From evidence to action. Design development interventions informed by what actually worked — and what didn’t.
Methods under review. Success percentages shown here count lessons, not independent projects — projects with longer reports contribute more extracted lessons. Lessons from highly rated projects are labelled “design principles” and those from poorly rated projects as “pitfalls”, but successful projects can describe mistakes and unsuccessful ones can contain effective components. These are historical associations for context, not validated forecasts of new project outcomes. A May 2026 World Bank paper applies similar text analytics to 7,500+ IEG evaluations; this is not a unique capability.
How this works. Select a development challenge and region. The Design Lab reads 3,043 lessons from 991 independently evaluated projects and organises them by evaluation outcome: lessons from projects rated satisfactory or above, and lessons from projects rated unsatisfactory or below. Every lesson links to the project it comes from and its independent evaluation rating.
Why this exists. Looking at evidence is not the same as doing something with it. This tool does the extraction and organisation so a designer can see what evaluators observed, rather than starting from assumptions.
Select a development challenge above to generate evidence-based design guidance.
Try: Energy → East Asia and Pacific | WASH → Sub-Saharan Africa | Agriculture → South Asia
Complete design brief available.See the first evidence-based design brief: Rural Electrification in Pacific Small Island States — 44 lessons from 16 countries, with risk assessment, implementation checklist, and monitoring framework.
About this tool
Source. 3,043 lessons extracted from 991 World Bank Implementation Completion Report Reviews (ICR Reviews). Each project has been independently evaluated and given an outcome rating from “highly unsatisfactory” to “highly satisfactory.”
Method. Lessons are classified as design principles when they come from projects rated satisfactory or above, and as pitfalls when they come from projects rated unsatisfactory or below. Moderately rated projects contribute implementation guidance. Sector and topic classification uses keyword matching (~72% coverage). All filtering and synthesis happens in your browser.
Limitations. Lessons are pattern-extracted, not manually curated. They reflect what World Bank evaluators identified, which is one perspective. Coverage is 991 of 7,300+ available ICR Reviews. The checklist is derived from patterns, not prescribed best practice. Local context always matters more than global patterns.