arm’s-length principle, a handful of OECD-recognized methods, and documentation you’re expected to have ready before they ask.
Say your company just signed its first intercompany SaaS licensing agreement. A US parent owns the platform. A European subsidiary sells access to local customers. Someone in accounting has to decide what the subsidiary pays the parent for the right to use the software.
That decision is transfer pricing, whether anyone in the room calls it that or not. Get the number wrong and you’re not just risking an audit adjustment. You’re risking double taxation, penalties, and a much longer conversation with two tax authorities instead of one.
Who This Actually Applies To
This guide is written for owners and finance leads at small and mid-size businesses that sell digital services across borders, not for tax directors at companies with in-house transfer pricing teams. If your company has one entity, one country, and no related-party transactions, none of this applies to you yet. Not yet, anyway.
It starts to matter the moment you have two or more related legal entities that transact with each other. That might be a parent and a subsidiary, sister companies under common ownership, or a holding structure set up for a foreign expansion. Common triggers for digital-services companies include:
- A US or EU parent licenses its software, algorithms, or brand to a foreign subsidiary that resells access locally.
- A shared services entity in one country provides cloud hosting, customer support, or platform maintenance to sister companies elsewhere.
- A regional sales entity earns a commission or margin for marketing and closing deals on behalf of the group’s core product company.
- Engineering or product-development costs are shared or cross-charged between entities that jointly build a platform.
None of these arrangements are improper. Multinational groups are allowed to structure operations this way, and tax authorities don’t object to the structure itself. What they check is the price.
Say a subsidiary keeps more profit than an unrelated distributor would earn doing the same job. Or a parent charges a royalty far outside what unrelated licensees pay for comparable technology. Either mismatch gets flagged.
The Arm’s-Length Principle, Applied to Digital Services
Every OECD member country and most non-member countries build their transfer pricing rules around one idea: related companies should price their transactions the way unrelated companies would. This is the arm’s-length principle, and it sits at the center of the OECD Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations, last fully updated in January 2022. In the United States, the same idea is codified in Internal Revenue Code Section 482 and the Treasury regulations under it. That’s the baseline everywhere.
Applying that principle to digital services is harder than applying it to, say, a shipment of auto parts. A physical product has a market price you can often observe directly. A cloud-hosting arrangement, a data-sharing agreement, or a bundle of platform IP rarely does. Comparable transactions between unrelated companies are scarce, and the value often sits in something intangible: code, algorithms, user data, or brand, rather than in a good you can point to.
That gap is exactly why the OECD spent BEPS Actions 8 through 10 rewriting the guidance on intangibles. The goal was to align where profit gets taxed with where the real economic activity happens. That means the people making decisions, taking on risk, and doing the development work. Not just the entity that happens to hold the IP on paper.
For a company selling digital services internationally, this matters directly. The country where your engineers build and improve the product has a legitimate claim to a share of the profit, even if a separate holding entity legally owns the intangible.
The Methods You’ll Actually Use
The OECD recognizes five traditional and transactional methods. In practice, most digital-services companies rely on two or three of them. Here’s what each one does and where it fits.
Comparable Uncontrolled Price (CUP)
You compare the price charged in the intercompany transaction to the price charged in a comparable transaction between unrelated parties. This is the gold standard when it works, but genuinely comparable digital-service transactions are rare. A company licensing niche analytics software to its own subsidiary usually can’t find a public license agreement for near-identical technology.
Comparable Uncontrolled Transaction (CUT)
A close cousin of CUP, used specifically for intangibles. You look for license agreements between unrelated parties involving comparable technology, comparable exclusivity, and a comparable market, then benchmark your royalty rate against them. Commercial license databases exist for this purpose, though usable comparables for newly released, highly specialized software are thin.
Resale Price Method
Useful when a subsidiary buys a finished digital product or service from its parent and resells it without material modification. You start from the resale price to the end customer, subtract a market-consistent gross margin, and that’s your arm’s-length transfer price.
Cost Plus Method
Common for centralized service functions: a shared IT or hosting center that supports the whole group adds a market-consistent markup to its costs and charges that amount to the entities it serves. Simple to apply, but only appropriate for routine, low-risk services, not for anything involving unique technology or real decision-making authority.
Transactional Net Margin Method (TNMM), called the Comparable Profits Method (CPM) in the US
By far the most-used method for digital-services companies, because it doesn’t require a matching transaction, only a matching level of profitability. You benchmark the tested party’s operating margin against a set of independent companies performing similar functions, then confirm the intercompany price lands the tested party’s margin inside that range.
Profit Split Method
Reserved for situations where both related parties make unique, valuable contributions, for example when a foreign subsidiary co-develops core platform features rather than just distributing a finished product. You split the combined profit based on each party’s relative contribution to value creation.
It’s data-hungry and harder to defend. But it’s also the method the OECD increasingly expects when two entities both hold genuine intangible value. That pattern shows up constantly in modern SaaS groups with distributed engineering teams.
Our honest read: profit split is the method most under-used exactly where it fits best. Companies default to TNMM because it’s easier to document, even when a distributed engineering team on the other side of the ocean is doing real, value-creating work that a simple margin benchmark doesn’t capture.
What OECD Guidance Says About Digital Intangibles Specifically
Hard-to-Value Intangibles
Two threads of OECD work matter most here. The first is the hard-to-value intangibles guidance issued under BEPS Action 8. It addresses exactly the problem digital-services companies run into: how do you price a license for an intangible, such as new software, a data asset, or an algorithm, when there’s no market history?
Its ultimate value depends on how well it performs after the deal is signed. The guidance lets tax authorities look at actual post-transaction results as evidence of what the parties should have priced at the outset, within limits, when the gap between projected and actual outcomes is large enough.
Amount B and the Push Toward Standardized Margins
The second thread is the OECD’s broader digital economy work, which produced the two-pillar reform package. Pillar One’s Amount B was finalized in February 2024 and applies for fiscal years starting on or after January 1, 2025. It standardizes returns for baseline marketing and distribution activities using fixed operating-margin bands, roughly 1.5% to 5.5% of sales, depending on industry group and asset intensity.
Worth flagging honestly: Amount B explicitly excludes services and digital transactions from its scope, so it won’t directly price your SaaS reseller’s margin. What it signals matters anyway. Regulators are moving toward standardized, harder-to-dispute benchmarks, and a similar approach for digital services isn’t out of the question in a future guidance cycle.
The IRS’s Parallel Move
On the US side, the IRS released Notice 2025-04 in December 2024. It introduces a Simplified and Streamlined Approach that mirrors Amount B and lets qualifying taxpayers elect a safe-harbor margin for in-scope distribution transactions, starting with tax years beginning on or after January 1, 2025. The notice was open for public comment through early March 2025, and its scope, like Amount B’s, centers on tangible-goods distribution rather than digital services.
Practitioners at firms including BDO have flagged this as a preview of where simplified, formula-based approaches for other transaction types could head next. That’s an interpretation from practitioners, not settled law.
Documentation: What Tax Authorities Expect to See
Good pricing without documentation is close to worthless in an audit. Paper trail wins audits, not intuition. Three layers of documentation come out of BEPS Action 13, and most countries with meaningful multinational activity have adopted some version of all three.
What the Three Files Cover
A master file describes the group’s global business, its overall transfer pricing policies, and where its intangibles live. A local file documents the specific intercompany transactions of the local entity, including the method selected and the benchmarking analysis behind it.
A country-by-country report is required for groups with consolidated revenue above roughly €750 million. It breaks down revenue, profit, tax paid, and employee headcount by country. That gives tax authorities a wide view of where profit sits relative to where people and assets actually are.
Most small and mid-size digital-services companies fall under the €750 million country-by-country threshold. That filing usually isn’t your problem. The master file and local file requirements often apply well below that line, though, and the exact trigger varies by country. Some require documentation for any related-party transaction over a modest dollar threshold; others only above a higher revenue floor.
Where the US Penalty Risk Bites
In the US, the stakes for weak documentation are concrete. Under IRC Section 6662(e), a substantial valuation misstatement penalty of 20% applies when a net Section 482 adjustment exceeds the lesser of $5 million or 10% of gross receipts. That climbs to a 40% gross valuation misstatement penalty once the adjustment exceeds the lesser of $20 million or 20% of gross receipts. Contemporaneous documentation prepared before the return is filed is generally your best defense against both.
A Worked Example: Pricing an Intercompany SaaS License
Numbers make this concrete faster than definitions do. This example is illustrative and simplified for teaching purposes. It is not an actual benchmarking study, and a real analysis would require comparable-company data pulled from a licensed database.
TechCo US owns the core platform IP for a project-management SaaS product. TechCo Distribution GmbH, its German subsidiary, markets and sells subscriptions to customers across the EU. The subsidiary handles local sales, marketing, and first-line customer support, but the underlying software, the servers it runs on, and the product roadmap all belong to the US parent.
| Item | Amount |
| Annual subscription revenue (German sub) | $18,000,000 |
| Local operating expenses (sales, marketing, support) | $9,000,000 |
| Platform license fee paid to US parent (proposed, 20% of revenue) | $3,600,000 |
| Resulting operating profit at German sub | $5,400,000 (30% margin) |
Run that 20% royalty rate through the numbers and the German sub keeps a 30% operating margin. That’s the problem.
Say an independent SaaS distributor performing comparable sales and support functions, no platform ownership, no product decisions, typically earns an operating margin in the 6% to 9% range under a TNMM benchmarking study. A 30% margin at the sub is a loud signal that the royalty is priced too low. Too much profit is staying in Germany relative to the limited function and risk the German entity actually carries.
What a Corrected Royalty Looks Like
Reworking the deal to target an 8% benchmark margin for the sub changes the math. An 8% margin on $18,000,000 of revenue is $1,440,000. Working backward, revenue minus opex minus target profit leaves $7,560,000 available to flow to the US parent as a license fee: a royalty rate of roughly 42% of net revenue rather than 20%.
That’s a large jump. It shows why picking the right benchmark range matters more than picking a method. Get the comparable set wrong and the “right” royalty rate can swing by double digits.
Two things stand out here. First, the US parent’s tax position improves under the corrected structure, which is the point: transfer pricing isn’t a tool for shifting profit to wherever the tax rate is lowest. It’s a mechanism for making sure profit lands where the value was actually created. Second, a 42% royalty on gross subscription revenue would itself need testing against comparable software license agreements using CUT, since an unusually high royalty rate can attract its own scrutiny.
Comparing the Methods at a Glance
Here’s how the six methods stack up for a typical digital-services group, ranked by how demanding the supporting documentation tends to be.
| Method | Best fit for | Documentation burden |
| Comparable Uncontrolled Price (CUP) | Directly comparable market transactions (rare for digital services) | Low, if comparables exist |
| Resale Price Method | Subsidiary resells a finished digital product unmodified | Medium |
| Cost Plus Method | Routine shared services: hosting, IT support, back-office | Low to medium |
| TNMM / Comparable Profits Method | Routine distribution, marketing, or support entities; the default for most SaaS groups | Medium |
| Comparable Uncontrolled Transaction (CUT) | IP licensing where similar third-party license deals can be found | Medium to high |
| Profit Split Method | Co-developed platforms where both entities hold unique, valuable intangibles | High |
Editorial rating: relative documentation and analytical burden by method (1-10 scale, Finance Fundamentals assessment)
CUP
3
Resale
5
Cost Plus
4
TNMM
6
CUT
7
Profit Split
9
Dashed line marks the documentation floor most tax authorities expect regardless of method chosen. This is an editorial estimate for planning purposes, not an official OECD or IRS scale.
Common Mistakes This Audience Makes
Small and mid-size digital-services companies tend to trip on the same handful of issues. None of this is exotic. Fix the basics first. None require a large tax department to avoid.
- Treating transfer pricing as a year-end afterthought. Pricing set after the fiscal year closes, purely to hit a target tax outcome, is exactly what invites scrutiny. Set it prospectively, before the transactions happen. Do it early.
- Using one royalty rate for every jurisdiction. A flat 15% license fee might be defensible in one market and clearly out of range in another, because comparable company data differs by country and industry.
- Skipping documentation because the company feels too small. Many countries set local file thresholds well below what founders assume, and the penalty exposure under rules like IRC Section 6662(e) doesn’t require a large balance sheet to apply.
- Ignoring the people, not just the paperwork. When engineers in one country build the product roadmap while a separate holding entity owns the IP on paper, that mismatch between substance and form is precisely what BEPS Actions 8 through 10 target.
There’s a fifth, quieter mistake: forgetting that remote and distributed teams create exposure of their own, separate from pricing. An engineer working from a country where the company has no registered entity can create a taxable presence issue that’s related to, but distinct from, transfer pricing. It’s worth reading alongside our guide to permanent establishment for remote workers if any part of your team works across borders.
Our honest take: most companies in this position don’t need a large advisory engagement to get transfer pricing right. They need a documented policy, a benchmarking study refreshed every two to three years, and the discipline to actually follow the policy they wrote down.
A Practical Checklist Before You Set or Reset Intercompany Pricing
Map every related-party transaction involving digital services, IP, or shared infrastructure.
Identify which entity performs which functions, holds which assets, and bears which risks.
Select a transfer pricing method that matches the actual economic substance, not the method that produces the lowest tax bill.
Commission or update a benchmarking study using independent comparables, refreshed on a regular cycle.
Draft or update intercompany agreements that describe the pricing and match how the business actually operates.
Prepare local file and master file documentation for every jurisdiction that requires it, before the return is filed.
Revisit pricing whenever the business changes meaningfully: new markets, new functions moved between entities, or a materially different revenue mix.
Key Takeaways
- The arm’s-length principle is the standard everywhere, but applying it to intangibles and digital services is harder than applying it to physical goods, because comparable transactions are scarce.
- TNMM, called CPM in the US, is the workhorse method for most SaaS and digital-services companies; profit split is reserved for genuinely shared value creation.
- OECD guidance under BEPS Actions 8 through 10 and the hard-to-value intangibles framework lets authorities look at actual post-deal results as evidence your original pricing was off.
- Documentation is not optional insurance. Under IRC Section 6662(e), inadequate US documentation can turn an adjustment into a 20% or 40% penalty on top of the tax itself.
That intercompany SaaS license from the opening scenario is a normal transaction, not a red flag by itself. What separates a defensible price from an audit target is a benchmarking study and a paper trail, not a lower number.
Frequently Asked Questions
Does transfer pricing apply if my company only has two small entities in different countries?
Yes, if those entities transact with each other. Size affects which documentation thresholds apply and how much scrutiny you’re likely to attract, not whether the arm’s-length principle applies in the first place.
What’s the difference between transfer pricing and permanent establishment risk?
Transfer pricing governs how much related entities charge each other for a transaction that’s already recognized as happening between two separate taxable entities. Permanent establishment asks a different question: whether activity in a country creates a taxable presence there at all, even without a registered local entity. Distributed teams and remote workers can trigger the second issue independently of the first.
Which transfer pricing method should a small SaaS company start with?
TNMM, or CPM under US rules, is usually the practical starting point for distribution, sales, and support functions, because it only requires benchmarking profit margins rather than finding a directly comparable transaction. Reserve profit split for cases where a subsidiary genuinely co-creates unique value, such as joint platform development.
How often should a transfer pricing benchmarking study be updated?
Most practitioners recommend refreshing the underlying comparable-company search every three years, with an annual check that the company’s own financial results still fall inside the benchmarked range. Update sooner if the business model, market, or function set changes materially.
Does the OECD’s Amount B guidance apply to digital services companies?
Not directly. Amount B, finalized in February 2024, covers baseline marketing and distribution of tangible goods and explicitly excludes services and digital transactions from its simplified margin bands. It signals a broader move toward standardized benchmarks, but SaaS and digital-services companies still need a traditional benchmarking analysis.
What happens if two countries disagree about the correct transfer price?
That’s a double-taxation risk, since both countries could tax the same profit. Tax treaties generally provide a mutual agreement procedure that lets the two tax authorities negotiate a resolution. Some countries also offer advance pricing agreements that lock in an approved method before the fact, reducing this risk considerably.
References
- OECD, “Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations 2022,” January 2022.
- OECD, “Guidance for Tax Administrations on the Application of the Approach to Hard-to-Value Intangibles,” BEPS Action 8.
- OECD, “Pillar One – Amount B” final guidance, published February 19, 2024, effective for fiscal years beginning on or after January 1, 2025.
- Internal Revenue Service, Notice 2025-04, December 18, 2024, introducing the Simplified and Streamlined Approach safe harbor.
- Internal Revenue Code Section 482 and Treasury Regulation Section 1.6662-6, substantial and gross valuation misstatement penalties.
- OECD, BEPS Action 13, Country-by-Country Reporting and transfer pricing documentation standards (master file, local file, CbC report; roughly €750 million consolidated revenue threshold).
- BDO, “IRS Releases Guidance on New Transfer Pricing Methodology,” analysis of Notice 2025-04.
- EY Global Tax Alert, “OECD releases final guidance on Pillar One Amount B on baseline distribution,” February 2024.
