The evidence generation trap: 5 questions pharma leaders should ask before building again 

Executive summary

The evidence generation trap leaves pharma teams rebuilding capabilities for each new study. Real-world data (RWD) is now widely available across pharma, but access alone has not solved the evidence challenge. Too often, each real-world evidence (RWE) question triggers another round of sourcing, contracting, governance, harmonization and analysis. The immediate study may succeed while the organization loses much of what it paid to build. Treating reusable capability as a second project outcome—alongside the evidence itself—can lower the cost and effort of answering the questions that follow. 

Pharma leaders are not buying data. They are buying confidence. 

I have participated in many meetings with life sciences leaders across a variety of departments and functions over the past year, and almost nobody has said, “I want to buy data.” They ask questions that sound like data questions: How many patients? Which hospitals? Which countries? 

Underneath is something more human: they face a consequential choice and want confidence that the evidence can support it. 

Should we invest? Where should we launch? Which sites can recruit the right patients? Is the evidence strong enough to submit, or should we change course?  

When those are the real questions, volume alone is not the answer. 

If we frame the conversation as a data transaction, price and record count dominate. If we focus on the outcome, the discussion becomes more useful: clinical depth, provenance, governance, timing, and whether the evidence can withstand scrutiny. 

The conversation moves from buying data to establishing whether the evidence is fit for purpose. 

The evidence generation trap 

Evidence generation can be understood as six connected steps: define the question; select and engage the right data partners; contract and govern access; harmonize and secure the data; generate the analysis; and carry the resulting capability forward. 

Steps one to five create the capability needed to answer the immediate question. The real value comes from carrying that capability into the next project. Yet many organisations still rebuild the same relationships, agreements, definitions, governance and analytical workflows every time. The study delivers an answer, but the methods, infrastructure and operational learning rarely extend beyond it. 

When a related need appears months later, teams repeat the sourcing, contracting, definitions, governance and analysis. They get another answer, but are no better equipped to produce the next one. 

This is the evidence generation trap: each study answers a question, but the organisation pays again for much of the same capability the next time around. 

Why more RWD does not always produce better RWE 

The instinctive response is often to acquire a larger dataset. But evidence generation is not primarily a volume problem — it’s a decision problem. 

We have seen projects stall because of one missing variable, not because there were too few patients. A database may contain millions of records and still lack the pathology detail, biomarker result, longitudinal outcome or treatment context needed to answer the question. It’s a little like owning an enormous phone book without the number of the person you need. 

Fit-for-purpose data changes the starting point. Instead of asking “What data can we buy?”, ask “What must the evidence enable us to do?” That clarifies the population, variables, provenance, methodology, timing and confidence required. 

Start with the decision — and build assets for reuse 

A mature evidence strategy begins with the business or scientific outcome, not the dataset. That changes the entire project design: who must be involved, what will be credible, which sources matter, what must be governed and when the answer is needed. 

It also forces an important commercial and operational question: what should this first project leave behind? 

A well-designed project should answer today’s question while making the next one faster, less expensive and less risky. That means designing reusable data assets deliberately: 

You do not need to build a huge enterprise platform before you know it works. Start with a focused pilot, but set it up so the relationships, methods and workflows can be used again if they add value.  That makes the pilot a test of the future operating model as well as the immediate evidence question. 

Proof before scale: how mature buyers reduce risk 

Pharma leaders rarely commit to enterprise-scale evidence infrastructure on the strength of a presentation. Nor should they. They buy proof before scale. 

For example, one precision-medicine team I was working with started with a fairly narrow question: how was a single biomarker being tested? As the conversation unfolded, it became clear that the challenge was much bigger. The information they needed sat across different countries, suppliers and asset teams. Every launch meant commissioning another study, agreeing another set of definitions and handing the work over again. The study answered the immediate question, but very little could be carried into the next one. 

Rather than trying to secure support for a large enterprise programme from the start, the team chose a focused pilot. It would deliver useful evidence at country and hospital level straight away, while also testing whether the same sites, governance and analytical environment could be used for the next biomarker, asset or market. 

A very different organisation, a neurotechnology company validating software on complex brain imaging, made the same buying decision. Cheaper imaging data existed, but it primarily came from healthy participants rather than the relevant patient population, and its provenance was not strong enough for validation. The company funded a small pilot first, not because the ambition was small, but because the pilot could reveal what scaling responsibly would require. 

The evidence needs were very different, but the pilots served a similar purpose. Each organisation came away knowing which relationships, governance arrangements and analytical capabilities could support the next decision — and which would need to change. 

What a reusable evidence foundation looks like 

What makes an evidence foundation reusable is what remains available after the first project: the data relationships, governance, definitions, analytical methods and working practices that teams can use to answer the next question. 

Its value compounds when teams can carry definitions, sites, contracts, secure environments and analytical logic across assets, indications and markets. The next use case moves faster because the expensive and difficult work does not disappear when the first report is delivered. 

The business impact is tangible. Teams spend less time duplicating work across studies, move through evidence cycles faster because they are not starting from scratch, and get more value from each investment across assets, markets and lifecycle questions. 

Five questions to ask before the next evidence project 

Before starting another evidence project, it’s worth pausing to ask five questions. They will help clarify what the work needs to achieve now and what should remain useful afterwards. 

AI raises the standard for evidence infrastructure 

AI increases the value of reusable evidence foundations because it can accelerate cohort design, analysis and interpretation across multiple questions. It also increases the cost of weak foundations: inconsistent definitions or uncertain provenance can be reproduced at greater speed and scale. The same infrastructure that supports reuse therefore becomes part of controlling AI-enabled evidence generation. 

AI does not remove the need for evidence discipline, but it does make weaknesses in the foundation more consequential. Fragmented, poorly harmonised or weakly governed inputs can produce outputs that are faster, but not necessarily more trustworthy. 

Responsible AI-enabled evidence generation depends on data provenance, transparent definitions, harmonisation logic, governance controls and auditability. These are not administrative ‘extras’. In regulated and high-stakes settings, they are part of the end product. 

Be ready for the next question 

Evidence teams face more complex questions, more fragmented data and less patience for slow answers. The challenge is to produce evidence that supports the decision at hand while retaining the relationships, methods and infrastructure that can support the next one. That means looking beyond the immediate study. The data, partnerships and analytical work should answer the question at hand without being discarded once the report is delivered. 

Making better use of the right data, and keeping the foundations already in place, helps teams answer the next evidence question faster and with greater confidence. 

Related reading: Explore the Evidence Insights series 

Read the feature: Narasimha Kumar explains how reusable data foundations are becoming essential to generate trusted, decision-ready evidence across the product lifecycle.

Download the white paper: In From Fragmented Data to Decision-Ready Evidence, we explore how fit-for-purpose data networks can help life sciences teams generate evidence faster, strengthen confidence in their analyses, and make faster, better-informed decisions across the product lifecycle.

Watch the expert interview: Tobi Oremulé and Narasimha Kumar discuss how fit-for-purpose data can help transform evidence complexity into decision-ready insights, and how organisations can build reusable evidence foundations that create value across the drug lifecycle.