Software and SaaS Transfer Pricing: From Licensing to Profit Splits
The transfer pricing of software and SaaS: the different economics of licensing versus subscriptions, the royalty benchmark problem, DEMPE, and when the profit split is the right answer.
Software and SaaS transactions price three different economics under one industry label: development services (building the software — a service return), licensing (the right to use the software — a royalty), and subscription/SaaS (the ongoing delivery of the software as a service — a service return with a technology-content question). Each has its own method, its own tested party, and its own examination pattern. The file that treats them as one transaction prices all three wrong.
The three fee streams, separated
| Fee stream | What is transferred | The TP character | The method |
|---|---|---|---|
| Development fee | The effort of building (the engineers’ work) | A service — the developer’s function | Cost plus / TNMM on OP/OC on the development service; the contract R&D characterization decides routine vs entrepreneurial |
| License fee (perpetual or term) | The right to use the technology | A royalty for the intangible | The intangibles machinery: CUP attempt, TNMM on the licensee, residual split for valuable IP |
| Subscription / SaaS fee | The ongoing service (hosting, access, maintenance, updates) | A service with technology content — the customer uses, it does not own | TNMM on the service provider’s function; the technology’s value sits in the platform owner’s return, not the subscription price |
The separation is the first documentation decision: the agreement’s fee structure (the development fee, the license fee, the subscription fee — or the blended price) must be mapped to the three economics, and a blended price is unbundled for the analysis (the documented split of the blended fee into its components, on the stated basis).
The royalty question, when it exists
Where a license fee is paid (the classic: the Indian entity licenses the platform to the group, or licenses a product to a related party), the intangibles machinery applies in full: the CUP attempt on comparable licenses (documented and usually failed on the specificity), the TNMM-on-licensee test (the Indian licensee’s residual return supporting the royalty), and the residual argument where the IP is genuinely valuable. The Indian examination specifics — the royalty character for TDS, the TPO’s testing of the licensee’s residual — are covered in the intangibles guide and in TDS on TP payments.
The SaaS-specific trap: pricing the subscription as if it were a royalty. The subscription fee is the price of the service — the hosting, the access, the maintenance, the updates — and the technology’s value is inside the provider’s return, not a separate royalty on the fee. Where the platform owner and the SaaS operator are different group entities, the operator’s return is benchmarked as a (technology-dependent) service, and the platform owner’s return is the platform’s economics — the two returns, the two benchmarks, and the allocation between them is the DEMPE question again (who built the platform, who operates the service, who owns the customer relationship).
DEMPE: where the software’s value is earned
The software fact pattern concentrates the value question in the Development/Enhancement functions:
- The captive that builds — the Indian entity that develops and maintains the platform for the overseas owner: the contract R&D characterization decides its return (the routine cost-plus, or the entrepreneurial participation). The Indian safe harbour for software development services (OP/OC ≥ 17%, 18% above the threshold) and contract R&D (software) (OP/OC ≥ 24%) are the prescribed positions where the facts fit — see the safe harbour guide.
- The entity that owns and exploits — the platform owner that commercializes: its return is the platform’s economics — the residual after the builders’ routine returns, on the DEMPE/control record.
- The SaaS operator — where a separate entity operates the subscription business (the customer relationships, the go-to-market): its return is the service return, benchmarked on the operating function — and the customer relationships it owns are an intangible of its own, which the file must acknowledge (the operator that builds the customer base is not a pure pass-through).
When the profit split is the right answer
The split is the answer where the software economics are genuinely two-sided: the Indian entity performs key development (not routine execution — the decision rights, the architectural control, the IP share), and the overseas entity commercializes. The one-sided TNMM on the Indian entity then prices the routine share and leaves the entire residual abroad — the misallocation the residual split exists to correct. See profit split vs TNMM for the decision test and the TPO’s response pattern, and the profit split guide for the mechanics.
The evidence is the same in every software split: the decision trail (who decided the architecture), the IP registry (who owns the code and the patents), the funding record (who paid when it failed), and the commercial record (who owns the customers and the revenue risk). Written contemporaneously, it is the file; written after the examination notice, it is the reconstruction.
The documentation, by fee stream
| Fee stream | The pack |
|---|---|
| Development | The development agreement (scope, decision rights, IP, milestones); the characterization memo (routine vs entrepreneurial); the cost base and the mark-up benchmark or the safe harbour position |
| License | The license agreement (scope, territory, term, the royalty base and rate); the CUP attempt and its failure; the TNMM-on-licensee test or the residual analysis; the DEMPE record |
| Subscription | The SaaS agreement (the service scope, the pricing, the customer-relationship ownership); the operator’s function analysis; the benchmark on the operating function; the platform-owner allocation where the entities differ |
The through-line for all three: the fee structure of the agreement is the economics, the economics are the method, and the method is the defence — the file that cannot map its invoices to the three fee streams cannot answer the first question the TPO asks.
See also
Run the screens as a study, not a spreadsheet
Quartyl applies the method, PLI and screening steps above as a pipeline — and keeps a documented reason for every exclusion.
Related docs
Intangibles TP: Licensing, Royalties, DEMPE and HTVI
The transfer pricing of intangibles: the royalty benchmark problem, DEMPE allocation, cost sharing, hard-to-value intangibles, and the Indian practice on royalties.
Read docContract R&D vs Entrepreneurial R&D: DEMPE and Cost-Based Returns
The line between routine contract R&D and entrepreneurial R&D: who performs the DEMPE functions, who controls the risk, who owns the IP, and what each side earns under each characterization.
Read doc