Burundi

  • Agile Approaches
  • Methodology
  • Overview
  • Findings
  • Research Indicators

Glossary

  • Computer Assisted Personal Interviews (CAPI)

    Computer Assisted Personal Interviews (CAPI) is a survey method where interviewers conduct face-to-face interviews using a handheld computer or tablet to ask questions and record responses. This technique enhances data accuracy and efficiency by allowing for real-time data entry and the use of complex question formats, making it effective for collecting detailed information in various research settings.
  • Computer Assisted Telephone Interviews (CATI)

    Computer Assisted Telephone Interviews (CATI) is a survey method where interviewers conduct telephone interviews using a computer system to manage questions and responses. This approach improves data accuracy and efficiency by reducing human error and allowing for immediate data entry. CATI is widely used in market research and public opinion polling to collect quantitative data from various respondents

Agile Data Indicators

Time

  • Survey Duration

    The mean amount of time it takes for a respondent to complete a survey from start to finish. It is calculated by taking the total time spent by all respondents to complete the survey and dividing it by the number of respondents. This metric is important for assessing the survey’s efficiency.
  • Number of call attempts

    Number of calls made before receiving an answer.
  • Response Rate

    The percentage of individuals that respond to the survey out of the total number of people invited to participate.

    It is a key indicator of the survey’s success and representativeness, reflecting how well the target population was reached and how engaged they were.
  • Survey Cost

    The amount of money saved on each individual survey compared to the traditional survey mode Computer Assisted Personal Interviews (CAPI) cost per survey.

    This metric helps assess the financial efficiency of data collection on a per-unit basis, reflecting how much less it costs to collect data from each respondent.

Cost

  • Data Collection Cost

    The reduction in expenses achieved during the data collection phase of a research project compared to the standard costs.

    It reflects the overall financial benefit of optimizing the data collection process.
  • Sample Size Reached

    Refers to the percentage of individuals that responded to the survey out of the total number of people in the sample.

Validity

  • Data Accuracy

    Inaccuracies or deviations between the recorded values and the true values of the variables being measured in a study.
    Sources of measurement errors include:

    • Respondent errors (e.g., misunderstanding questions, providing inaccurate responses).
    • Interviewer errors (e.g., inconsistent question delivery, leading the respondent).
    • Instrument errors (e.g., faulty survey tools, miscalibrated equipment).
    • Environmental factors (e.g., distractions during data collection).

    Minimizing measurement errors improves data quality and ensures that the findings of a study are more accurate and reliable.
  • Replacement Rate

    The percentage of individuals who were successfully reached by using phone numbers from a secondary or replacement list over the total number of realized interviews.

Methodology

In this project COSA collaborated with JNP Coffee, a specialty coffee importing company working directly with smallholder farmers in Burundi. With a genuine commitment to improving farmer livelihoods, including specific partnerships with women’s cooperatives through the International Women’s Coffee Alliance (IWCA) local chapter, JNP has been working to build a stronger evidence base for its operations and to honor its commitments to prominent buyers such as Nestle around living income standards. COSA assisted JNP to map its farmers with reliable, and timely data. Farmers, buyers, cooperatives, and policymakers alike lack access to accurate information on what it costs to produce coffee, how much income farmers earn from it, and how their overall economic situation evolves throughout the year. Without reliable cost and income data, buyers cannot set prices that reflect the economic realities of production, cooperatives cannot design targeted support programs, and farmers themselves cannot understand or optimize their own financial position. Beyond generating this evidence, the project was designed to test a fundamental question: can mobile technology replace or meaningfully complement traditional enumerator-based surveys in low-income, low-connectivity, low-literacy settings?

Survey

Building on prior successful work in Rwanda, where COSA collaborated with ICO, GIZ, and NAEB to develop a survey to assess cost of production and actual income for coffee farmers, COSA proposed applying a similar but more technology-driven approach in Burundi. Supported by the Gates Foundation through the ADAIM initiative, this project represents the first comprehensive assessment of cost of production and actual income for coffee farmers in Burundi supplying JNP Coffee.

Sampling

The sample was randomly selected from a list of 1,500 farmers provided by JNP Coffee. Those 1,500 farmers were selected proportionally to the different provinces and the gender of the farmers. The sample accounted for both educated and uneducated household heads with at least one member in the household able to read and write.

The sample was originally composed of 400 farmers, with 309 male and 91 female farmers. However, 3 farmers passed away, 2 did not participate in the study, 2 lost their phones, and 1 never participated in the activity module, resulting in a final sample of 394 farmers. Among these, 304 are male and 90 are female farmers. Of the 394 farmers in the final sample, 383 participated in the in-person survey, representing a response rate of 97%, of whom 293 are male and 90 are female.

Technology

COSA deployed the AgileAg App, a mobile platform for real-time, farmer-reported data collection, alongside a traditional in-person survey and Coffee Washing Station (CWS) coffee sales administrative records. The AgileAg App collected high-frequency real-time data throughout the full coffee production cycle, from planting through harvesting and selling (October 2024-July 2025). The App was built with the specific constraints of the Burundian context in mind: very low literacy levels, intermittent internet connectivity, and limited prior exposure to digital tools. A central component of the project was what COSA calls data democracy: the systematic sharing of useful information back to farmers based on their own inputs. Farmers received notifications when their coffee production costs exceeded recommended benchmarks, along with agronomic insights tailored to their reported activities. This reciprocal exchange was intended both to incentivize ongoing participation and to create genuine value for farmers beyond the research objectives of the project.The App was deployed alongside two other data sources: a traditional in-person survey conducted by trained enumerators, and administrative records from Coffee Washing Stations (CWS), which track every delivery made by every farmer throughout the season. This three-source design allowed the project to assess how the App’s data quality compares to established alternatives.

Risk, Resilience and Climate

Business Development

Productivity

Climate Smart and Regenerative Agriculture

Producer livelihood