Digital twins aren’t failing because the technology isn’t ready – they’re failing because organisations aren’t.

In this instalment of Capgemini’s ‘Future of Oil & Gas’ series, we explore why digital twins so often stall in Proof‑of‑Concept and what it takes to turn promising pilots into trusted, decision‑making operational capabilities.

Digital twins (virtual replicas of physical assets or processes, running off real data) have become one of the most widely piloted technologies in the oil & gas sector — yet paradoxically, one of the least successfully scaled. Across upstream, midstream, and downstream operations, pilots consistently demonstrate impressive gains in production, reliability, efficiency, and safety. But organisational reality tells a different story: fewer than a quarter of digital twins ever progress into full operational deployment, and many stall permanently in Proof‑of‑Concept (PoC) mode.

Why? It turns out the core barriers have little to do with technology readiness — and everything to do with mindset, governance, and execution.

This blog explores why digital twins get stuck, what must change, and how Capgemini helps operators turn promising prototypes into embedded operational capabilities.

Done right, digital twins are incredibly valuable

When implemented well, digital twins deliver substantial and measurable value. They provide a dynamic, data‑driven virtual representation of critical assets, enabling organisations to understand and optimise operations in ways not previously possible.

Real‑world examples from the sector prove this:

Whether improving yield, reducing outages, minimising helicopter trips, preventing failures, or accelerating workflows, digital twins repeatedly prove their capacity to unlock multi‑million‑dollar results. The value is clear. So, why aren’t these successes scaling across the enterprise?

Why digital twins get stuck in pilot mode

Drawing on Capgemini’s deep industry expertise and hands‑on experience delivering digital twins across the oil & gas value chain, we’ve identified a consistent set of reasons why so many promising initiatives struggle to progress beyond the pilot stage. While the technology itself is increasingly mature, organisational realities – spanning integration, governance, trust, and operating‑model alignment – often prevent pilots from becoming scalable, decision‑bearing operational tools. The following themes outline the most common factors that hold digital twins back from achieving their full enterprise value.

The market‑wide integration failure

No major vendor has yet delivered multiple, scaled, decision‑bearing twins across diverse assets. Each pilot tends to be bespoke – tailored to one platform, one piece of equipment, or one refinery unit. That makes it hard to replicate success across dozens of facilities.

Failure to meet all four pillars of a “true” digital twin

Most pilots succeed in demonstrating one or two pillars (often real‑time data or simulation), but few deliver all four:

  • Bidirectional, with data sent from the physical asset to the digital model and received in return
  • Continuously calibrated to reflect and stay synchronised with real-time operational states
  • Institutionally trusted by engineers and regulators as a source of truth
  • Decision‑bearing, not just providing visualisation

Without all four, twins are just compelling experiments, not operational tools.

The value perception gap

According to EY’s Future of Energy survey, only 14% of companies believe digital twins are living up to expectations, despite widespread pilots.

This gap exists because PoCs often focus on proving functionality, not delivering business outcomes. When the ROI isn’t clearly quantified, production leaders deprioritise it in favour of immediate operational pressures.

Data siloes and quality issues

Digital twins require integrating data from engineering, operations, maintenance, etc. which are often sitting in different siloes in different formats. In oil & gas particularly, data can be incomplete or unclean. And let’s face it, trust is huge. As soon as end-users see that the twin’s data is wrong, they will distrust the whole system. Which is why Shell’s team implemented the Pin my data feedback loop – to actively maintain data quality and thus user trust. The trouble is, not all companies have processes like this, and many PoC twins slowly diverge from reality as underlying data isn’t maintained, leading to abandonment.

Integration with live systems is also technically challenging – connecting a new twin platform to, say, a refinery’s DCS data can raise cybersecurity and networking concerns that slow down projects significantly. 

The pilot‑to‑production “Valley of Death”

A pilot might be done with a subset of data, maybe in a test environment, to prove a concept. Scaling it means connecting to all assets, automating data feeds, dealing with user access control for potentially hundreds of users, hardening the system for cybersecurity, etc. These tasks are not trivial and often not budgeted or planned for in the pilot.

When the pilot ends, teams can be unsure how to proceed – they proved value on one asset, but replicating to 50 assets could be non-linear in effort. Without commitment, projects stall.

Also, PoC stall happens when initial results are promising but not overwhelmingly so, leading management to adopt a “wait and see” stance rather than push forward aggressively. Perhaps a pilot saved £1M in a small area – impressive, but in a company with billions in expenses, some might question if it’s worth scaling right away or if they should try other pilots. This cautious approach can lead to “pilotitis”: always doing pilots, never fully rolling out.

What needs to change?

There are ways out of PoC purgatory but scaling digital twins requires a fundamental shift from tech‑led experimentation to decision‑centric operational transformation.

We recommend the following path forward for success:

  1. Reframe digital twins as “Decision Twins”

This aligns the technology to the decisions that matter most – e.g.,

  • “How do we reduce compressor failures by 20%?”
  • “How do we cut turnaround duration by 5%?”
  • “How do we maximise Crude Distillation Unit margin every hour?”

2. Build a scalable data and governance foundation

Successful operators treat digital twins as living assets requiring:

  • A clean, governed, continuously updated data foundation
  • Clear ownership in operations, not digital innovation teams
  • Defined model assurance and recalibration workflows
  • Integrated knowledge management

3. Start with a Minimum Viable Twin (MVT)

Over‑engineering kills adoption. A twin should begin with the smallest surface area required to deliver value – one unit, one decision, one workflow – then expand in manageable increments.

4. Embed twins into operational routines

A twin that sits on a separate portal will never scale. A twin that changes the shift handover, informs the daily production meeting, reduces offshore maintenance visits, and updates the maintenance backlog becomes indispensable.

Twins must be operational capabilities, not products.

5. Engineer a repeatable path to production

We recommend your five-step credible path to production needs:

  1. A clearly defined decision domain
  2. A validated, continuously calibrated model
  3. Hybrid automation with humans in the loop
  4. Engineering for scale – repeatable, standardised, secure
  5. Full workflow embedding and organisational adoption

This is precisely how operators like bp, Equinor, Shell, and Cosmo succeeded with their flagship projects.

Capgemini can help you break out of PoC Purgatory

We bring a uniquely integrated capability set that directly addresses the barriers identified in the industry:

  • Deep engineering and data integration expertise – We combine asset engineering, OT/IT convergence, and data platform modernisation – exactly what’s needed to build the data foundations where most twins fail.
  • Proven success across the oil & gas value chain – Our work spans predictive maintenance, lifecycle optimisation, emissions reduction, operations management, and engineering governance — all areas where digital twins unlock measurable gains.
  • Leading partner ecosystem – We work with all major digital twin platforms, including Cognite, AVEVA, Hexagon, Kongsberg, IBM, and Microsoft, ensuring vendor‑agnostic, scalable solutions.
  • Decision‑centric transformation frameworks – Capgemini’s “Decision Twin” approach ensures focus on value, not technology.
  • Change management and operating model redesign – We help organisations embed twins into real workflows, redesigning processes and governance to ensure sustained adoption.

Digital twins have already proven they can deliver millions in value – increased production, reduced downtime, enhanced safety, lower emissions, and more. But to unlock these benefits at scale, the industry must shift from experimenting with technology to transforming decision‑making.

By reframing twins around decisions, investing in robust data and governance foundations, starting small, and embedding outputs into daily operations, oil & gas companies can finally break out of PoC purgatory.

And with Capgemini as your business and technology transformation partner – combining engineering depth, digital excellence, and operational know‑how – you can transform digital twins from isolated prototypes into the powerful operational capabilities they were always meant to be.