Outcome Mapping: measuring the change you can actually influence

The organising concept is the boundary partner: an individual, group or organisation the programme interacts with directly and can plausibly influence, but does not control. A research programme's boundary partners might be a ministry's policy unit, three universities and a farmers' federation — not "smallholder farmers in the region", who are beyond the boundary.

Outcome mapping divides the world into three spheres. The sphere of control holds the programme's own activities and outputs. The sphere of influence holds the boundary partners, whose behaviour the programme can affect but not dictate. The sphere of interest holds the development impact everyone is ultimately after, which the programme contributes to alongside forces it does not touch.

The method then makes a deliberate choice: monitoring and evaluation concentrate on the sphere of influence. Outcomes are defined as changes in the behaviour, relationships, activities or actions of boundary partners. Not attitudes, not awareness, not satisfaction — observable changes in what people and organisations actually do.

That definition is the method's discipline and the source of most of its value. "Increased awareness of climate risk among planning officials" is not an outcome in outcome mapping. "The planning department now requires a flood-risk annex in every district submission" is.

Outcome mapping was developed at Canada's International Development Research Centre in the late 1990s by Sarah Earl, Fred Carden and Terry Smutylo, and published as Outcome Mapping: Building Learning and Reflection into Development Programs (2001). Earl and Carden set out the reasoning behind it in Learning from complexity: the International Development Research Centre's experience with Outcome Mapping.

The problem it was built for was specific to research-for-development funding: IDRC funded research whose value lay in being used by others, sometimes years later, through chains the programme neither owned nor observed. Logframes handled this badly, producing either trivial output indicators or impact claims nobody believed.

It has since been applied well outside that origin — in agricultural knowledge systems, as in Research that matters: outcome mapping for linking knowledge to poverty-reduction actions, and in health-system work, as in Outcome mapping for health system integration.

A five-year regional programme worked on climate-resilient planning. Its logframe promised "improved resilience for 400,000 people" — a claim no evaluation could have supported.

The outcome-mapping redesign named six boundary partners: two ministries, a national meteorological service, a university department, a municipal association and a network of community organisations. Each got an outcome challenge and 8–12 progress markers.

Three years in, the outcome journals showed a pattern nobody had anticipated. The ministries had moved briskly to the top of "expect to see" and stalled — attending everything, changing nothing, because the relevant decisions sat with provincial offices the programme had not identified as partners. The municipal association, given far less attention and budget, had reached two "love to see" markers, having rewritten its model planning bylaw on its own initiative.

The programme reallocated: it added provincial planning offices as boundary partners, cut ministry engagement to maintenance, and put staff time behind the municipal association. None of that would have been visible in a results framework reporting activities delivered and people reached — and the graduated markers were what made the stall legible as a stall rather than as slow progress.

• Boundary partners defined as beneficiaries. If the programme does not interact with them directly, they are in the sphere of interest, and putting them in the sphere of influence reintroduces exactly the attribution problem the method was built to sidestep.
• Progress markers that are not behaviours. "Improved understanding" cannot be observed. "Cites the risk assessment in the annual plan" can.
• Markers that are programme activities in disguise. "Attends our training" is about the programme; "applies the method without us present" is about the partner.
• A flat ladder. If every marker is at roughly the same difficulty, the graduation does no work and monitoring becomes a checklist.
• Journals as reporting. Filled in retrospectively at reporting time, the journals lose their function as a reflection instrument and become a more laborious activity report.
• Too many partners. Eight or more boundary partners usually means the programme has not made a strategic choice. Each one carries a real monitoring cost.

Outcome mapping is assessed on the specificity of its behavioural definitions and the honesty of its monitoring. The questions a reviewer asks are: could an outsider observe whether this progress marker has been met? Do the journals record markers that were not achieved, and unexpected outcomes that no marker anticipated? Did the programme change what it was doing in response?

A set of outcome journals in which every partner advances steadily and nothing surprising happens is evidence of a monitoring system that is not being used honestly. Genuine journals contain stalled partners, abandoned strategies and outcomes nobody planned.

Outcome mapping deliberately declines to measure impact. For programmes whose funders require impact claims, it has to be paired with something else, and presenting boundary-partner behaviour change as impact would misrepresent it.

It is also heavy. Twelve steps, three journals and a full marker set per partner is a substantial monitoring apparatus, and small programmes routinely adopt fragments — usually progress markers alone — rather than the whole system. That is legitimate and worth saying out loud rather than pretending to a fidelity that was never there.

And it depends on partner cooperation. Progress markers describe partner behaviour, so monitoring means asking partners about themselves, which is awkward where the relationship is funder-to-grantee and the partner has an interest in the answer.

Outcome harvesting is the retrospective sibling: same definition of an outcome, no requirement that anyone set it up in advance, which makes it the right tool for a programme that never used outcome mapping.

Theory of change work sits upstream and is compatible — outcome mapping can be read as a theory of change with an explicit boundary drawn around what will be monitored.

Most Significant Change complements it well, supplying participant-selected narrative where progress markers supply structured observation.

Contribution analysis is what to reach for when a funder needs an explicit causal claim rather than a record of partner behaviour.

The journals are the data. Over five years, six boundary partners and quarterly reflection cycles, they become a substantial qualitative corpus that has to be read across partners and across time to answer questions like "which strategies preceded movement on the harder markers?".

That is a longitudinal coding problem, and doing it by re-reading Word files is where most outcome-mapping programmes quietly stop learning from their own monitoring. Evidano supports outcome mapping as a named methodology, so journal entries can be analysed against the marker structure rather than as undifferentiated text. Defining the partners and markers stays a strategic judgement the team has to make.