Following a summary and excerpt of the very interesting study “Digital and Data-Driven Agriculture: Harnessing the Power of Data for Smallholders” co-published by @GFAR (Global Forum on Agricultural Research and Innovation), @GODAN and @CTA (Authors listed at the end of this article)

While I strongly recommend to read it all, the reader may find in this article find a paper summary including the key delivered messages (according to my opinion).

Digital Agriculture

Digital Agriculture defines as “having the right farm data, at the right time, to make a better decision“.

For farmers, data needs to be transformed into information, and this information used with experience as knowledge. A core process is learning. This is not linear and applied or adapted based on the farmer’s own knowledge base. These learning and adaptation capacities are critical for farmers to bene t from data-driven agriculture opportunities.

Generally, the ability to manage and effectively use data and information reflects the power of farmers in agri-food systems and their participation in market chains.

Challenges for Smallholders

Smallholders farmers are significantly more affected by unequal or insufficient information, unpredictable environmental changes in rainfall, soil erosion, scarcity of human labor with increased cost, yield loss due to pests and insects, increase in cultivation costs, as well as poor and perhaps exploitative supply and market chain management.

Larger farmers (with access to better infrastructure and supporting farmer organizations) are, generally, better placed on these issues. Generally, the ability to manage and effectively use data and information reflects the power of farmers in agri-food systems and their participation in market chains.

Larger size farms seem indeed to be linked to the higher adoption of precision agriculture. A number of reasons for this include:

  • Cost and affordability for larger operations
  • Greater efficiencies and return on investment over a larger site or multiple sites
  • Availability of expert or trained personnel, some of whom can use their expertise over multiple sites.
  • A higher level of support from vendors,

The implication for smallholders is that they need to find ways to obtain the same type of adoption advantages.

As precision farming affordable for small farms are just emerging, smallholders will continue to operate at a disadvantage until the technologies become more smallholder-ready and the farmers more data-savvy.

As public and private institutions themselves become more data-driven, there is a danger that smallholders, unless included in the data environment, will fall out of the normal systems. Lacking data on their situations, smallholders are unlikely to convince banks, government and credit providers, for example, that they are good investments.

One possible solution might be to create networks of farmers that collect farmers’ data and use it to create credit models that allow banks to vet farmers for loans, grants or other public entitlements.

The trend in economically-developed markets (rapidly spreading in developing countries) is for corporate-dominated agri-food chains. These tend to have few value chain nodes that closely link producers, supermarkets, and consumers. From a data management perspective, having fewer nodes reduces the complexity of improving efficiencies of information flows. It also makes it easier to manage and control, and potentially exclude, actors in the agri-food system (such as farmers).

For smallholder farmers to benefit and overcome these challenges, data-driven agriculture must aim at symmetrybalance, and equity in the data flow and its use among all actors and stakeholders.

There is an urgent need for data-driven applications and systems that are smallholder-ready.

Making data-driven agriculture smallholder-friendly needs to be guided by critical factors that drive agri-food and data systems. Getting these ‘right’ will help ensure that data-driven agriculture really benefits smallholder farmers:

  • Appropriate policies and institutions;
  • Positive incentives;
  • Capacities at different levels;
  • Available, affordable hardware, software and data. All the necessary tools including hardware such as sensor-equipped farm machinery, software such as applications for the local farming system and cropping pattern, and data needed by the applications needs to be affordable for the smallholder sector.
  • Needs-driven data and software. More applicable software, models and data can overcome the challenges related to the relevance of data and the usefulness of software and web applications. This also entails more realistic and focused research, not only related to agricultural data guiding the farmer but also presenting data and information on more user-friendly platforms. It is important that farmers or their representatives are involved in the design of apps, tools, and software. tools, and software.
  • Accessible infrastructure

Data system drivers to enable data-driven agriculture among smallholders farmers

APPS FOR FARMERS: successful apps must provide localized and specific solutions and therefore use more localized and specific data. Data-driven mobile applications should be built on user requirements and during application development, the user should always be placed at the center of the designThe most appropriate approach is to develop platforms where all apps can be standardized to be interoperable with the sharing of data and information and offer farmers solutions with a basket of options that are location specific. Such platforms will also enable a level playing field for (local) entrepreneurs to offer data and information services.

OPEN DATA: in such open approaches, data should be more open and more accessible; smallholders should also get fair returns from their data. This is not just about farmers selling their data. It could also mean negotiating cheaper and better services. Putting a price on data is not always the best solution as the rich will always have an advantage over the poor, so risks of monopoly can increase. The open data movement can also add value through its work on data standards.

BIG DATA, INTERNET OF THINGS: An important aspect of big data is that it allows precise ‘analytics’ of this data to discover trends and meaning that can be used to model, build and deliver targeted services. Big Data platforms are essential to handle the amount of data generated by the Internet of Things (IoT), which is all the data coming from all the interconnected ‘things’ that send data over the Internet. In agriculture, big data and IoT are currently mainly associated with information collected by sensors, satellites or drones combined with genomic information or climate data, which can all help farmers optimize their farm operations.

Developing Data Ecosystems for Smallholder Farmers

1. Aggregating farmer data through joint action that empowers and gives voice to farmers

Cooperatives, producer organizations, farmer organizations have important roles to promote and enable aggregation of farmer initiatives, including managing, sharing and using data and information.

Finding, testing and implementing sustainable business models for farmers and farmer organizations to benefit from data is an urgent action. Farmer-representing organizations need to transition towards digitally-smart organizations or ‘data cooperatives’ that are able to broker data-driven interactions between and among their members and with external organizations of all types.

Besides any local aggregation of farmers, the ‘virtual’ aggregation of farms could facilitate more synchronized access to farm inputs, processes, outputs and logistics to participate in markets through the use of ICTs.

2. Establishing trust centers, mechanisms and platforms to enable and regulate open data sharing at different levels

Setting up trust centers and platforms (commodity-specific, value-chain segment-specific regional, national…) would help bring together and facilitate necessary joint efforts and actions to improve the ways that data is accessed, shared and applied in agriculture and nutrition. Trust centers could take on several important roles and issues, such as:

• Innovation and experimentation: Test and disseminate innovative solutions to ensure efficient and fair sharing of data such as: state-of-the-art technologies to ensure efficiency (like big data platforms) mechanisms to ensure transparency and recognition of ownership (like blockchain); engage technology providers and incubators to become part of the trust center or negotiate outsourced technology solutions.

Agri-food system

Farming and agri-food systems are complex, dynamic systems with hundreds of component sub-systems, each with multiple actors and stakeholders, each with different roles and expectations. Key stakeholders of the Agri-food system:

  • input suppliers
  • producers
  • processors
  • traders to consumers

These agri-food systems are bound together through flows of finance, commodities, and information.

Keys are natural agricultural resource capability, the potential to farm inputs, consumption or wastage within and across the chains.

Imperfect flows, of data and information, can cause turbulence and failure in agri-food related markets. Transparency of these flows with more inclusive and open access to these data can help mitigate the risks of such turbulence and failure.

Data is thus both a critical input as well as a valuable product in the modern agri-food system.

Open data enables and empowersOpen data may also pose a threat if collected and applied without clearly defined principles rules, and ethics, making farmers even more vulnerable to the asymmetries of financial, commodity and information flows in agri-food chains, undermining farmers’ livelihoods.

Types of “digital agriculture” data

1. LOCALIZED DATA: generated and collected INSIDE the farm, for use only INSIDE the farm

They include:

  • soil data (soil form; soil depth; nutrient composition),
  • seed and fertilizer use,
  • date of sowing,
  • production practices,
  • water use,
  • etc.

Data that farmers have about their immediate location. This data can be generated and managed by the farmer or by an agent acting for the farmer. It would normally be ‘owned’ by the farmer.

The data collected, managed and used in farming can be static, such as land ownership and farm eld boundaries, or dynamic, which changes over time. Some dynamic data can have very short (daily weather data) or very long lives (soil nutrient values).

2. IMPORTED DATA: generated and collected OUT OF the farm, for use INSIDE the farm

Examples are climatic data and market prices that have been interpreted and customized for on-farm use.

This data is usually owned, managed and controlled by a third party and made available, directly or through intermediaries, to farmers (and their representatives). Farmers do not ‘own’ this data unless they have purchased it; they have permission to exploit it. These services are often private or semi-private and they seek effective business models which often include explicit sales of data.

3. EXPORTED DATA: generated and collected INSIDE the farm for use OUTSIDE the farm

This is usually processedaggregated or combined (often anonymized) with other data and information generated elsewhere and is used by various actors and stakeholders, such as governments or private companies.

Other users include market intermediaries, farm input and service providers including banks, insurance agencies, farm advisory services, scientists, other farmers and their associations etc. They each have their own need for data ‘from’ farms.

While collected from farmers (or their farms using sophisticated tools like drones or remote sensing), this data and the products and services it generates are usually owned by a third party. It can be made more widely accessible by making it ‘open’. There are growing concerns to safeguard this data for farmers to ensure it is not exploited at the cost of the farmer. Some data collectors, such as scientists, have adopted ethical guidelines to make sure the data they collect does not exploit the farmers who provide it.

Some private companies have procedures that copy data they collect (through smart equipment and sensors for instance) to data repositories that the farmers can own and access.

4. ANCILLARY DATA: generated and collected (IN and) OFF the farm, mainly for use IN the farm

A large proportion of ‘agricultural’ data such as government statistical and research data are generated using various data sources, in another value chain component perhaps, and may have little direct on-farm application.

Data in crop farming cycles

For the data to be relevant and useful its context, it needs to be associated. Data loses its context if separated from the information a farmer needs and uses. This is sometimes seen, for example, when farmers are given raw weather data without the context of what the data means for a farming operation. For example, farmers do not see the usefulness in just rainfall data. Its use must be interpreted in the agronomic context of whether the pattern of rainfall is adequate to initiate sowing of a particular crop at a particular time. The context of rainfall data may be different if the crop, for example, is rice or cotton. Data should, therefore, become information in an adaptive context.

Example of data by types

Uses for Data in Farming


What: What to produce, when to produce, where to produce? For whom to produce? What operations to do when and where on the farm?

Data types: imported data, localized data

Monitoring and assessment

What: How is the product growing? What is the status of the natural agricultural resources?

Data types: localized data, (exported data).

Event management and intervention

What: Which action should be taken and when?

Data types: localized data, (external data like weather forecasts, growth models or market conditions)

Autonomous action through ICTs

What: for example, switching on water pumps to irrigate fields when soil humidity falls below a target amount, opening or closing windows in glasshouses, …

Data types: localized data, (external data like weather forecasts, growth models or market conditions)


What: what will be the economic, environmental or social return/effect on the investment/action?

Data types: imported data (like market data, consumption statistics, land and water use, potential payment schemes for environmental services)


What: how much will be the crop or animal yield? How much profit?

Data types: imported data, localized data (to enable prediction models: the more data the farmer (or the farmer’s adviser) has, the more accurate the forecast is).

Tracking and tracing

What: where is the product, item, resource or material? What is its source and where will it go next?

Data types: imported data (exported data – the data shared from the farm can also become essential for the overall tracking data flows (e.g. farm identification, farming practices, the agricultural input used))

Negotiating and market access

What: where are the consumers? What do they want? Who else is selling the same product? Which market is surplus and which glut? Which service providers can I best work with?

Data types: localized data, imported data (in addition: localized ‘metadata’ on farmers and their farms can be powerfully aggregated by farmer organizations using joint actions to negotiate better deals for their members).


Exported data streams present all the risks and benefits related to the sharing of one’s own data in public.

It is important to note though that localized data streams, when digital, are always likely to be exported. Depending on how farmers collect and manage and share their data, it can be exported (willingly or not) in various ways.


For farmers, data needs to be transformed into information, and this information used with experience as knowledge. A core process is learning. This is not linear and applied or adapted based on the farmer’s own knowledge base. These learning and adaptation capacities are critical for farmers to bene t from data-driven agriculture opportunities.

  1. Availability: do data exist? can farmers find these data?
  2. Accessibility: can data be opened (protocols, licenses, permissions)? can data be read (formats, encoding)?
  3. Interoperability: can data be exchanged among different platforms and systems (formats, standards, semantics)?
  4. Reusability: can be combined with other data and reused (licensing, metadata, provenance)?
  5. Usefulness: Accuracy, Scale, Timeliness, Trustworthiness, Relevance
  6. Affordability: the costs of access and use of data and services.
  7. Applicability: concerns the relevance of data and services to the specific needs and capabilities of farmers.
  8. Appropriation: applies to the capacities of farmers to ‘appropriate’ or take ownership of data and information in a collective manner.
  9. Effective use: concerns the abilities of farmers to find, understand, interpret and use data and information effectively.

Original Paper Authors:

Ajit Maru, Consultant

Dan Berne, Lagom Ag Initiative

Jeremy De Beer, University of Ottawa & Global Open Data for Agriculture and Nutrition

Peter Ballantyne, Technical Centre for Agricultural and Rural Cooperation Valeria Pesce, Global Forum on Agricultural Research and Innovation

Stephen Kalyesubula, iLabs@Mak Project, Makerere University

Nicolene Fourie, Council for Scienti c and Industrial Research, South Africa

Chris Addison, Technical Centre for Agricultural and Rural Cooperation

Anneliza Collett, Department of Agriculture, Forestry and Fisheries, South Africa

Juanita Chaves, Global Forum on Agricultural Research and Innovation

With thanks to: Mark Holderness (GFAR), Martin Parr (GODAN), André Laperrière (GODAN), Hugo Besemer (Wageningen University)

Share this story, choose your platform!