Lessons from Financial Services and Healthcare
Every CIO has sat through the demo. The pilot works, the steering committee is impressed, and someone proposes rolling it out enterprise-wide within the quarter. Then, more often than not, momentum evaporates. The research on why this happens is now extensive and consistent enough that it should reshape how transformation programs are designed from day one, not treated as a post-mortem exercise.
The Scale of the Problem
McKinsey’s global survey of 1,793 respondents found that more than eight in ten organizations had undertaken digital transformation efforts in the previous five years, yet fewer than one-third of transformations succeeded in improving performance and sustaining those gains over the long term (McKinsey). Only 16 percent of respondents reported transformations that both improved performance and equipped the organization to sustain the change, with another 7 percent seeing performance improve without the gains holding (McKinsey). Even in digitally native sectors such as high tech, media, and telecommunications, success rates did not exceed 26 percent, and in more traditional industries including oil and gas, automotive, and pharmaceuticals, success rates ranged from just 4 to 11 percent (McKinsey).
Boston Consulting Group’s research, drawing on its own experience with 70 leading companies and a survey of 825 senior executives, arrives at a similar figure: roughly 70 percent of digital transformations fall short of their objectives, with only 30 percent succeeding (BCG). BCG’s more granular breakdown is instructive for program design: 30 percent of transformations land in the “win zone,” meeting or exceeding target value with sustainable change, while 44 percent fall into a “worry zone,” creating some value but missing their targets (BCG). Notably, BCG found that getting a defined set of success factors right flips the odds of success from 30 percent to 80 percent, which suggests the failure mode is structural and addressable rather than random (BCG).
One pattern from the McKinsey data deserves particular attention from large enterprises: respondents at organizations with fewer than 100 employees were 2.7 times more likely to report a successful digital transformation than respondents at organizations with more than 50,000 employees (McKinsey). Scale itself is a headwind. This has direct implications for how large Singapore banks, hospitals clusters, and statutory boards should sequence transformation, favoring smaller, bounded rollouts with real ownership over sweeping enterprise-wide mandates announced from the top.
Financial Services: The Legacy Core Banking Trap
Banking is a useful lens because the pilot-to-scale gap is not primarily a technology problem there, it is a legacy dependency problem. IDC research from May 2024 found that nearly three-quarters of banks globally continue to run on legacy core banking systems, even as 98 percent of banks plan to upgrade those systems within the next three years to accelerate digital transformation (Fintech Singapore). The same research found that 52 percent of banks reported major restrictions from legacy infrastructure impacting delivery of new digital products, 65 percent identified the absence of real-time processing capabilities as a key challenge, and 49 percent cited inflexibility to configure or customize products as a major issue (Fintech Singapore).
This is precisely why a promising AI-driven credit scoring pilot or a slick new mobile onboarding flow so often stalls before enterprise rollout. The pilot is built on modern middleware or a sandboxed environment, but scaling it requires the core ledger, settlement, and customer data systems underneath to keep pace, and those systems were frequently not designed for real-time interaction. Encouragingly, 93 percent of banking executive teams surveyed expressed willingness or full commitment to core system change, and 53 percent of large banks globally aim to deploy more than 40 percent of total workloads to the cloud (Fintech Singapore). Intent is not the constraint. Sequencing is.
For financial institutions operating in Singapore, this legacy modernization work also cannot be separated from regulatory obligation. The Monetary Authority of Singapore treats cloud services operated by third-party providers as a form of outsourcing, and its 2021 advisory on public cloud adoption makes clear that financial institutions remain ultimately responsible and accountable for maintaining effective oversight and governance of their engagement with cloud service providers, regardless of how much operational control is delegated (MAS). MAS has stated it has no objection to financial institutions adopting cloud services, and cites the potential for economies of scale, enhanced operational efficiency, and cost savings, but this comes with an expectation of governance discipline set out in the Guidelines on Outsourcing and the Technology Risk Management Guidelines, developed jointly with the Association of Banks in Singapore through a Cloud Services Implementation Guide (MAS). A pilot that ignores this governance layer will not survive contact with a production rollout involving customer data.
Healthcare: Interoperability as the Scaling Constraint
Healthcare surfaces a different but related version of the same problem: transformation pilots that work brilliantly for one department or one hospital do not automatically scale across a health system, because the underlying patient record infrastructure was never built to be shared. Singapore’s approach to this problem is instructive precisely because it tackled the infrastructure question before layering point solutions on top. The Next Generation Electronic Medical Record (NGEMR) programme is one of three major national healthcare programmes designed to transform care delivery, and it delivers an advanced, centralized electronic medical record providing a single patient record across institutions under the National Healthcare Group and the National University Health System (Synapxe). As of July 2024, NGEMR had completed implementation across 37 healthcare institutions, spanning key partners of both healthcare clusters (Synapxe). That is not a pilot. It is a deliberately sequenced, multi-year national infrastructure rollout, built on a single-record architecture from the outset rather than retrofitted after individual hospitals each built their own systems.
This matters for any CIO watching a promising clinical AI or patient engagement pilot succeed in one facility and wondering why it stalls when proposed system-wide. The constraint is rarely the pilot’s core logic; it is almost always the absence of a shared, interoperable data layer beneath it. Singapore’s national HealthTech agency, Synapxe, has continued to invest specifically in interoperability standards and integrated data pipelines as a precondition for scaling clinical technology across institutions (GovInsider), which is the healthcare equivalent of the core banking modernization financial institutions are now belatedly pursuing.
What Separates Programs That Scale
Across both sectors, the pattern is consistent. Programs that scale treat the pilot as a proof of a narrow hypothesis, not a proof of enterprise readiness, and they invest in the unglamorous infrastructure work (core systems, interoperable data layers, governance frameworks) before or alongside the pilot, not after it succeeds. Programs that stall tend to treat the pilot’s success as sufficient evidence to greenlight a scale-up that the underlying architecture, regulatory posture, or organizational structure was never designed to support. The transformation leaders getting this right are not necessarily the ones with the most sophisticated pilots. They are the ones sequencing infrastructure and governance work early enough that scaling becomes an execution exercise rather than a discovery process.
