Introduction 

Digital Green is a global tech nonprofit that leverages technology and data to help smallholder farmers build prosperous and resilient communities. Since 2008, it has reached over 3.6 million farmers in Ethiopia. For more than a decade, Digital Green has partnered with the Ethiopian Government’s agricultural extension system, training over 20,400 frontline workers using video based extension channels and co-creating digital solutions. Digital Green has launched FarmerChat, an AI-powered agricultural assistant that delivers real-time advisories in farmers' own languages. The organization works to improve farm productivity and increase farmers’ incomes by facilitating critical advisory services and enhancing the efficiency of extension systems through partnerships with public, private, and civil society actors. 

Farmers' adoption of agricultural advice from FarmerChat is a critical outcome in Digital Green's Theory of Change, leading to downstream impacts such as higher yields, increased income, and climate change resilience. Adoption is currently tracked via inapp surveys aligned to the seasonal calendar. Consistently, approximately 70% of active users across geographies report adopting at least some recommended advice; however, two important gaps remain in understanding the true adoption picture. First, the accuracy of this self-reported data has not yet been independently verified and it is not known whether farmers who said they adopted actually applied the recommended practices on their farms. Second, the adoption behavior of active FarmerChat users who did not respond to the in-app survey remains unknown, leaving open the question of whether the reported figure is representative of all active users or only of those who chose to respond. 

This Terms of Reference (TOR) outlines the scope, methodology, and requirements for a consultancy to validate the accuracy of in-app adoption reporting through on-theground field verification, and to assess on-farm adoption among active FarmerChat users who did not respond to the in-app survey. 

2. Objective of the Verification  

2.1. Overall Objective 

The general objective of this verification is to validate the accuracy of self-reported adoption data collected through FarmerChat's in-app surveys, by comparing in-app responses with on-the-ground observations, and to assess whether active FarmerChat users who did not respond to the in-app survey adopted any of the recommended practices on their farms.  The research will generate a practical adjustment factor to adjust future adoption reporting and SROI calculations, thereby strengthening the credibility and accuracy of Digital Green's impact measurement. 

2.2. Specific Objective 

• Determine the verified adoption rate — Establish the true proportion of farmers who actually applied FarmerChat's recommended practices on-farm; 

• Calculate false positive and false negative rates — Quantify the proportion of farmers who reported adopting but did not (false positives), and those who reported not adopting but did (false negatives), to understand the direction and magnitude of self-reporting bias. 

• Understand drivers of adoption and barriers to non-adoption — Identify and document the key factors that influence farmers' decisions to adopt or not adopt FarmerChat recommendations, including perceived usefulness of the advice, access to inputs and resources, extension support, cost considerations, trust in the recommendations, understanding of the advice, and relevance to farmers' production contexts. 

• Identify contextual factors influencing reporting accuracy — Explore factors such as recall period, farm conditions, type of agricultural practice, access to inputs, and extension support that may explain gaps between self-reported and verified adoption across both adopter and non-adopter groups. 

3. Scope of Work

The consultant will be responsible for the full verification cycle, including: 

• Programming a single comprehensive quantitative verification questionnaire in ODK (Open Data Kit), applicable to both adopters and non-adopters, with branching logic to guide enumerators through the relevant sections. The questionnaire will cover application or adoption of agricultural practices, barriers and enablers,observable evidence, as well the source of information about these practices.  The ODK tool will be piloted and refined before full fieldwork deployment; 

• Conducting independent in-person field verification visits (~20–30 minutes per farmer) and document evidence of adoption, if it exists, including taking photos. Enumerators will also collect short qualitative data and will need to be able to take notes; 

• Once the data are collected, perform backend matching by comparing answers from the in-field observation with the answers received previously; ● Performing data cleaning, analysis, and reporting key findings. 

Data collection will be conducted across 4 regions (Oromia, Southern Ethiopia, Central Ethiopia, and Sidama), covering 10 zones, 12 woredas, and 25 kebeles, with sampling geographically clustered within these administrative units to ensure logistical feasibility and adequate representation across diverse agro-ecological and farming contexts. 

4. Methodology 

4.1. Research Approach 

The verification will adopt a mixed methods approach using a single comprehensive structured questionnaire applied consistently to all 400 farmers. This ensures consistency across the sample, enables reliable calculation of the adjustment factor, and supports defensible SROI reporting. A brief set of open-ended questions will capture additional details from the farmers, narrative context and illustrative quotes. 

Data from field interviews will be later analyzed and compared with previous survey responses to produce a complete picture of adoption accuracy for each respondent. 

4.2. Sample 

Target Sample: A total list of 400 farmers will be provided by Digital Green, including both survey respondents and active FarmerChat users who did not respond to the in-app survey.  We will not disclose before the data collection which category (adoption/non adoption) the farmer belongs to in order to mitigate bias. 

4.3. Field Procedures 

Digital Green team will support coordination with the Development Agents in each region and woredas who will notify farmers that they have been selected for the survey. The research team will not be expected to do cold calls or visit any farmers without prior notification. A close coordination with DAs, who are trusted agents in rural communities, will help ensure that field visits and observations can happen smoothly.  

4.4 Data Matching and Analysis 

After all 400 field visits are completed, the analysis will be conducted using Farmer ID as the unique key to compare: 

• In-app adoption survey responses (yes/no, practice category, crop or livestock type; 

• Field verification findings. 

This triangulation produces a complete verification record per farmer, from which the the following classification of practices can be created:  

Classification 

Definition 

Confirmed 

Observation and/or interview independently corroborate the practice. 

Crop/livestock type matches. Attribution to FarmerChat is established. 

Partially confirmed Practice is corroborated but attribution is uncertain (farmer may have learned it elsewhere), or crop/livestock type doesn’t fully match. Unconfirmed No observable evidence and farmer does not provide a convincing account of the practice. Contradicted Evidence or testimony contradicts the survey response (e.g. farmer denies changing the practice they reported). Cannot assess Farmer absent, practice is non-observable and testimony is ambiguous, or data quality issues. 

Primary indicators will include:  

1. Practice-level verification rate: For each practice category, what proportion of claims were confirmed or partially confirmed? This is the most granular and informative measure. 

2. Farmer-level verification rate: Proportion of “Yes” respondents classified as confirmed or partially confirmed overall. 

3. False positive rate: Proportion classified as contradicted or unconfirmed (over-reporting). 

4. False negative rate: Proportion of “No” respondents where the enumerator found evidence of practice adoption that was not reported in the survey. Disaggregate into: (a) adoption with FarmerChat attribution (farmer used the app but answered “No”), (b) adoption with indirect FarmerChat link (learned from another FarmerChat user), and (c) adoption from other sources (genuine non-FarmerChat adoption). This rate is as important as the false positive rate for assessing survey instrument quality. 

5. Attribution strength: Proportion of confirmed adopters who spontaneously named FarmerChat vs. only confirmed when prompted vs. attributed to another source. 

6. Adjusted Adoption Rate: (In-app adoption rate × Verified 'Yes' Rate) + (In-app nonadoption rate × False Negative Rate) 

7. Non-respondent adoption rate: Among active FarmerChat users who did not respond to the in-app survey, the proportion found to have adopted any of FarmerChat's recommended practices on their farms during the verification visit. 

Some disaggregation analysis (by geography, crop, etc) may also be added - this may be discussed with the DG team during the analysis phase.

5. Deliverables & Timeline 

Deliverable  

Description  

Indicative Timeline 

ODK for Quantitative Verification 

ODK-programmed quantitative verification questionnaire with branching logic and a template for taking notes prepared 

19 June 2026 

Data Collection 

Field-level in-person verification visits conducted with all 

400 sampled farmers using the ODK-programmed quantitative verification questionnaire and qualitative questions. Daily data quality checks will be conducted throughout fieldwork to ensure completeness and consistency. 

24 June 2026 

Fieldwork Progress Update 

Summary update on data collection status, initial observations, challenges encountered, and preliminary trends across the 4 regions, 10 zones, 12 woredas, and 25 kebeles. 

6 July 2026 

Draft Findings Report A draft report prepared following the structure and guidance provided by Digital Green, covering all key findings and insights. 

13 July 2026 

Final Findings Report A final report incorporating Digital Green team feedback, covering all key findings and insights. Final cleaned and validated raw data (for reference, QA and any future analysis) submitted alongside the report. 

20 July 2026 

Job Requirements

Team Composition, Expertise, and Technical Skills 

• A multidisciplinary team with expertise in agricultural extension, impact measurement, research methodology, and rural development relevant to smallholder farming contexts. 

• Strong understanding of digital agricultural advisory services, including mobilebased platforms, and their integration with agricultural extension systems. 

• Experience in behavioral and adoption research, including understanding of social desirability bias, blinded data collection, and field verification methodologies, including structured observation protocols and triangulation of multiple data sources. 

• The team must include or be led by professionals with a background in agriculture or agricultural extension, capable of supervising and qualityassuring field verification of on-farm practices. 

• Proven capacity to analyze quantitative data and calculate adoption metrics, false positive/negative rates, and to produce high-quality, actionable findings 

• Experience with collecting and analyzing qualitative data 

• Demonstrated ability to work in rural Ethiopia and coordinate effectively with the selected regions and kebeles. 

• Field enumerators must hold a minimum diploma or degree in agriculture, agricultural extension, agronomy, or a related field, with practical knowledge of the practices being verified (fertilizer application, pest and disease control, and livestock management, etc..), prior experience in rural Ethiopian farming communities, fluency in local languages of the study areas. 

7. Client Responsibilities (Digital Green) 

• Provide a list of sample farmers and geographic details (plus a back up/replacement list of farmers) 

• Provide a survey instrument including quantitative and qualitative questions Facilitate communication and coordination with key stakeholders, including extension agents, frontline workers, and local woreda/block agriculture offices. 

• Share database for comparison/matching analysis 

• Monitor the overall quality of the consultant's work to ensure adherence to this Terms of Reference.   

8. Application and Instructions 

8.1. Proposal Format and Content 

The consultant is expected to prepare the proposal in a clear and efficient manner, providing a concise description of their capacity to meet the requirements outlined in this Terms of Reference (TOR). 

Qualifications and Related Experience (General Part) 

This section should outline the consultant's background, organizational profile, and relevant experience in conducting large-scale quantitative and mixed-method field verification or evaluation studies in the agriculture and rural development sector. It should highlight previous work related to digital agricultural advisory tools, adoption verification, agricultural practice monitoring, and field-based impact measurement in smallholder farming contexts. Include details on team composition, key personnel qualifications, examples of similar studies, and client references. 

Technical Approach / Work Plan (Technical Part) 

This section should present a clear methodology for delivering the verification, including the data collection approach, ODK tool design, field verification procedures, backend data matching and analysis plan, and approach for calculating key indicators. A detailed work plan with timelines aligned to the deliverables schedule in Section 5 must be included. Cost Details (Cost Part) 

This section should provide a transparent breakdown of all anticipated costs, including personnel, travel, data collection, analysis, reporting, and any applicable taxes. Costs should be presented clearly in Ethiopian Birr (ETB) linked to the activities described in the Technical Approach. 

 How To Apply

8.2. Submission of Proposal 

The proposals must be submitted no later than June 10, 2026, 2:00pm (8:00 local time) to the following address: Digital Green Foundation – Ethiopia Office, Yeka SubCity, Woreda 07, House Number 934/01, Haile Gebreselassie Avenue Waryt Building, 

6th Floor, Addis Ababa, Ethiopia 

8.3. Evaluation Criteria 

• Technical proposal =70% (Technical Capability 50% and Demonstrated qualification and experience = 20% 

• Financial proposal= 30% 

9. Payment Terms 

The payment terms for the consultancy services will be negotiated and agreed upon in a separate agreement. 

Job Requirements Digital Green seeks a consultant to verify the accuracy of self‑reported adoption data from its FarmerChat platform. The role involves designing and piloting an ODK questionnaire, conducting 400 in‑field verification visits across four Ethiopian regions, collecting photographic evidence, matching field observations with app responses, and producing analytical reports on true adoption rates, bias metrics, and contextual drivers. How to Apply Submit your proposal by June 10, 2026, 2:00 pm local time to the Digital Green Foundation – Ethiopia Office at Yeka SubCity, Woreda 07, House Number 934/01, Haile Gebreselassie Avenue, Waryt Building, 6th Floor, Addis Ababa, Ethiopia.