ILRI Scientist Dr John Recha on Why Climate-Smart Agriculture Programmes Struggle to Change Farmer Behaviour

This is the second article in our series drawing on field research across Western and Mount Kenya. In the first, we explored how farmers learn about climate-smart agriculture. In this piece, we examine why learning does not always translate into sustained behaviour change. Drawing on our field findings and insights from ILRI scientist Dr John Recha (PhD), we explore the role of experiential learning, peer networks, co-created demonstrations, follow-up, and bundled support in helping climate-smart practices take hold.

After we published our findings on how farmers learn about climate-smart agriculture practices, we wanted to go further. To ask a harder question: if farmers are already learning and adapting, why do so many climate-smart agriculture programmes still struggle to shift behaviour?

Dr John Recha (PhD), a Scientist in Climate-Smart Agriculture and Policy at the International Livestock Research Institute (ILRI), shares his insights. He has spent over 15 years researching climate-smart agriculture across Eastern, Central and Southern Africa. 

He observes that the real challenge is whether programmes are designed to turn that learning into sustained practice. 

Climate change is not a future problem for farmers. It is a present one.

Climate change is not a future problem for farmers. It is a present one.

When our team spent time with farmers across Western Kenya and the Mount Kenya region, they were not talking about climate change in abstract terms. They were describing what they saw in their fields, season by season.

In Western Kenya, maize farmers spoke about planting windows that no longer held - rains arriving weeks late or stopping abruptly before crops had matured. In the Mount Kenya region, horticultural farmers described water stress and prolonged dry spells that were rare a generation ago.

Dr Recha situated these experiences within a broader regional pattern observed across Eastern, Central, and Southern Africa. Climate change, he noted, does not affect all farmers in the same way. Its effects vary according to crop system, agroecological zone, water access, and market context. That variation is precisely why programmes transferred unchanged from one context to another routinely underperform.

Taken together, our fieldwork and Dr Recha’s wider research point to the same conclusion: farmers are not passive in the face of these shocks. They are already adapting - switching to drought-tolerant varieties, adjusting planting schedules, intercropping to spread risk, and changing how they manage water.

Yet adaptation is often reactive. Farmers adjust after a failed season, a pest outbreak, or an unexpected dry spell, rather than having the support to anticipate risks and prepare for them.

That distinction has direct implications for programme design. Programmes that reach farmers only after losses occur are already one season behind. The opportunity is to strengthen anticipatory learning and support - helping farmers interpret emerging risks, test appropriate practices, and make informed decisions.

What helps climate-smart practices become sustained behaviour

We asked Dr Recha: based on his experience, what are the most effective ways for farmers to learn and adopt climate-smart agricultural practices? His responses mapped closely onto what we observed on the ground: 

Learning by doing: Dr Recha pointed to experiential learning as the foundation. When farmers try something on their own land and see results, the knowledge sticks. It is retained, adapted, and shared. This is why knowledge that arrives as agricultural extension advice, from a text message, a pamphlet, or a one-off training, often fails to change behaviour. It does not engage the farmer as a practitioner.

Social learning and peer networks: "Farmer-to-farmer exchange is the strongest driver of adoption," Dr Recha said, "because of the trust that already exists." This mirrors what we saw across both regions. Peer learning was the most common and most trusted source of agricultural information - built on shared experience and social proximity. When a neighbour's farm thrives, people notice. When a practice fails, they notice too.

However, social learning has limitations. Our research surfaced this clearly. It is largely passive - observation-driven rather than structured dialogue. When farmers talk, the conversation tends to focus on how much fertiliser or which seed variety they used. 

The deeper system of bundled practices, such as the combination of soil fertility management, timing of planting, seed variety selection, and agronomic practices like crop rotation, that makes climate-smart farming work, rarely transfers completely through peer-learning.

There are also few reliable mechanisms for fact-checking or correcting incomplete advice once it begins circulating through peer networks.

Seed varieties learning event by Dr Recha.

Co-created demonstration plots: This is where Dr Recha was most emphatic. Demo farms are widely used in agricultural programmes, but he drew a sharp distinction between demonstrations done for farmers and demonstrations done with them.

"Without farmer involvement in the designing and monitoring of these demo farms, demonstrations consistently underperform," he said. When farmers are involved from the start - choosing what to test and tracking results, the demonstration becomes their knowledge, not someone else's prescription.

This gap between delivering knowledge and co-creating it with farmers is why many climate-smart agriculture programmes fail to produce lasting behaviour change.

On the role of digital tools: Digital tools are becoming increasingly important to how climate-smart agriculture programmes reach farmers at scale. Dr Recha’s perspective is clear about both their potential and their limits.

“A hybrid model combining digital alerts with local, in-person demonstrations is the most effective approach,” he said.

Digital tools can extend reach and deliver timely information, including weather alerts, planting advisories, pest warnings, and market prices. But many climate-smart practices are not only knowledge-based; they are practical and context-dependent. Farmers often need to see a practice demonstrated, try it in their own conditions, interpret the results, and adjust.

A text notification can prompt action. It rarely builds that practical capability on its own.

This maps closely onto what we found in our research. Farmers are increasingly connected, but digital tools are still used primarily for communication and financial services rather than agricultural decision-making. For many farmers in our study, the channel was available, but the trust, specificity, and contextual relevance needed to influence farming decisions were not yet established.

Digital tools are therefore most effective when connected to trusted human channels, reinforcing what farmers learn from extension officers, peer networks, agronomists, and demonstration farms rather than being treated as a substitute for them.

What organisations are getting wrong when designing climate-smart agriculture programmes for farmers

Drawing on his 15 years of research across Eastern, Central, and Southern Africa, Dr Recha identified the gaps he sees often in how organisations design and deliver climate-smart agriculture programmes: 

Overinvestment in technology, underinvestment in learning systems

"Many programmes lead with inputs, e.g. improved seed varieties, fertiliser and tools - treat these as the solution. They are not. Inputs without a functioning learning system around them produce inconsistent results at best, and can cause harm at worst," Dr Recha said.

Treating farmers as passive recipients

"Farmers are not passive," Dr Recha said. "But the systems around them often treat them that way." Farmers carry indigenous knowledge built over generations. Knowledge of how to read seasons, manage pests, and spread risk. Programmes that ignore this and deliver prescriptions based on assumptions tend to produce surface-level adoption that does not change farmer behaviour.

One size fits all learning approach

Kenya's farming landscape is not uniform. The crop systems, rainfall patterns, and market dynamics of the agroecological zones of Western Kenya differ significantly from those of the Mount Kenya region. Our research made this clear: in Western Kenya, maize dominates two distinct seasons, with the second season highly prone to drought. In Mt. Kenya, farmers manage year-round production of diverse horticultural crops through irrigation, helping them remain resilient. 

A programme designed for one context and applied to the other will fail. 

One-off training without structured follow-up

"Behaviour change requires continuous training and structured follow-ups, not just sessions focused on output," Dr Recha said. This is one of the most common and costly mistakes in agricultural programming. Our research revealed this as well: farmers who had been introduced to calibrated fertiliser application tools through NGO programmes had largely stopped using them - not because the tools were inappropriate, but because the follow-up had not sustained the practice. When programmes exit, adoption often reverses. 

Single practices instead of bundled knowledge

"A single practice, like only using improved seed varieties, is ineffective. What drives results is a bundle: testing the soil, planting on time, using suitable certified seeds, preparing the land for the crop, and proactively managing pests and diseases. These practices are interdependent. Adopting one without the others produces unpredictable results, and farmers who experience disappointing outcomes have little reason to continue," Dr Recha said. 

Ignoring social norms and behavioural drivers

Farmers make decisions based on more than information. They observe what their neighbours are doing, respond to what worked last season and what they trust, assess the benefits, and apply what they can afford.

Programmes that ignore these drivers and measure achievement by the number of farmers trained rather than by actual change in behaviour and practice are using inappropriate indicators. "Monitor the level of adoption and change in farmer behaviour," Dr Recha said, "not just how many farmers attended a training."

What this means for organisations designing climate-smart agriculture programmes

A clear picture emerges from combining our field research with Dr Recha's insights. Farmers are learning and adapting. The knowledge infrastructure around them just needs to be designed to match how they think, decide, and act.

For microfinance institutions, donors, and NGOs working in the agriculture and climate space, a few principles stand out clearly:

Co-create with farmers: Programmes that involve farmers in the design of demonstrations, training content, and tools produce more effective results. Approach farmers’ knowledge as an asset to be integrated.

Bundle climate-smart practices: No single intervention changes farmers’ behaviour sustainably. Combining knowledge, inputs, market linkages, and financial products into coherent bundled packages gives farmers both the information and the right conditions to act on it.

Design for existing learning systems, not just the content: The question is not only what information farmers receive, but in what contexts. Trusted human channels like peer networks, field demonstrations, extension officers, agronomists, and agrovets remain the most effective. Digital tools amplify these channels; they do not replace them.

Measure change in behaviour, not attendance: The metric for success in climate-smart agriculture is not the number of farmers trained. It is how many changed what they do on their farms, and whether those changes held after the programme or training ended.

Do follow-ups: One-off engagements with no follow-up are not programmes, they are experiments. Unfortunately, many organisations have treated farmers as exactly that.

In conclusion: Designing for learning that lasts

The farmers we met across Western Kenya and the Mount Kenya region are already observing changing climate conditions, testing responses, and making difficult decisions with the knowledge and resources available to them.

The opportunity for organisations is to build on farmers' agency.

By designing programmes around how farmers actually learn and decide: through practical experience, trusted relationships, reinforcement, and evidence that a practice can work in their own context.

Our fieldwork and Dr Recha's research reach the same conclusion: the gap between climate-smart agriculture knowledge and sustained practice is as much a programme design problem as an awareness problem. How programmes are designed, delivered, reinforced, and measured determines whether learning becomes lasting behaviour change.

The real measure of success is not how many farmers are reached or trained. It is whether they can test new practices, apply them correctly, and sustain them long after direct programme support ends.

At Spindle Design, we help organisations ground climate-smart agriculture programmes in farmers' realities and translate research into practical programme, product, and delivery decisions. If you are designing or strengthening a climate-smart agriculture initiative and want to better understand how farmers learn, decide, and adopt, reach out to us at hello@spindledesign.co