The publication
The decentralized evaluation, commissioned by the WFP Mali Country Office and carried out by the consultancy i-APS, assessed an integrated resilience package across 20 priority communes in Mali, covering 2018 to 2024. Data collection ran from March to April 2025 and the final report, in French, is dated December 2025.
The evaluation combined qualitative and quantitative methods across seven evaluation questions, applying WFP and United Nations Evaluation Group ethical standards, with women-only focus groups and a 400-household survey analysed in SPSS.
Disclosed, not hidden
The use of Evidano was named directly in the report’s data analysis methods, alongside the team’s coding system and a codebook built in Taguette. This reflects the kind of transparent, citable AI use that holds up to methodological scrutiny.
How Evidano fit the workflow
The publication describes a content analysis workflow built around themes, sub-themes, and a researcher-built codebook.
- 1Raw qualitative data
Interviews, focus groups, and written interview notes.
- 2Coding and structuring
A codebook built in Taguette organized transcripts into themes and sub-themes, including gender questions.
- 3AI-assisted analysis with Evidano
The team’s interview notes were analyzed using Evidano as part of the qualitative process.
- 4Triangulated findings
Qualitative insight was triangulated with survey data to strengthen the reliability of conclusions.
“The evaluation team wrote interview notes, which were also analyzed using the AILYZE analysis software.”
“Par ailleurs, l’EE a rédigé des notes d’entretien, lesquelles ont également été analysées à l’aide du logiciel d’analyse AILYZE.”
“La triangulation entre sources de données et méthodes de collecte a été systématisée tout au long du processus, afin de renforcer la fiabilité des conclusions.”
Why researchers look at this
Notes and transcripts coded by hand. Slow to compare across many interviews and groups. Hard to trace each theme back to evidence quickly.
Themes and codes surfaced faster across large qualitative datasets, with researchers reviewing and interpreting the output and the tool named transparently in the methods.
Move interviews, focus groups, and notes toward themes and codes.
Transcription and translation in 100+ languages for multilingual fieldwork. This evaluation was conducted and reported in French.
Use your own codebook and review every output. AI assists; you decide.
End-to-end encryption, no third-party data sharing, no use of your data for AI training.