Theory of Change: how funders map the path from funding to impact

What is a theory of change?

A theory of change is a funder's articulation of how the grants it makes will produce the outcomes it cares about. It maps the causal path from inputs, money, time and expertise, through activities and outputs, to outcomes and ultimately the longer-term impact the funder is trying to achieve.

At its core it is a causal claim: a hypothesis about cause and effect. It answers the question "why do we believe that funding this kind of work, for this kind of organisation, in this context, will produce the change we want?"

‍

What does a good theory of change include?

A strong theory of change does three things. It identifies the change the funder is trying to achieve, names the intermediate changes required along the path to it, and makes explicit the assumptions that must hold for the pathway to work.

That last element is what separates a theory of change from wishful thinking. By stating the assumptions openly, inputs, animating conditions, external factors, a funder can test whether the logic actually holds as evidence comes in, rather than discovering years later that a hidden assumption never did.

‍

How do funders use a theory of change?

For funders, a theory of change is a design and evaluation tool, not a paperwork exercise. It shapes program design by clarifying what level of intervention is needed, and it shapes evaluation by defining which outcomes are worth measuring in the first place.

Many funders also expect grantees to hold their own theory of change, or to produce one on request, and use it to design reporting that collects evidence on the model, rather than asking for activity metrics that never connect to the outcomes that matter.

‍

Applying a theory of change in a grant platform

A grant management software platform turns a theory of change from a static document into the backbone of measurement. Each intended outcome becomes a KPI that grantees report against, so grantee reporting collects evidence directly on the model rather than on disconnected outputs.