What makes an evidence brief decision-ready?

The defining question is not “How much literature did we collect?” It is “Can the intended reader understand the evidence, its limits and the next defensible action?”

1. Begin with the decision

Before searching, state who needs the brief, the decision they face, the date of that decision and the consequences of getting it wrong. A broad topic such as “digital education in Uganda” is not yet a decision question. A more useful frame identifies the programme, learners, implementation choice, setting and time horizon.

2. Use a method proportionate to the claim

A rapid search can support a rapid evidence assessment if its sources, dates, eligibility and limitations are recorded. It cannot honestly be presented as exhaustive merely because the report is polished. The method and label should match the time, coverage and review actually completed.

3. Link material claims to traceable evidence

The reader should be able to distinguish sourced fact, analysis, inference and recommendation. Citations belong near the claims they support. Extraction tables, data provenance or appendices may preserve the technical trail without overloading the main brief.

Concise writing is not the removal of uncertainty. It is the organisation of uncertainty around a decision.

4. Appraise usefulness, not prestige alone

Peer-reviewed studies may offer stronger methods for some questions, while current government data, programme records or implementation guidance may be essential for others. Source appraisal should consider relevance, design, recency, transparency, bias and transferability to the intended setting.

5. Make uncertainty visible

Use plain statements about evidence gaps, conflicting findings, imprecision, missing groups, weak measurement and limits to transferability. If the evidence cannot distinguish between two choices, that is a finding. It may justify a pilot, additional data collection or a reversible decision rather than a false conclusion.

6. Separate findings, implications and recommendations

Findings report what the sources or analysis show. Implications explain what those findings may mean for the decision. Recommendations propose action and should state assumptions, risks and confidence. Keeping these layers separate allows a reader to disagree constructively.

7. End with action and an update trigger

A useful brief identifies the recommended action, responsible actor, immediate next step, monitoring question and conditions that should trigger review. Evidence changes. A dated brief should explain when new data, policy, safety information or implementation experience would justify an update.