Method

Scope. Design.
Deliver. Measure.
Verify.

The framework every Torchbearer programme runs through. It is published in full because a method that cannot be examined is not a method — it is a claim.

Students working at their desks

Most programmes fail at stage one — designed against an assumption instead of a baseline

Why a framework

Good intentions
are not a method

Social and environmental programmes fail in predictable ways. They are designed around an assumption nobody tested. They measure activity because activity is easy to count. They report a change without establishing what things looked like beforehand. They depend on staff who leave.

None of that is caused by a lack of care. It is caused by the absence of a structure that forces the awkward questions early, when they are still cheap to answer.

The five stages below are that structure. Each has a defined output and a standard that has to be met before the next stage begins.

The five stages

Scope

Establish what is actually true before designing anything. Who is affected, how many, what they currently do, and what has already been tried in the same district by someone else.

Output: a written problem statement, a population baseline, a map of existing provision, and an explicit list of what will not be measurable.
Standard: the problem statement is agreed with the local partner before design begins.

Design

Write the programme as a causal chain — this input, producing this behaviour change, producing this outcome — and attach an indicator to each link. Indicators are drawn from named SDG targets and from the frameworks the partner already reports into, so results arrive without needing translation.

Output: programme design document, measurement plan with instruments, safeguarding and data-protection plan, cost-per-outcome model, risk register.
Standard: nothing is procured until the cost per outcome is stated and the sensitivities that would break it are written down.

Deliver

Execution through local organisations and locally recruited teams. Product and education are delivered together — distribution on its own is not an intervention and is not reported as one. Delivery data is captured as the work happens, not reconstructed afterwards.

Output: delivery against plan, operational data, incident and safeguarding log, variance notes.
Standard: every person delivering on our behalf, including partner staff and volunteers, is inducted on the safeguarding and conduct standard first.

Measure

Baseline at intake, endline at a fixed interval, same instrument, same population. Where a control or comparison group is feasible we use one; where it is not, the reporting says so rather than implying a counterfactual we did not test. Attrition is reported as a number.

Output: measured change against each indicator, attrition and exclusions, and the confidence we are entitled to claim.
Standard: the people who delivered the programme do not sign off its results.

Verify

Assemble the evidence so a third party can reconstruct the result without speaking to us: raw instruments, method notes, exclusions, and the reasoning behind every adjustment. Where scale or partner requirements warrant it, verification is performed by a party with no stake in the outcome.

Output: a verification pack suitable for audit, assurance or disclosure use, and a published programme review including what underperformed.
Standard: the pack is complete enough that a sceptical reader can disagree with our conclusion using our own data.

What we publish
for every programme

The same set, whether the programme went well or badly. Published on the Evidence page as each programme completes its endline.

ItemWhat it containsWhen
Problem statementThe tested position at scoping, including what we could not measureAt design
Indicator setEvery indicator, with its SDG target and national framework mappingAt design
Cost per outcomeModelled at design, restated against actuals at closeDesign and close
BaselinePopulation, instrument, date, sample and methodBefore delivery
EndlineMeasured change, attrition, exclusions, confidence claimedAt interval
Programme reviewWhat worked, what underperformed, what we would changeAt close
Verification packRaw instruments, method notes, adjustments and reasoningAt close

Participant-level data is never published and never shared with a funder or partner in identifiable form. Consent for imagery is taken separately from consent to participate, and can be withdrawn at any point.

The framework in full

The complete document — stage gates, instrument templates, safeguarding standard and data-protection position — is provided to partners as part of any engagement, and to anyone else who asks.

Request the framework See the alignment mapping