Translate funder requirements, standards, and methods into approved measures, traceable evidence sources, visible gaps, and collection-ready decisions.
SOPACT ACADEMY · IMPACT MEASUREMENT · DEFINE
To turn reporting requirements into evidence you can collect, begin with the report—not a framework catalog. Extract each required claim, measure, breakdown, narrative, and method from the funder’s documents and the authoritative standards it names. Connect each requirement to an approved organization measure, then identify the quantitative, qualitative, documentary, and longitudinal evidence that can support it. Record gaps and human decisions. The result is a Reporting-to-Evidence Map that guides collection and generates different reports from the same governed evidence.
This lesson is for nonprofit, foundation, program, MEL, grant, and impact leads who encounter terms such as SDG, IRIS+, GRI, Five Dimensions of Impact, SROI, schema, and data dictionary—but need a practical way to use them without first becoming standards specialists.
What you will produce
In short: Reporting requirements arrive in different languages and formats, while the evidence lives across different people and systems. A shared evidence map translates both sides: what each audience asks for and which governed evidence can support the answer.
The goal is not one universal report. It is one governed evidence base that can produce audience-specific outputs without redefining the evidence each time.
In short: A data dictionary is the governed version of your shared evidence language. It records what each item means, how it is measured, where it comes from, who owns it, and which reporting requirements it can support.
You do not need to begin with field IDs or database types. Begin with the confusion the dictionary prevents. “Participant,” “completed,” “employed,” “person with a disability,” and “retained at 90 days” must mean the same thing when they appear in a program form, partner PDF, board dashboard, and funder report. Chapter 5 approved those meanings. This chapter connects them to reporting requirements and evidence sources.
The technical implementation—stable IDs, allowed values, validation, missing-data rules, access classification, and version history—still matters. It belongs in the supporting guide, How to Build a Data Dictionary for Program and Impact Data, after the team understands the practical job.
In short: No. They are connected layers with different jobs. Treating them as synonyms makes the system harder to govern and easier to misapply.
In short: They are not peer metric lists and they are not interchangeable. Each contributes a different kind of reporting or measurement guidance. Use only what applies to the decision and document the source and version.
| Source | Plain-language job | What not to assume |
|---|---|---|
| SDG indicators | Official global indicators and metadata used to follow progress on the 2030 Agenda | Selecting an SDG icon does not create a program measure or prove contribution |
| IRIS+ | Impact-investing guidance, Core Metrics Sets, taxonomy, and qualitative and quantitative metric catalog | A similar metric name is not an exact mapping |
| GRI | A modular sustainability-reporting system with Universal, Sector, and Topic Standards | GRI is not merely a list of impact-program metrics |
| Five Dimensions of Impact | Organizing norms for considering What, Who, How Much, Contribution, and Risk | A framework does not supply every field, instrument, or reporting rule |
| SROI | A social-value methodology involving stakeholders, outcomes, materiality, valuation, transparency, verification, and avoiding overclaiming | SROI is not simply assigning money to an output and calculating a ratio |
| Custom measures | Organization-, service-, community-, or theme-specific evidence external sources do not capture | Custom does not mean ungoverned or less credible |
The official UN SDG metadata repository, IRIS+ overview, GRI Standards, Impact Frontiers norms, and Social Value International guidance describe these purposes in detail. Standards and methods evolve; record the version and revalidate before relying on a mapping or formal claim.
In short: Many reporting claims require numbers and context together. Useful evidence can be structured, qualitative, documentary, and longitudinal. The shared structure must preserve the source, date, person or entity, program stage, access rule, and citation—not force everything into a numeric column.
In short: Translate each reporting requirement into an approved measure or question, then identify the source and gap. Do not begin by adding every available disability, employment, or SDG indicator to a survey.
Illustrative scenario: A workforce program must report how participants with disabilities enroll, complete training, enter employment, and retain employment at 90 days. The funder also wants evidence about barriers and the program’s response, plus references to selected SDG and IRIS+ measures.
| Reporting requirement | Approved measure or question | Evidence source | Mapping status | Gap or next action |
|---|---|---|---|---|
| Participants with disabilities enrolled | Confirmed enrollment, disaggregated using the approved disability measure | Consent + intake + enrollment record | Candidate | Approve instrument, language, consent, threshold, and access |
| Training completed | Completion under Chapter 5’s approved event and denominator | Attendance + assessment + completion record | Organization-approved | Confirm partner sites use the same rule |
| Entered employment | Employment started within the approved window | Participant record + employer verification email | IRIS+ candidate | Compare population, window, numerator, denominator, and version |
| Retained at 90 days | Employed on the 90-day follow-up date under the approved rule | Follow-up record + employer or participant confirmation | Organization-approved | Separate unreachable from not retained |
| Barriers participants experienced | “What made it difficult to participate or remain employed?” | Open text + interview + case note | Custom qualitative | Preserve exact words; human-review themes |
| How the program responded | Barrier → action → owner → date → result | Case action, meeting note, referral, partner email | Custom narrative | Link each response to evidence and timestamp |
If the reporting purpose requires comparable disability disaggregation, use a suitable approved instrument rather than asking staff or AI to infer disability from narrative evidence. The Washington Group Short Set on Functioning was developed for censuses and surveys to produce comparable functioning data; WHO’s WHODAS 2.0 is a standardized assessment of functioning and disability. Neither is a universal substitute for deciding the reporting purpose, population, consent, administration mode, language, safeguarding, and local applicability.
In short: AI can reduce the first-pass reading burden by extracting candidate requirements, comparing definitions, inventorying evidence, and flagging gaps. It cannot decide which standard applies, establish compliance, validate causality, or approve a sensitive measure.
A team can build this map manually in a spreadsheet and shared folder. Friction grows when each funder has different documents, evidence is spread across staff and partner systems, the same requirement recurs under different wording, and a report must be regenerated after new evidence arrives.
Sopact Sense can hold the grant documents, official reference extracts, approved organization definitions, forms, PDFs, interviews, emails, and report evidence in one governed workspace. It can extract candidate requirements with citations, apply approved mappings, flag missing or contradictory sources, and generate different audience views from the same evidence. People validate the standards mapping, approve definitions, review sensitive interpretations, and retain final authority for the report.
For high-stakes reporting, test extraction and qualitative coding across languages, document overrides, restrict access to sensitive participant data, establish retention rules, and keep the source text and citation available for review. AI should make the work more traceable—not make unsupported claims sound more polished.
Identify the report and authoritative sources, extract each measure, breakdown, narrative, method, and document requirement, connect each to an approved organization definition, identify the evidence source and collection moment, and record gaps and human decisions. Test whether the resulting evidence can support the required output with citations and visible limitations.
It is the organization’s shared evidence language written down and governed. It says what every approved measure or evidence item means, how it is calculated or interpreted, where it comes from, who owns it, and where it can be used. Technical fields implement that agreement; they are not the starting explanation.
Not necessarily. Different reports may legitimately use different standards, frameworks, and methods. Govern the organization’s definitions once, then document how each external requirement relates to them. Do not force unrelated requirements into one framework or claim equivalence when populations, boundaries, periods, or calculations differ.
AI can propose candidates and show definition differences, but it should not make the final selection. A responsible owner must confirm the reporting purpose, applicable source and version, population, unit, period, boundary, calculation, and disclosure requirements against the official material.
No. A custom metric may be essential when an external standard does not capture the service, community, or decision. Credibility comes from a clear definition, appropriate collection, stakeholder relevance, documented calculation, traceable evidence, consistent use, and honest limits—not from attaching a familiar label.
Yes, when its role and limitations are clear. Preserve the source, author or speaker, date, context, access rule, and citation. A document can support a claim, reveal a requirement, or provide narrative context, but it is not automatically accurate or sufficient merely because it exists.
Not automatically. Preserve the participant’s exact words. An approved coding method may group responses into themes, counts, or ordered categories, but the code and source text should remain linked and interpretations should be human-reviewed. Never infer a quantitative value or sensitive attribute that the respondent did not provide.
No. SROI is a methodology for analyzing social value. It requires stakeholder involvement, evidence of outcomes, judgments about materiality and value, attention to contribution and overclaiming, transparency, and verification. The evidence map can preserve inputs for a later SROI analysis, but it should not manufacture a monetary value before the method is properly applied.
Turn each evidence requirement into a collection moment inside the real workflow: enrollment, attendance, service delivery, follow-up, interview, case note, partner report, or document review. Define who collects it, what validation occurs, how identity and time are linked, and when the evidence becomes available for analysis.
Author: Sopact (Unmesh Sheth). Drafted August 2026. Standards and methodologies change; verify the authoritative version and applicability before formal reporting, assurance, regulatory filing, or SROI analysis.
Next: You now know what every report requires, which approved measure answers it, where the evidence lives, and what is missing. Put that plan inside enrollment, service delivery, follow-up, interviews, and partner workflows in How Do You Collect Clean Evidence Inside the Workflow? →
ChatGPT, Claude, and Gemini are fine for a quick test — but not for an answer you'll put in front of a funder or board. When it has to hold up, run it in Sopact Sense.
Build your evidence map →
