M&E Specialist with over 7 years of experience in research, programme evaluation, and data-informed decision-making within the non-profit and government sectors. Contributed to more than 10 national evaluations, primarily focused on government policies related to economic development and transformative equity.
Publicado el 07/12/2025
My experience with the Three Horizons (3H) Initiative at the International Evaluation Academy (IEAc) has reinforced a key lesson: evaluation practice must become more transformational and shift away from conventional approaches that have long been shaped by Global North ideologies.”
In trying to find solutions on how evaluations can be fit for purpose, various points on localising evidence were brought up.
Firstly, localising impact evaluations means shifting power towards local actors and local knowledge. Components such as evaluation questions should be shaped by communities, local governments, and indigenous knowledge holders.
Secondly, particularly in the Global South context, approaches such as storytelling, oral histories, communal dialogue, and participatory narrative methods should sit alongside quantitative and experimental designs. These reflect how many African communities make sense of change and offer culturally grounded insights that traditional methods often miss.
Last but not least, respect for cultural protocols, indigenous and community consent ensures that the evaluation serves the people it studies, not external agendas.
Using the Three Horizons framework while centring African and indigenous knowledge can help create evaluations that are culturally rooted, locally owned, and better aligned with the futures communities themselves envision.
South Africa
Sibongile Sithole
Evaluation Consultant
Publicado el 26/10/2026
Indeed, the real value of AI isn’t just doing the same work faster, but freeing up evaluators’ time for context, interpretation and judgement. Whilst this may be ideal, AI remains exclusive in many cases. Most grassroots and development communities, where donor-funded programmes are usually situated, lack access to AI, yet they are often the same people most affected by funding cuts. It is therefore important to consider how we can avoid their marginalisation, because the very same AI that can be useful could further widen the divide. If AI-supported evaluations draw mostly on documents and datasets produced by donors and implementing agencies, community voices risk being even further removed from the evidence base.