Iterating Learning & Impact Through Questions We Grow

Evidence informs learning, and applied learning magnifies impact

As an organization we endeavor at all times to test and utilize evidence to inform our decision-making, improve our platforms and increase our impact.

As articulated by Michael Kremer, experimentation allows us to “test new approaches, refine them, and scale up the most effective solutions” - this is our lodestar.

To achieve this goal, we gather and analyze data, and deploy tests and evaluations to generate evidence of impact. We aim not only to filter out success from failure, but to learn and to make this learning widely available.

We value failure just as much as success. No question is too big (or too small) to fail, and relative failure may spark new questions that lead to unforeseen successes.

Our underlying theory in all of our questioning is that improved communications - farmers receiving the right information at the right time - and farmer knowledge leads to improved behavior, which leads to improved agricultural outcomes. We aim to establish more and more evidence along each point of this theory of change. When we are confident that we have isolated an effective approach, we then leverage our partnerships and systems to achieve and generalize the impact at scale.

A/B testing – experimentally comparing two or more service design options to assess which is preferred or more effective – allows for near instantaneous upgrading of content and service delivery to improve user experience and deliver more appropriate information. The use of rigorous assessment tools such as randomized controlled trials (RCTs) provide opportunities to systematically understand impact and refine our model over time. We also measure impact of our service on farmer behavior and agricultural outcomes whenever possible. Furthermore, gathering farmer feedback throughout the testing cycle helps us to identify opportunities to further refine our services.

The process of learning to which we aspire leverages local context and farmer feedback, rigorous testing and evidence, as well as economic theory to advance the quality of advisory content, our service delivery and impact.

A critical advantage of digital systems is that mobile phones allow for the collection of large datasets from users which can be harnessed for rapid experimentation and analysis. In turn, digital systems themselves are capable of being rapidly iterated to improve user experience and impact.

The following three cases highlight examples of our research and learning agenda in 2019:

Case 1: Leveraging A/B Tests to improve tools & access to information

Case 2: Testing the use of SMS to Promote Improved Nutrition

Case 3: Testing Remote Sensing to Estimate Yields