Horizon subcontracting: test the action task before you price the provider
The first classification question is not whether an invoice will be issued. It is whether the external provider will implement an action task. That answer controls the justification, budget line, selection evidence and beneficiary accountability.
A Horizon Europe budget can place an external invoice in the wrong category long before procurement begins. The common drafting mistake is to decide that an activity is a 'purchase' because a supplier will bill for it. The controlling question is different: will the provider implement an action task described in the project?
The European Commission's Annotated Grant Agreement distinguishes subcontracts from purchases on that basis. A subcontract concerns implementation of action tasks. A purchase covers other goods, works or services needed to support implementation. The same type of service can fall on either side depending on the project's actual task design.
That classification affects more than the budget heading. It determines what must be described and justified in Annex 1, how the estimated cost appears in Annex 2, what selection evidence must be retained and which beneficiary remains accountable for delivery.
The beneficiary does not transfer grant responsibility to the provider. The Annotated Grant Agreement states that beneficiaries remain responsible towards the granting authority for work performed by subcontractors and other participants. A contract can allocate commercial remedies, but it does not remove the beneficiary's obligations under the grant.
Selection also requires evidence. Subcontracting costs must be based on best value for money or, where appropriate, the lowest price and must avoid conflicts of interest. Competitive selection is described as the safest default, but the governing test is whether the beneficiary can prove compliance under its applicable rules if reviewed or audited.
The standard Grant Agreement option says subcontracting may cover only a limited part of the action. The Annotated Grant Agreement also notes that this option can be unselected for actions where subcontracting is a key or large part, such as some infrastructure or procurement-oriented actions. The selected call and grant options therefore matter.
Core coordinator tasks are a separate boundary. The Annotated Grant Agreement says the coordination tasks listed under Article 7 cannot be subcontracted or outsourced, apart from the specific delegation exceptions described there. A proposal should not build its accountability model around an external coordinator-for-hire.
PRINCEPS recommends a seven-question subcontracting decision gate before the work plan and budget are frozen:
1. Action-task test — does the provider perform work that the proposal defines as an action task, or supply a supporting good or service?
2. Necessity test — why is external implementation necessary, and why can the beneficiary or consortium not perform the task with its own resources?
3. Scope test — is the outsourced work limited, precise and consistent with the selected call and Grant Agreement options?
4. Responsibility test — which beneficiary owns the task, technical acceptance, reporting, records, corrective action and consequences of failure?
5. Selection test — what process and market evidence will demonstrate best value for money or the lowest price and absence of conflict of interest?
6. Annex and budget test — do Annex 1, the work-package narrative, Table 3.1g, Annex 2 and any lump-sum cost assumptions describe the same scope and estimate?
7. Rights and evidence test — does the contract secure access, intellectual-property, data, confidentiality, audit, record-retention and verification rights needed for the grant?
This is a PRINCEPS control gate, not an official Commission table. It is deliberately narrower than a general consortium-role or country-funding screen. Its purpose is to decide whether an external provider is implementing an action task and to make the resulting procurement and accountability trail coherent.
The action-task test should be applied to the actual Description of Action, not the provider's commercial label. A laboratory analysis, survey, software build, event service or technical study may be a supporting purchase in one project and a subcontracted action task in another.
The proposal should also avoid naming a preferred provider without a defensible basis. Where a provider is known at application stage, the narrative still needs to explain the task, estimate and selection logic. Where it is not known, the task and cost can be identified without pre-selecting the entity, subject to the applicable procedure.
For African field implementation, do not use subcontracting as a substitute for a substantive consortium role where the organisation is expected to shape methodology, govern data, make project decisions or own results. Conversely, do not place a genuine supplier in the consortium solely to avoid procurement controls. Classify the real contribution.
Before submission, reconcile the external-work list with work packages, deliverables, milestones, person-months, purchase costs, subcontracting justifications, ethics and data measures, intellectual-property provisions, risk controls and the budget. A mismatch in one section can undermine both credibility and cost eligibility.
This analysis is strategic and editorial. It is not legal, procurement, audit or cost-eligibility advice and does not guarantee acceptance of a proposed cost. The live call, selected Grant Agreement options, institutional procurement rules, national law and granting-authority guidance remain controlling.
Official sources
Verify the underlying development.
Editorial note: This is PRINCEPS analysis for general information. It does not replace the official work programme, topic conditions, submission system, grant rules or professional advice specific to an application.