What "evidence-based" actually means
The term has been diluted to the point of meaninglessness. Here's how we use it, and what it requires of you.
The phrase “evidence-based” appears on the homepage of nearly every consultancy in our category, including, until recently, ours. The phrase has been diluted to the point of meaninglessness.
When we say a decision is evidence-based, we mean four specific things.
1. The evidence is fit for the question
A descriptive study answers “what is happening?” An evaluative study answers “what worked?” A causal study answers “what would have happened without us?” These are three different questions requiring three different study designs. Evidence-based decision-making starts with matching the design to the question, not with whatever data is already on hand.
2. The evidence is recent enough to be true
A 2014 randomised controlled trial of a literacy intervention in Ontario tells us less about a 2026 literacy intervention in rural Alberta than a 2024 quasi-experimental study from a similar context. “Recent” is not a hard threshold; it is a contextual judgment about how fast the underlying conditions change.
3. The evidence is honest about uncertainty
A point estimate ($3.20 per dollar) without a sensitivity range ($2.40–$4.10) is a marketing number, not a research finding. We report ranges. We report assumptions. We report the things that would invalidate the finding if they turned out to be wrong.
4. The evidence is contestable
If a finding is true, it should survive scrutiny. We pre-register hypotheses where we can. We publish methods alongside conclusions. We invite re-analysis. “Evidence-based” without contestability is just authority-based with extra steps.
Most of what gets called “evidence-based” in our sector meets one or two of these criteria. Meeting all four is rare. It is also the actual job.
Article body — approximately 1000 words. Expand during content production phase.