Evidence-Based Management: Metrics Over Gut Feel

WHY IT EXISTS Evidence-Based Management (EBM) was created to move organizations beyond decisions based on opinion, hierarchy, or 'the way we've always done it.' It provides a systematic way to reduce risk and increase the likelihood of achieving desired outcomes by grounding choices in observable reality, not just assertions.

THE MENTAL MODEL Think of EBM like a doctor diagnosing a patient. A good doctor doesn't just guess; they gather evidence like symptoms, lab results, and medical history. They use this evidence to form a hypothesis, prescribe a treatment, and then monitor the results to see if it worked. EBM applies this same scientific rigor to business and software development decisions.

HOW IT WORKS EBM is structured around four Key Value Areas (KVAs) to provide a holistic view of performance. First, Current Value (CV) measures the value your product delivers to users right now. Second, Unrealized Value (UV) considers the potential future value if you could perfectly meet all customer needs. Third, Time to Market (T2M) measures how long it takes to deliver a new idea to customers. Fourth, Ability to Innovate (A2I) assesses how effectively your organization can adapt and deliver new capabilities. Teams set goals in these areas, measure their current state, run experiments, and measure again to see if the experiment achieved the desired outcome.

WHEN TO USE IT Use EBM when you need to make high-stakes decisions, justify investments, or break out of a 'feature factory' mindset where activity is mistaken for progress. It is ideal for product management, process improvement, and any situation where you must connect team activities directly to business outcomes and customer value.

WHEN NOT TO USE IT EBM can be overkill for low-risk, easily reversible decisions. In a true crisis where immediate action is required and no data is available, you may have to rely on expert intuition. It is also ineffective in a culture that punishes failed experiments or demands certainty before any action is taken.

ONE CANONICAL EXAMPLE A team wants to improve user retention. Instead of guessing, they use EBM. Their Current Value measurement shows 30-day retention is 20%. They identify Unrealized Value by hypothesizing that a better onboarding flow could improve this. They run an experiment by shipping a new, simplified onboarding flow to 10% of new users. They then measure the retention of the test group against the control group. If the new flow shows a statistically significant lift, they have evidence to roll it out to everyone. If not, they've learned something valuable with minimal investment and can try a new hypothesis.