Most Significant Change (MSC)

What MSC is — and what it is not

Most Significant Change is routinely mistaken for “collecting testimonials”, and the mistake inverts the technique. MSC is a system: stories of significant change are collected from the people closest to the programme in answer to a deliberately open question; panels at successive organisational levels then select the most significant of those stories, recording the criteria and reasons for every choice; and the results of selection are fed back down to the people who told and gathered the stories [1][2]. The story is the raw material. The documented deliberation about which change matters most — and why — is the product.

Two design features do most of the work. First, the open question: MSC does not ask about pre-defined indicators; it asks, in essence, looking back over this period, what do you think was the most significant change in this domain — and why do you consider it significant? That final clause makes every story arrive with a value judgement attached, from the person best placed to make it [1]. Second, the hierarchical selection: forcing a group to choose one story from many, and to say why, converts private assumptions about what the programme is for into an explicit, recorded argument. Organisations discover in these panels that their field staff, managers and board hold different theories of success — which is exactly the discovery the technique exists to produce [1].

MSC therefore belongs to the participatory family: valuation authority is deliberately distributed to storytellers and mixed panels rather than reserved to an evaluator [3]. Where it sits in that family’s control spectrum depends on panel composition — a point commissioners choose, and should choose knowingly (see participatory and empowerment evaluation).

Origins and canon

Rick Davies developed the technique in the mid-1990s for a large NGO programme in Bangladesh, as an approach to monitoring change in a programme whose outcomes were too diverse and unpredictable for a fixed indicator set; Jess Dart’s subsequent adoption and adaptation in Australian settings carried it into wider practice. Their jointly written 2005 Guide consolidated a decade of use across development agencies and remains the reference of record [1][2].

The ten steps, compressed to what teams actually run

The Guide sets out ten implementation steps; in practice they cluster into four working phases, and it is the clusters that teams should plan around [1]:

1. Setting up (steps 1–3). Raise interest and recruit the people who will make the system run; define domains of change — broad, deliberately loose categories such as “changes in people’s lives” or “changes in participation” that tell storytellers where to look without telling them what to find; and fix the reporting period.
2. Collection (step 4). Gather stories through interviews, group sessions or written accounts — who was involved, what happened, when and where, and why the storyteller considers the change significant. Attribution as experienced by the storyteller is captured in the story, not imposed on it.
3. Selection and feedback (steps 5–6). Panels review the stories at each level — typically field teams shortlist, programme management selects again, and a senior or steering group makes a final selection — each panel documenting its criteria and reasons. Results, including the reasons, are fed back to storytellers and lower levels.
4. Strengthening the system (steps 7–10). Verify selected stories where the stakes warrant it; quantify where the stories point to countable change; run secondary analysis across the accumulated story set; and revise domains, questions and panel design in the light of experience.

The selection discussion is the analysis

A common implementation failure is to treat panel selection as administration — a vote, a tally, done. The Guide’s insistence on documenting why a story won is what turns selection into analysis [1]. A recorded rationale (“we chose this story because it shows a change the programme did not plan, sustained without our input”) is a datum about organisational values that can itself be tracked over time: when the criteria used by panels shift, the organisation’s operative theory of success has shifted, whether or not any strategy document says so. Facilitation therefore matters — panels need enough time to disagree, and a scribe whose record captures the argument, not just the result. The accumulated stories-plus-rationales become a corpus for the secondary analysis of step 9, analysable with the standard tools of qualitative methods.

Verification, consent, and the ethics of stories

Stories carry names, places and identifiable lives, which imposes duties indicator tables do not [1]:

• Verification without accusation. Selected stories — especially those that will travel into reports — should be verifiable, and the Guide provides for checking them at source. The visit’s framing matters: it confirms details and gathers depth; it does not treat the storyteller as a suspect whose account must be broken.
• Layered consent. Consent to tell a story to a field officer is not consent to appear, named, in a donor report or on a website. Consent should be re-confirmed at the point of wider use, with anonymisation as the default where there is any doubt.
• Feedback as respect. People who give their stories and never hear what became of them have been mined, not engaged. The feedback step is an accountability obligation as much as a system-design one.

Limits — stated as bluntly as the Guide states them

MSC’s own canon is unusually honest about what the technique cannot do, and the honest list is short but hard [1][2]:

• Positive-story gravity. Storytellers and panels drift toward success unless the design resists it — a domain explicitly for negative changes, prompts that ask for changes “for better or worse”, and panels willing to select an uncomfortable story. An MSC system that has never surfaced a negative change is measuring its own social dynamics, not the programme.
• No prevalence claims. A selected story shows that a change of this kind happened and that the organisation values it. It says nothing about how often it happened, or to how many people. “Our most significant change story shows X, therefore the programme achieves X” is the technique’s standard abuse, and quantification (step 8) exists precisely to send countable claims back to counting.
• Selection effects all the way down. Who is asked, who writes, and who sits on panels all shape the corpus. Documenting these choices is the available control.
• Complement, not replacement. MSC catches the unexpected and the humanly meaningful — exactly what indicator systems miss; indicator systems provide coverage and comparability — exactly what MSC lacks. Run together, each covers the other’s blind side.

MSC is often mentioned alongside outcome harvesting, which also works from evidenced outcome descriptions rather than pre-set indicators — but harvesting is a different technique with its own verification workflow, explained in the Monival outcome harvesting guide rather than here. Within this site’s own cluster, the natural pairing is Outcome Mapping, whose behaviour-change monitoring MSC-style stories can enrich; how selected stories survive into formal reporting is treated under evaluation reports.