AFRICA / CIVIC PARTICIPATION / 2026
ENFRPT
CIVIC TECH
FUND AFRICA

Practical application

Track public promises without overstating the evidence

A reviewable promise register: source, scope, evidence, response and correction.

The procedures and examples below are learning tools. They describe a possible approach, not a test conducted by the named initiative.

Exact promise and source → Scope and deadline → Qualifying evidence → Assessment and response → Correction history
Track public promises without overstating the evidence · Original booklet · May 2024 Open the full-size diagram
  1. Exact promise and source
  2. Scope and deadline
  3. Qualifying evidence
  4. Assessment and response
  5. Correction history

Define a promise that can be checked

The May 2024 booklet describes SIVIO’s Citizen Watch PCT as an accountability knowledge base and PromessasCV as a way to examine electoral promises. For a new register, preserve the original statement and its source. Separate an aspiration from a specific commitment. Record the responsible institution, place, timeframe and observable deliverable only when the statement actually contains them. Mark missing information as unspecified instead of filling gaps with assumptions.

Use an evidence standard

Create a short rule for each status before judging a promise. “No evidence found” means a documented search found no qualifying material; it is not proof that nothing happened. A procurement announcement may support a planning status, while completion requires evidence of the promised deliverable. Keep conflicting sources and explain how they affect the assessment. Date every observation so readers can distinguish a recent review from an old source.

Allow a response and correction

Give the relevant office a way to provide evidence or challenge an error. Label its response separately from the assessment. Publish corrections with a reason and retain the previous status in a change log. Avoid ratings based on popularity, a political affiliation or social engagement. When comparing promises, explain why they are comparable and which differences in scope or timeframe make a ranking misleading.

Maintain the register

Assign a reviewer and a next-review date. Archive copies or stable references where permitted, and protect private material. Publish the assessment method with the register. Explain that a selected list is not the complete programme of an institution. If a promise cannot be assessed from the available documents, say so. A transparent unresolved record is more useful than an unsupported score.

Working template

  • Exact promise and source
  • Scope and deadline
  • Qualifying evidence
  • Assessment and response
  • Correction history

Illustrative example

A hypothetical promise to repair ten classrooms has a dated statement and an institution’s procurement notice. That supports a procurement stage, not ten completed repairs. An editor would seek completion evidence before changing the final status.

Build your project brief

Sources and provenance

Update a promise record without turning a claim into completion

A promise tracker needs a rule for changing its public wording, not just a coloured status. The African Charter on Democracy, Elections and Governance provides a context of transparency, accountability and participation. It does not prescribe the categories below. This is a practical record-keeping exercise, not an official assessment, a funding opportunity or an audit of a real authority.

Follow one invented commitment

Fictional commitment P01: “Provide six access ramps at six named public buildings by 30 September 2026.” In a real record, preserve the original wording, speaker, date, URL and list of buildings. Here the buildings and documents are deliberately unnamed models. The unit is a completed ramp at a specified building, not a contract, a payment or a photograph.

  1. Version 1, commitment recorded: keep the target of six and the deadline. Outcome evidence has not been supplied. Do not call the promise broken merely because the tracker has not yet received documents.
  2. Version 2, procurement recorded: model document D1 describes a contract for six ramps. Add it as procurement evidence. Public wording becomes “Contract reported for six; completion not established.” Ordering work is not completing it.
  3. Version 3, partial completion claimed: model document D2 reports four completed ramps. Save the four relevant locations and the actual passage. Wording becomes “Completion claimed for four of six; site-level verification pending.” Do not silently change the target from six to four.
  4. Version 4, conflicting account received: model report D3 questions completion at two of the four locations. Record which locations, date and type of evidence it provides. The account is a disputed claim, not proof that two ramps are absent. Keep the conflict visible and request the records needed to resolve it.

A completed status line

P01, review 11 November 2026: target = six ramps; deadline = 30 September 2026; procurement = reported for six in D1; completed work = claimed for four in D2; conflict = D3 disputes two of those locations; verification = unresolved; next action = compare location-specific acceptance documents and dated observations; next review = 18 November 2026. Every date in this paragraph belongs to the invented case.

Copy these fields for your own ledger: record ID ___; original promise ___; unit and places ___; deadline ___; evidence and exact passage ___; competing account ___; wording before/after ___; reason for change ___; checker ___; next review ___. Preserve earlier versions. A corrected source may justify a new conclusion later; the revision history should show why, rather than imply that the tracker always knew the answer.

The fictional target remains six. A contract for six, a claim of four completions and a dispute at two locations are different evidence states.
Model P01: neither procurement nor a reported completion is independent verification. © Civic Tech Fund Africa · CC BY 4.0.

African Union · African Charter on Democracy, Elections and Governance