The number that has not moved
The figure is well known in the industry: 70 to 90 per cent of large capital projects fail to deliver on time, on budget, or to full functional specification. It is cited in boardrooms, procurement committees, and project management frameworks across the sector. The most rigorous evidence sits at the upper end of that range: IPA's benchmarking of industrial megaprojects (≥ US$1 billion) finds that roughly one in five succeeds against its sanctioned business case — and that most failures surface not as a cost blow-out at the investment gate, but as an asset that never reaches its design production in the first two years after start-up.¹
What is less well understood is why this number has not improved materially over three decades of process improvement, digital project controls, and contractor market consolidation.
The answer is not execution alone. It is contract structure.
Most project failures in EPC and EPCM engagements are not pure execution failures that better project management would have prevented. They are V&V failures — the owner had no viable mechanism to verify whether a contracted obligation was actually delivered, to detect a shortfall while it could still be corrected, or to recover value once it could not.
It is worth being precise about the causal claim here, because the headline figure invites a looser one. The 70-to-90-per-cent number measures failure to deliver. What contract structure controls is adjacent and decisive: whether that failure becomes visible in time to act on, and whether the owner holds a defensible position when it does not. The two are linked — an engagement whose obligations cannot be verified will keep failing, undetected, however well it is run. The typology of the contract, not the quality of its execution, determines how defensible the owner's position is when something goes wrong.
EPC vs EPCM: the risk decision most owners take too late
The distinction between EPC and EPCM is frequently reduced to a procurement question: who manages the contractors?
In EPC (Engineering, Procurement, and Construction), the contractor bears full responsibility for outcomes against a defined specification. The owner accepts a price; the contractor accepts the risk of delivery. In EPCM (Engineering, Procurement, and Construction Management), the owner retains risk and engages the EPCM contractor as a management layer — co-ordinating specialist contractors who contract directly with the owner.
This distinction matters for cost and schedule accountability. But the more consequential distinction is epistemic: in EPC, the owner can, in principle, point to contractual specifications and measure deviation. In EPCM, accountability is distributed across a management layer, and the standards for verifiable delivery are more difficult to define and enforce.
Neither model is inherently superior. Both require the owner to answer a prior question at contract award: Under what circumstances will we be able to recover value if this engagement fails to deliver?
Most owners answer this question after the dispute begins, not before.
The three typologies: V&V exposure by service model
Around the prime EPC or EPCM contract, the owner assembles a portfolio of support engagements — the ancillary contracts through which specialist work, operational functions, and people actually enter the project. Three typologies recur across the industry, and they carry fundamentally different risk profiles when assessed against V&V exposure. These are also the contracts the owner controls most directly — which is precisely why their exposure is the owner's to manage, and the owner's to answer for when an engagement fails to deliver.
Specialist consultancy delivers a defined output: a study, specification, risk assessment, methodology, or expert report produced by a named individual or team against terms of reference. V&V exposure is low. The deliverable exists. It can be reviewed, audited, and challenged. Its conclusions are attributable to specific inputs and a declared methodology. In arbitration or mediation, a specialist consultancy deliverable generates the highest evidentiary value of the three typologies: the output is traceable, the responsibility is clear, and the standard for verification was established before delivery began.
Managed services delivers an ongoing operational or technical function under a framework agreement — typically measured by activity metrics: hours worked, incidents responded, system uptime, processing volumes. V&V exposure is medium. Results exist, but accountability is diffuse. When something goes wrong in a managed services engagement, the owner faces a structural problem: it is rarely clear which decision, by whom, under which instruction, with what information, produced the adverse outcome. Recovery in a dispute requires reconstructing an accountability chain that was never designed to be transparent. Managed services contracts frequently contain language that accurately describes the scope but provides no mechanism for the owner to demonstrate what was and was not delivered.
Staff augmentation embeds individual resources in the owner's team, billable by time, with no defined deliverables and no contractual results obligation. V&V exposure is maximum. There is nothing to verify — no result to examine, no obligation to challenge, no evidentiary trail to present. This typology maximises latent exposure precisely because it presents as the lowest-risk option at the time of contracting: flexible, scalable, easy to terminate. What it does not offer is any mechanism for the owner to recover value when performance falls below expectation, because no standard of performance was ever contractually defined.
The structural failure in EPC and EPCM project portfolios is the absence of V&V exposure analysis across these three typologies. Owners choose typologies on cost and flexibility grounds. They do not, as a matter of routine, map each component of an engagement against its V&V exposure profile before the contract is signed.
The question the owner never asks
The standard question at contract award is: What will this cost?
The question that changes the defensibility of the owner's position — and by extension, the risk profile of their capital portfolio — is: Under each service typology in this engagement, what is our ability to verify delivery and recover value if the engagement fails to deliver?
This is not a legal question, though its implications are legal. It is an engineering governance question. Answering it requires mapping each component of the engagement to its typology, assessing the V&V exposure profile by component, and making an informed risk allocation decision before the contract is signed.
The FIDIC suite provides distinct frameworks for the principal risk positions: EPC/turnkey, where the contractor carries delivery risk against a lump sum (Silver Book); plant and design-build, where the contractor designs to the owner's requirements (Yellow Book); and traditional, owner-designed construction (Red Book).² Notably, the core suite has no flagship EPCM form at all — EPCM management sits closer to a professional-services agreement than to any of these — which is itself part of the problem. The frameworks are sound. They do not, however, require the owner to conduct V&V exposure analysis as a pre-condition for contract execution. That gap is where project portfolios accumulate latent risk — risk that is invisible until a dispute reveals it.
Case: how typology mapping changed the outcome (anonymous)
In a recent upstream capital project in Latin America, the owner engaged four contractors under a hybrid EPC/EPCM structure. Two were specialist consultancies. One was a managed services provider for instrumentation and control integration. One was staff augmentation for site supervision and commissioning support.
Prior to contract execution, JR Engineering Company conducted a V&V exposure assessment across all four engagements.
The managed services contract contained a seven-figure grey area: the scope was defined entirely by activity metrics, and no verification mechanism existed for the primary integration deliverable — the demonstrable, tested function of the integrated control system. Had the engagement proceeded on those terms, the owner's ability to recover value in the event of underperformance would have been limited to termination; there was no contractual standard against which underperformance could be measured.
The staff augmentation contract had a six-figure exposure of a different character: two of the embedded resources had been making system design decisions — decisions with long-term maintenance and integrity implications — with no documentation trail, no formal authority, and no traceability back to the project specification.
Both contracts were restructured before mobilisation. The managed services scope was supplemented with a defined functional acceptance test. The staff augmentation resources were placed under a defined scope-of-work protocol with mandatory documentation requirements.
Twelve months later, a significant scope variation arose in the project. In both cases, the owner had a documented, traceable position. The dispute was resolved in mediation. No arbitration was required.
The V&V exposure assessment cost less than three per cent of the exposure it neutralised — the seven-figure and six-figure grey areas above, priced out before they could mature into a dispute.
What changes when you ask the right question
The 70 to 90 per cent figure will not improve until the industry changes the question it asks at contract award.
The failure rate is not a function of execution capability, project management maturity, or contractor market quality. It is a function of a systematic absence of V&V exposure analysis at the point where contractual typology decisions are made.
Specialist consultancy, managed services, and staff augmentation are all legitimate instruments. The question is whether the owner understands, at the moment of contracting, what each typology offers and what it withholds — specifically in terms of verifiability, accountability, and recoverability under adversarial conditions.
That is a question of engineering governance. And it has a methodology.
JR Engineering Company applies SE-grade V&V to contract structuring across EPC and EPCM engagements — verification that every contracted obligation is declared in a form that can be tested, and validation that the resulting accountability chain holds under adversarial conditions. The discipline is borrowed from systems engineering by design: here the object under test is not the finished asset but the contract structure, examined before it is signed rather than after it fails.
References
- Merrow, E. W. (2011). Industrial Megaprojects: Concepts, Strategies, and Practices for Success. John Wiley & Sons. IPA's database of industrial megaprojects (≥ US$1 billion) records a success rate of roughly one in five against the sanctioned business case; among failed projects, the majority show serious, persistent production-attainment shortfalls in the first two years after start-up.
- FIDIC. Conditions of Contract for Construction (Red Book, 2nd ed., 2017); Conditions of Contract for EPC/Turnkey Projects (Silver Book, 2nd ed., 2017). Fédération Internationale des Ingénieurs-Conseils.
- Society of Petroleum Engineers (SPE). Project-management and execution-strategy technical literature (OnePetro conference proceedings and SPE technical papers on capital-project delivery).
- Hogan Lovells. Construction & engineering disputes commentary on EPC/EPCM contract structures (e.g. EPC or EPCM contracts? and related publications, hoganlovells.com).
- Ramírez, J. (2023). 3 Common Contract Types for Professional Services in EPCM Projects: Examples and Objectives. LinkedIn Pulse, May 2023.