The Project Experience


These implementation case studies can help projects understand what to expect from the MAD experience.



Header Image
Ano Yonda holds a tablet with Mark Aik and Francis Kui inspect holes in the base of a canarium tree that have been left by a boring (left) and Juponse Bokosou (3rd from left)  underneath canarium trees located at the Kerevat NARI research station outside Rabaul in East New Britain province of Papua New Guinea. Earlier in the year a large group of NARI staff participated in the commcare training delivered by AgImpact as part of an ACIAR short research assignment.

During the MAD Research Series, four projects were supported by Oikoi to scale out the adoption of Mobile Acquired Data (MAD) in research work across the region. Each of these provides a unique case study into the value and costs in adopting digital data collection including feedback from projects on the additional workloads, difficulties in using the technology and perspective from their frontline workers. The projects covered different types of studies, environments and with teams of varying technical aptitude. These pages provide an overview of each project (at a glance), along with what research activities adopted MAD technology, the difficulty reported by project staff, the experience designing, building and deploying their apps, and the associated benefits, challenges and costs.



A scenic view of rice production in the Vietnam Highlands
Vietnam Vegetables - At a Glance


Lao Cai province, Hanoi, Ho Chi Minh and Son La cities, Vietnam
Systems Studied Smallholder vegetable production
Research Activities 2 x 2,500 household surveys (rural and urban, baseline and endline);
Lead Institution  The University of Adelaide
Partner Institutions

Australia: University of Queensland, NSW DPI, Department of Foreign Affairs and Trade

Vietnam: Vietnam Women's Union, Institute of Policy and Strategy for Agriculture and Rural Development, Vietnam National University of Agriculture, Fruit and Vegetable Resource Institute, Soil and Fertiliser Research Institute, National Institute of Medicinal Materials, Plant Protection sub-department of Lao Cai

Disciplinary focus Biophysical (vegetable resource and disease management practices) and socioeconomic (value change analysis and market settings)
Project stage during MAD implementation 3rd year of 4 year project, start of associated PhD study 
Level of MAD experience of project staff High technical capacity among researchers but untrained in MAD at the commencement of adoption
MAD activities Rural household survey with short adoption time and urban household survey with a medium (2-3 months) adoption time
MAD feedback mechanism In-app issue reporting form, real-time feedback via electronic messaging, daily reports to research managers, Prof Wendy Umberger and enumerator team as data was being collected
Key Lessons
  • Adopting MAD apps for this project reportedly saved an equivalent 1250 enumerator days of work 
  • Third party app builders should complete the application in a project space accessible to the project team. Additionally, there should be a requirements gathering and planning phase to ensure the project's needs will be met by the end product. 
  • When outsourcing enumeration have an exchange of policies before the application is built. 
  • High general technical capacity among researchers dramatically reduces the MAD training burden. 
  • Ad-hoc feedback and updates to the application (as opposed to scheduled, batched updates) increase the time required for testing, issue resolution and updates to translations.
  • The use of devices for digital data collection should be leveraged to provide a platform for real time communication between enumerators, field supervisors and researchers.



group of men high five each other
Vanuatu Beef - At a Glance
Project Title Increasing the productivity and market options of smallholder beef cattle farmers in Vanuatu
Location Espiritu Santo, Vanuatu
Systems Studied Smallholder beef production
Research Activities On- and off-farm participatory research, demonstration and training, livelihoods and production systems monitoring
Lead Institution The University of Queensland
Partner Institutions

Australia: Queensland Department of Agriculture, Fisheries and Forestry, Australian Ministry of Agriculture, Southern Cross University

Vanuatu: Ministry of Trade, Commerce, Industry and Tourism, Vanuatu; Agricultural Research and Training Centre, Vanuatu Agriculture College, Ministry of Agriculture, Livestock, Forestry, Fisheries and Biosecurity

Disciplinary focus Multidisciplinary; Livelihoods analysis, cattle economics and value chains, cattle and forage production
Project stage during MAD implementation 1st year of a 4-year project
Level of MAD experience of project staff High technical capacity among researchers, exposure through the Masterclass, but untrained in MAD at the adoption
MAD activities Small scale (36-100 households) livelihoods survey (baseline/endline), farm system survey (baseline/endline), longitudinal cattle production monitoring was developed with a 2-month lead time
MAD feedback mechanism Informal phone conversations and emails, printouts of cattle information sheets for farmers
Key Lessons
  • It is important to consider practices in the field when designing application workflows
  • Building in real time smallholder feedback strengthens the relationship between enumerators and smallholders and improves local perceptions of the project
  • Difficulties in linking data between forms and cases can result from major changes to application structures after collection starts and a lack of real-time data monitoring
  • Benefits of MAD are more obvious at scale (e.g. time saving increases with more fieldwork and longer surveys), but smaller scale projects could maximise benefits of MAD by selecting key parts of their data collection for MAD implementation
  • Face to face intensive training is an effective method for rapid upskilling in app building




Anam Afzal (left) and Sobia Majeed interview a farmer during the commcare survey. The AVCCR team have been working in Jaguwala village for a number of years and due to their close relationships with farmers they conducted the pilot test of their commcare surveys with the farmers of Jaguwala village.
Pakistan Dairy - At a Glance
Project Title Improving smallholder dairy and beef profitability by enhancing farm production and value chain management in Pakistan
Location Punjab and Sindh provinces, Pakistan
Systems Studied Smallholder dairy and beef production within mixed crop-livestock systems
Research Activities Meetings and discussions with potential partner organisations for scale-out activities of whole-family extension approach. Desktop survey for initial evaluations of organisation's management and objectives in extension. Review of current dairy-beef value chain literature in Pakistan. Value chain actor interviews to map beef value-chain of Punjab and Sindh. Review current understanding and efficiency in the dairy and beef operations on smallholders farms 
Lead Institution University of Melbourne, Australia 
Partner Institutions

Pakistan: University of Animal and Veterinary Sciences (Lahore), Sindh Agriculture University 

Australia: University of Melbourne, Charles Sturt University 

Disciplinary focus Value chains research, extension and adoption
Project stage during MAD implementation First year of a four-to-five-year project
Level of MAD experience of project staff Project leader had one year of MAD experience, research staff were new to MAD
MAD activities Survey of value chain actors (farmers, processors & NGOs), monitoring and evaluation of partner organisation extension activities, tracking farmer adoption after extension implementation.
MAD feedback mechanism Surveys provided comparison of farmer-responses with baseline data collected in a previous project. Extension advice for different farmer responses was also pre-programmed into the MAD application and was provided during data collection. 
Key Lessons
  • Allowing project teams (including a least one in-country) time to work together to learn application building under basic guidance hep to build local capacity and places emphasis on their understanding of how MAD applies to their work/research.
  • Important to have clear targets about what is needed in app design otherwise the building process is drawn out. 
  • Start small with one or two builds and data collection activities and take the team through the whole process. This will help to build confidence/understanding. 



Myanmar Department of Agriculture workers are trained to use a CommCare application for the project's endline survey
Myanmar Rice - At a Glance
Project Title Diversification and intensification of rice-based systems in lower Myanmar
Location Ayeyarwady delta, Myanmar
Systems Studied Smallholder rice production
Research Activities Household surveys, benchmarking farmers' fields and establishment of best practices, on-farm rice and rice-pulse trials, and postharvest management training. 
Lead Institution International Rice Research Institute, Philippines (IRRI)
Partner Institutions Myanmar: Department of Agriculture, Department of Agricultural Research
Disciplinary focus Agricultural Economics and Social Science
Project stage during MAD implementation Extension year (4) of 3-year project
Level of MAD experience of project staff Experience with 2 MAD platforms, SurveyBe and CSPro, prior to training in CommCare app building
MAD activities 250 x farmer data sheet (high level endline), 250 x household survey (in-depth endline), monitoring of participation in farmer training and meetings
MAD feedback mechanism In-country researchers reviewed the information daily and debriefed in person with enumerators. 
Key Lessons

MAD skills within an institution.

  • Embedded calculations in apps increase the speed of data analysis and provide valuable feedback to farmers.
  • Surveys developed for one project can become structural template for use in similar projects.
  • Large surveys should be broken down by topic into multiple forms for targeted testing and to minimise the risk of data loss.