Our recent LinkedIn poll asked a simple question: What is the first step in a digital transformation journey?
The response was decisive. Seventy percent of respondents identified strategy definition as the starting point, while only 20% selected system automation and 5% chose data lake creation or tool implementation. Those results reinforce what we're seeing across the life sciences industry. Organizations that begin with technology often struggle to realize value because the strategic foundation isn't in place. We agree, and would go one step further: strategy is also the foundation for successful AI adoption.
The focus on artificial intelligence is moving quickly from experimentation to expectation. Executives are under pressure to identify AI opportunities, launch pilots, improve productivity, and demonstrate measurable business impact. Yet the evidence from our experience and recent AI implementation research is clear: major AI investment without digital transformation discipline is unlikely to deliver sustainable, transformational value. The organizations most likely to realize value from AI are not the ones moving fastest to buy tools. They are the ones first building the digital, data, and operating foundations that make AI usable, compliant, and scalable.
MIT NANDA’s State of AI in Business 2025 offers a sobering benchmark: despite $30–40 billion in enterprise GenAI investment, 95% of organizations report zero return, and only 5% of integrated AI pilots deliver meaningful value. The report also finds that while more than 80% of organizations have explored or piloted tools such as ChatGPT and Copilot, most of the benefit remains at the individual productivity level rather than producing measurable enterprise impact.
AI does not create transformation on its own. It depends on the organization’s strategy, data quality, process maturity, governance, technology landscape, and ability to adopt new ways of working. In that sense, digital transformation should not be viewed as a separate initiative that delays AI. It should be understood as the foundation for successful AI investment. MIT’s pilot-to-production data show how quickly value erodes when AI is not embedded in workflow-ready environments.
For life sciences leaders, this matters because AI is not being deployed into a neutral environment. It is being introduced into highly regulated, data-intensive, cross-functional processes spanning R&D, quality, manufacturing, clinical, and regulatory operations. If those processes are fragmented, data is inconsistent, and governance is weak, AI will amplify the problem rather than solve it. Research on AI project failure consistently points to the same root causes: poor strategy, poor data quality, weak integration into business processes, and insufficient governance.
The first reason digital transformation should come before major AI investment is the need for strategic clarity. Many AI initiatives fail because they begin with technology rather than the business problem they are trying to solve. A digital transformation strategy helps shift the question from “Where can we use AI?” to “Which business outcomes are we trying to improve?” That distinction matters. A strong strategy clarifies the organization’s current state, identifies inefficiencies and pain points, defines the desired future state, and creates an actionable roadmap.
A successful digital transformation should begin by assessing the existing technology landscape, data infrastructure, workflows, business requirements, risks, and opportunities before selecting or implementing new technologies such as AI. Clear, specific use cases should be defined as part of the transformation that link to a strategic goal, define the objective and business questions, identify required data, determine governance needs, assess technology requirements, define skills and capacity, establish risks and mitigations, and set KPIs for success. Without clearly defined use cases, AI pilots can become isolated experiments with no path to scale. With clear cases of use, leaders can evaluate whether an AI investment is feasible, valuable, measurable, and aligned with broader transformation priorities.
MIT NANDA’s research reinforces this point. The report found that successful buyers demand process-specific customization, evaluate tools based on business outcomes rather than software benchmarks, and expect systems to integrate with existing processes and improve over time.
The second reason is data readiness. AI is only as reliable as the data beneath it, assuming it can access, interpret, and learn from. A major AI investment made before addressing data readiness can produce unreliable outputs, poor adoption, and limited scalability. If data is siloed, inconsistently defined, difficult to access, poorly governed, or lacking sufficient quality, AI tools may generate answers that are technically impressive but operationally unusable; the classic “garbage in, garbage out” problem. For life sciences organizations, where data often sits across labs, quality systems, manufacturing platforms, and document-heavy workflows, digital transformation creates the conditions for trusted AI through data standards, governance, interoperability, and better access to fit-for-purpose data.
The third reason is dependency on process modernization. AI can automate work, but it should not be used to automate broken processes. If current workflows are fragmented, manual, inconsistent, or poorly understood, AI may simply accelerate inefficiency. In life sciences, critical work happens across tightly controlled processes that require traceability, repeatability, and accountability. AI should not be layered onto broken or overly manual workflows; those workflows should first be simplified, integrated, and redesigned so that AI can be applied where it can improve cycle time, quality, decision support, or exception handling.
The fourth reason is the need to plan for adoption and governance. In regulated industries, AI cannot be treated as an isolated innovation experiment. Governance, validation, explainability, security, and change management must be built in from the start. This is particularly important as organizations move from simple productivity tools to AI embedded in workflows, decision-making, quality processes, or regulated operations. AI governance should not be added after deployment. It should be designed into an operating model from the beginning. Digital transformation provides the structure to manage new risks before AI is scaled. That includes data and AI governance programs, compliance support, risk evaluation, performance monitoring, validation planning, user requirements, traceability matrices, operational risk assessments, and controlled release processes. Organizations can avoid being stalled by a reluctance to adopt due to risk-aversion by establishing and clearly communicating the new business-compliant business processes while promoting adoption and the expected benefits. There is a lot of adoption in the shadows at most companies due to the uncertainty around policies and tools. This means an organization is missing a huge opportunity to learn from one another and to increase the likelihood that individual experimental use is producing significant gains.
The executive takeaway is straightforward: AI should be treated as an accelerator of digital transformation, not a substitute for it. In life sciences, the path to AI value runs through strategy, process redesign, trusted data, governance, and adoption. Without that foundation, AI remains a promising pilot. With it, AI becomes a scalable business capability.