Fish Don’t Talk About Water: Why Good Programmes Fail, and We Never Know Why

When a carefully designed programme falters, our first instinct is always the same: we look for a flaw in implementation. The training was weak. The team was stretched. The funding arrived late. We rarely pause to ask the harder question — what if the problem was not how we delivered the programme, but the idea it was built on in the first place?
There is a fundamental difference between two kinds of failure, and we routinely confuse them. The first is implementation failure: the idea was sound, but we did not execute it well. The second is theory failure: we did everything exactly as planned, and still the change we expected never came — because our understanding of how change happens was wrong. When we mistake one for the other, we fix what isn’t broken, and repeat the same error in the next programme.
Every programme we design is loaded with silent assumptions about how change occurs — assumptions we cannot see precisely because they have become part of how we think. There is a phrase that captures this well: “fish don’t talk about water.” The water surrounds them entirely, present in every moment, and for that very reason they never notice it. So it is with our assumptions: we build every decision on them, yet we overlook them because they seem self-evident. As the Hivos guidance on theory of change puts it plainly — we are all biased. An assumption, at its core, is a belief we take to be true without ever checking it.
Let me make this concrete. For years, many disability programmes carried a simple unspoken assumption: if we provide a child with a wheelchair, the child will go to school. The wheelchair is an assistive device, and assistive devices sit squarely within the WHO CBR Matrix — so the logic seemed sound. But I have watched, up close, the chair sit unused in a corner while the child stayed exactly where they were. The barrier was never the absence of a chair. It was a school with no ramp, a teacher who did not know how to respond, a family afraid of the neighbours’ eyes. The assumption we built the programme on — “the device is enough” — was itself the point of failure.
This is where the real value of a theory of change lies. It is not a tool for drawing a tidy diagram at the design stage. It is a way of dragging these silent assumptions into the light before they cost us years of work. When we force ourselves to write down what we assume — “we assume that providing the chair will lead to school enrolment” — the assumption becomes visible. And once it is visible, it becomes testable. Now we can ask: do we have evidence for this? Under what conditions does it hold?
Not all assumptions carry equal weight. Some are marginal; others are critical — if they fail, the whole programme fails. The critical assumption is the one that deserves our attention: the one that combines a high likelihood of being wrong with serious consequences if it is. Our task is not to verify every assumption, but to identify the few on which success depends, and to monitor them closely throughout implementation — not just at design.
This brings us to what we measure. When we track outputs alone — chairs distributed, sessions delivered — we are confirming that we followed the plan, not that change occurred. When we track our assumptions alongside our outputs, we learn: was our theory of change correct, and what did we miss if it wasn’t? The difference between counting and measuring is the difference between proving we worked and understanding whether we made a difference.
In the end, good programmes rarely fail because we delivered them carelessly. They fail because we never stopped to ask what we were assuming. The assumption we cannot see is the one that defeats us. So before you launch your next programme, ask the question we tend to avoid: what are we assuming is true here — and have we actually checked?