The evidence model

The proposed evidence model distinguishes the source document, the statement extracted from it and the reviewer's interpretation. Keeping these separate is important when one document supports several different claims or when later evidence changes an interpretation.

A minimum useful item

Field	Purpose
Statement and category	What is being reported: a result, risk, change, decision, lesson or question
Source and passage	Where the statement came from and what the source actually says
Program and outcome	Which program objective the item may help explain
Date and context	When the observation occurred and which setting it describes
Review status	Whether the proposal is pending, approved, edited or rejected
Limits and related items	Relevant uncertainty, contradictions and connected records

From a claim to its trail

Consider an illustrative update reporting improved outreach attendance. The record should connect that statement to the original update and the schedule change it mentions. It could also connect a stock-shortage risk and a question about who was reached. Each relationship should state its meaning.

The sequence “a change occurred, then a result was reported” is not the same as evidence that the change caused the result. MapToc should help a reviewer investigate the relationship while keeping alternative explanations available.

Evidence gaps belong in the record

Sometimes the important information is what the team cannot yet establish. A source may lack a denominator, omit a location, contain an unverified account or disagree with another observation. Recording that gap is more useful than filling it with an AI-generated explanation.

Continuity and revision

The design calls for source attribution and review history to remain available when records are corrected. Exact versioning, retention and export arrangements must be resolved in implementation and pilot agreements. A source that is later removed or becomes inaccessible should not remain silently presented as fully traceable.

An outcome-focused record

Separate activities, outputs, observed outcomes and longer-term impact. “1,240 women reached” describes reported reach; it does not establish improved access or health. Preserve qualitative changes, negative findings, unintended effects and disagreement alongside positive results.

An outcome entry should retain the intended change, population, place, period, relevant observations, supporting and contradictory sources, assumptions and gaps. Source quality, method, coverage and corroboration should be inspectable rather than collapsed into an unexplained AI score.

Funding relationships identify the grant, allocation, period and relevant summary or approval. They describe context and resource decisions; they must not imply that every dollar caused a particular outcome. Contribution pathways remain interpretations open to revision.