Scaling approaches and pathways for agricultural innovations and technologies: Emerging lessons from East and Southern Africa Scaling for Impact | 1 Document Description: Technical Report Date: December 2025 CGIAR Acknowledgments This work was undertaken as part of the CGIAR Scaling for Impact (S4I) program. We gratefully acknowledge the support and funding provided by the CGIAR Trust Fund. We extend our sincere appreciation to our partners, stakeholders, and collaborators whose expertise, insights, and commitment have contributed significantly to shaping this work. Their contributions have been instrumental in advancing CGIAR’s ambition to scale proven innovations across food, land, and water systems, fostering impact that is inclusive, sustainable, and transformative. We recognize the continued support and collaboration of national and regional partners, whose engagement ensures that the solutions developed are responsive to local needs, strengthen innovation systems, and contribute to building more resilient agrifood systems. We also acknowledge additional support from the SIFAZ project funded by the European Union in Zambia (Grant No 660 FED/2019/400-893) and the CGIAR Regional Initiative on Diversification in East and Southern Africa. Marcel Gato, AoW 5 co-lead under the CGIAR Scaling for Impact program and Adane Tufa from the International Institute of Tropical Agriculture (IITA) provided useful comments and feedback. We are grateful to the many implementing partners who supported fieldwork in the study countries, which provided the data used for the case studies. To learn more about CGIAR Scaling for Impact (S4I) program, please contact: scaling@cgiar.org Scaling for Impact | 2 Scaling for Impact (S4I) is a CGIAR program (2025–2030) that tests, refines, and scales innovations in food, land, and water systems. It works to align those innovations with stakeholder needs to achieve transformative impact. Website: https://www.cgiar.org/cgiar-research-porfolio-2025-2030/scaling-for-impact/ About CGIAR Scaling for Impact (S4I) program CGIAR is a global research partnership for a food secure future. Visit https://www.cgiar.org/research/cgiar-portfolio to learn more about the initiatives in the CGIAR research portfolio. About CGIAR © 2025 CGIAR System Organization. This publication is licensed for use under a Creative Commons Attribution 4.0 International License (CC BY 4.0). To view this license, visit https://creativecommons.org/licenses/by/4.0. Scaling for Impact | CGIAR 3 Hambulo Ngoma - CIMMYT Zimbabwe Mark Manyanga - CIMMYT Zimbabwe Kelvin Mulungu - CIMMYT Zambia Esau Simutowe - CIMMYT Zambia Eurelia Kazekula - CIMMYT Zambia Mitelo Subakanya - CIMMYT Zambia Blessing Mhlanga - CIMMYT Zimbabwe Timothy J. Krupnik - CGIAR Marenya Paswel - CIMMYT Kenya Christian Thierfelder -CIMMYT Zimbabwe AUTHORS SUGGESTED CITATION Ngoma, H., Manyanga, M., Mulungu, K., Simutowe, E., Kazekula, E., Subakanya, M., Mhlanga, B., Krupnik, T. J., Marenya, P., & Thierfelder, C. (2025). Scaling approaches and pathways for agricultural innovations and technologies: Emerging lessons from East and Southern Africa: Technical report. CGIAR Science Program: Scaling for Impact. CIMMYT. Radia Rowshan, Communication Analyst, CIMMYT DESIGNED BY PHOTOS © CGIAR / CIMMYT Photo: CGIAR https://creativecommons.org/licenses/by/4.0 https://creativecommons.org/licenses/by/4.0 Scaling for Impact | CGIAR 4 Contents Page Acknowledgments 2 Acronyms 5 Abstract 6 1. Introduction 7 2. What scaling approaches work? A brief literature review 10 2.1 Scaling drivers 12 3. Targeting scaling approaches to agricultural innovations: A conceptual framework 13 4. Scaling dimensions and innovation characteristics: Putting the framework into practice 16 5. Case study: A synthesis of scaling approaches from CIMMYT 20 5.1.1 Reach through mother and baby trial approaches 21 5.1.2 Reach through regular demonstrations 21 5.1.3 Reach through small trial seed packs 21 6. Effectiveness and efficiency 23 7. Ranking of scaling approaches 26 8. Conclusion and implications 27 References 29 CGIAR Acronyms Scaling for Impact | 5 AWD Alternate Wetting and Drying AID-I Accelerated Innovation Delivery Initiative CA Conservation Agriculture CATI Computer Assisted Telephone Interviews CGIAR Consultative Group on International Agricultural Research CSA Climate Smart Agriculture CSAIP Climate Smart Agriculture Investment Plan FAO Food and Agriculture Organization FFS Farmer Field School FMNR Farmer Managed Natural Regeneration GRZ Government of Zambia ICT Information Communication and Technology IFPRI International Food Policy Research Institute ISFM Integrated Soil Fertility Management NGO Non-Governmental Organisation PATs Pan African Trials PICS Purdue Improved Crop Storage PPPs Public Private Partnerships SIFAZ Sustainable Intensification of Smallholder Farming Systems in Zambia SIP Sustainable Intensification Practice SRI System of Rice Intensification SSA Sub-Saharan Africa CGIAR Abstract Keywords: Agricultural innovations; Scaling approaches; Scaling pathways; Complexity; Knowledge- intensity Transforming food systems and enhancing rural livelihoods in East and Southern Africa hinges on several necessary systemic changes including improved access to input and output markets, credit, institutional and policy innovations, enabling policy environment, enhanced market access and widespread adoption of agricultural innovations. A critical, yet often overlooked, challenge is selecting the appropriate pathways to effectively scale different types of agricultural innovations. This paper introduces a nuanced conceptual framework for scaling that categorizes agricultural technologies based on two key characteristics: technical complexity and knowledge intensity. This framing makes explicit potent delivery mechanisms for scaling different innovations. We posit that an innovation's position within this framework can inform the most effective scaling approach and dimension, whether it be scaling out (reaching more users and communities), scaling up (influencing policies, markets, and institutional frameworks), or scaling deep (impacting cultural practices and norms). To test this hypothesis, we analyze case studies of conservation agriculture (CA), improved seeds, and mechanization across six African countries (Ethiopia, Kenya, Malawi, Tanzania, Zambia, and Zimbabwe). Our analysis demonstrates that the success of scaling efforts is directly linked to the strategic alignment of the scaling approach or delivery mechanism with the innovation's specific characteristics. Less complex and low-knowledge intensive innovations can be scaled through mass media, SMS, and occasional demonstrations. Complex but low-knowledge intensive innovations may require road shows, videos, and field days. In contrast, complex and knowledge-intensive innovations such as CA and sustainable intensification demand participatory approaches like farmer field schools, mother and baby trial approach, lead farmer model, living labs, and regular extension visits. Knowledge-intensive but less complex practices, such as some forms of agroforestry benefit from structured training, follow-ups, and intensive demonstrations. With respect to scaling dimensions, less complex, low-knowledge-intensive innovations are amenable to rapid scaling out, whereas highly complex and knowledge-intensive technologies demand a focus on scaling deep and scaling up to create a lasting, enabling environment. Small seed packs, roadside demonstrations and mega demonstrations ranked highest in terms of reach, affordability, and market-oriented approaches. Overall, we see a very close alignment between on the ground practices in the selected countries and the predictions by the scaling framework. It is worth noting that the scaling framework developed in this paper will need to be adapted for use in different settings and contexts. The scaling dimensions and pathways are not value chain agnostic as market access and the role of private sector is crucial for sustainability. This paper provides a structured guide for researchers, policymakers, and development practitioners on the design of more targeted and impactful scaling strategies, ultimately accelerating the transition to more resilient and productive agricultural systems. Scaling for Impact | 6 CGIAR 1. Introduction Expanding the reach and use of agricultural innovations is crucial to support the transformation of food systems, improvements in rural livelihoods, and to build resilience. Because smallholder farmers produce over 70% of the food in sub–Saharan Africa (FAO, 2014; Ricciardi et al., 2018), scaling agricultural innovations among this farming segment is crucial. This is important because smallholders are resource-constrained, making the adoption and dissemination of technologies a critical development priority. In this context, scaling encompasses activities and approaches to expand the reach, use, benefits, and impact of agricultural innovations beyond initial trial sites to larger populations in the same or different geographical locations (Hartmann and Linn, 2008). Scaling could also imply mainstreaming of innovations into policy processes. Scaling can take different forms, such as scaling up, scaling out, scaling deep, and scaling down (Woltering and Boa-Alvarado, 2021; Rivera et al., 2025). Scaling up embeds successful innovations into policy processes, institutions, and governance systems, ensuring long-term structural change. Scaling out expands the geographical or horizontal spread to achieve a wider reach and adoption across various locations. Scaling deep focuses on strengthening the quality, sustainability, and cultural relevance of innovations so that they meaningfully transform farmer practices and values. Scaling down highlights the importance of tailoring innovations to specific local contexts. It emphasizes matching and adapting innovations to the needs and capacities of target groups. We distinguish between scaling approaches and scaling pathways. A scaling approach refers to methods or strategies to expand the reach and adoption of an innovation (Woltering and Boa-Alvarado, 2021). This includes different strategies or delivery mechanisms to introduce and popularize agricultural innovations at farm, community and policy level. The aim is to grow, expand, replicate, spread, and transfer successful innovations from pilots to larger contexts. Scaling pathways refer to the actual steps and the sequence taken to make an innovation succeed at scale, in a holistic, context-specific and integrated manner (Kohl, 2023; Woltering and Boa-Alvarado, 2021). It includes building a supportive and enabling environment necessary for sustained and long-term impact at scale. A scaling approach is a method for scaling, while scaling pathways represent parts of an overall roadmap for scaling an innovation in a given context that addresses ancillary success factors key for sustained impact at scale (Woltering et al., 2019; Woltering and Boa-Alvarado, 2021). Scaling pathways explain how actions lead to outcomes in a theory of change. They clarify the sequence linking implementation to results. This supports reflection on whether expected transformations are credible. Common scaling approaches used in East and Southern Africa (ESA) include participatory farmer approaches such as farmer field schools (FFS), farmer learning centers, lead farmer, and mother and baby trial approaches. Other, less participatory or experiential- based learning approaches include digital or ICT-enabled agricultural advisory platforms such as apps, SMS, TV, radio, interactive voice response and WhatsApp chatbots. Small trial seed packs are popular for varietal technologies. Scaling pathways tend to vary depending on the innovation and the scaling ambition. Taking conservation agriculture (CA) as an example, scaling pathways may include market-smart conditional subsidies that crowd in the private sector through public-private partnerships and input supply chains, or policy-driven initiatives deployed via national extension services (Aker et al., 2016; Braun and Duveskog, 2011). Private sector is critical to provide market opportunities for agricultural produce and to invest in downstream value addition and processing, and to participate in input provision under conditional subsidies. Private sector can also provide financing through contract farming and other means. Such interventions create demand for agricultural production which in turn can sustain CA adoption. Other examples include bundled innovation pathways, which combine various innovations in bundles designed to achieve set targets with clear impact pathways. An example is the scaling of mechanization using mechanization hubs in Zambia. These hubs focus on agriculture value chains from land preparation to post-harvest and bundle mechanization services with good agronomic advisories. Among other things, the hubs were designed to contribute to the Zambian government’s targets of increasing maize production to 10 million metric tons, and soybean and groundnuts to 1 million metric tons per year (GRZ, 2024). The selection of which pathways to chart and follow is contingent upon the type of innovation being advanced and the enabling environment in which scaling takes place. Scaling for Impact | 7 CGIAR Scaling approaches and pathways interact strongly. Several scaling approaches and scaling dimensions can be embedded in a given scaling pathway. Figure 1 presents an example of how scaling pathways for agronomy and farm mechanization innovations may overlap with scaling approaches. Several salient elements underpin any scaling pathway: scaling approaches, an enabling policy environment and institutional support, strengthened market access, financial services, and well-functioning value chains (Gebreyes et al., 2021; Hartmann and Linn, 2008). An enabling and institutional support framework could include having the relevant policies or strategies to support targeted value chains (Gebreyes et al., 2021). For example, to increase agricultural production through scaling mechanization, the Zambian government crafted a mechanization strategy to guide mechanization interventions in country (GRZ, 2024). Enabling and institutional support frameworks could also include economic incentives to crowd in private sector investments, access to finance, cross-sectoral coordination, R&D support and quality standards in target value chains (for inputs, produce or machinery), among other measures. Another important element focuses on enhancing market access, access to finance and strengthening value chains to ensure that the innovations being scaled are integrated into markets. It suffices to mention that this delineation is only illustrative as the three blocks interact in complex ways as reflected by the connecting arrows. Scaling for Impact | 8 Photo: CGIAR CGIAR . Scaling for Impact | 9 Agronomy and mechanization scaling pathways Scaling approaches Enabling policy and institutional support Market access and value chain strengthening • Demonstrations • Mother and baby trials • Farmer field schools • Road shows • Lead farmer approach • Radio and television • Capacity building for service providers • Digital tools and decision support systems • Knowledge networks/ communities of practice • Monitoring and feedback loops • Policy framework or strategy • Bundled innovations • Extension services • Strengthened supply chains for spare parts • Access to finance • Rural mechanics • Coordination across actors • Incentives (tax rebates and conditional subsidies) • R&D for mechanization • Standards and safety regulations • Public-private partnerships • Clear land tenure systems • Capacity for local fabrication/maintenance • Monitoring and feedback loops • Enhanced market access for products • Value chain strengthening Market intelligence • Contract farming • Input supply networks • Aggregation and storage • Market information systems • Risk mitigation and management • Service bundling •Joint ventures • Monitoring and feedback loops Figure 1: Relationship between scaling approaches and pathways using agronomy and mechanization as examples CGIAR 1 Despite considerably increased attention to the importance of scaling innovations, numerous challenges persist. These include limited access to credit and inputs, weak extension systems, socio-cultural barriers, poor infrastructure, and a lack of long-term institutional support (Douthwaite et al., 2003; Gebreyes et al., 2021; Makate, 2019). Although numerous innovations generated by agricultural R&D systems demonstrate potential in controlled and managed environments (Woltering and Boa-Alvarado, 2021), their sustained uptake at scale, under smallholder farmer conditions, often remains low (Wigboldus et al., 2016). This synthesis paper develops a conceptual framework to identify scaling approaches feasible for different types of innovations. We provide case studies to assess and compare the reach and effectiveness of various scaling approaches. In assessing reach, we are cognizant of the fact that reach does not always imply behavioral change or sustained adoption, but only under special conditions such as informed household decision-making, access to quality information, minimal constraints related to credit and insurance, limited externalities, and persistent use that signals that benefits outweigh costs (Brooks et al., 2025). While our review is broad, we limit the case studies to three main scaling approaches for which we have data: demonstrations implemented by extension systems and seed companies, small seed packs, and mother-and-baby demonstrations. We assess the extent to which these scaling approaches increased reach for CA, mechanization, and improved seeds across Ethiopia, Kenya, Malawi, Tanzania, Zambia, and Zimbabwe between 2021 and 2025. This study informs future scaling strategies while improving the reach and impact of agricultural transformation. Scaling for Impact | 10 2. What scaling approaches work? A brief literature review Scaling agricultural innovations from pilot projects to widespread adoption is a complex yet crucial process. We synthesize the current literature on the major scaling approaches and pathways used in agricultural development, specifically targeting smallholder farmers. The goal of the review is to highlight the functionality of different scaling approaches and pathways, where they have been effective, and for which types of innovations they have worked or, indeed, not worked as expected. Participatory and horizontal pathways: Demonstration plots, FFSs, and farmer research committees engage local farmers in participatory learning and adaptive experimentation. These approaches often serve as platforms for peer-to-peer learning and enhance credibility through visible results under local conditions (Braun and Duveskog, 2011; De Roo et al., 2019; Wall et al., 2019). For example, the adoption of push-pull pest management technologies in East Africa increased to over 80% among farmers exposed through field days and participatory demonstrations (Gwada, 2021; Khan et al., 2011). The promotion of CA through demonstration plots managed by host farmers in Malawi’s Nkhotakota district significantly enhanced adoption rates among smallholder farmers (Pangapanga-Phiri et al., 2024). Host farmers played a central role in facilitating field days, which helped build farmer confidence and reduced misconceptions about CA practices (Pangapanga-Phiri et al., 2024). This is in line with complementary evidence from a quasi-experimental study in Malawi where farmers who participated in season-long, farmer-led demonstration plots were more likely to adopt more components of integrated soil fertility management (ISFM) compared to those who only attended a one-day field event (Maertens et al., 2021). This suggests that engagement through participatory learning models is more effective in promoting the uptake of knowledge-intensive agricultural technologies. However, these participatory approaches are expensive, raising the need to balance cost efficiency and potential reach. Furthermore, in Malawi and Indonesia, the FFS were relevant and helped farmers to adapt their agricultural practices to changing weather and climate circumstances (Van Den Berg et al., 2020). In the Philippines, using Alternate Wetting and Drying (AWD) for growing rice reduced water use by 21–50% without reducing crop yields, but its adoption relied a lot on policies and good cooperation among institutions (Evangelista et al., 2026). Seed companies like SeedCo benefited from the Pan-African Trials (PATs) through demonstrations of new varieties across a wide array of landscapes which in turn boosted awareness and adoption (Douthwaite et al., 2003). Market-based and commercial pathways: Market-based scaling depends on public-private partnerships (PPP), value chain linkages and economic incentives to disseminate new ideas. For instance, the public and private partnership between CGIAR centers and private seed companies managed to scale over 300 improved agricultural technologies resulting in a 64% to 72% increase in adoption, indicating that PPPs are an effective scaling pathway in East and Southern Africa (World Bank, 2020a). Other examples of market-based scaling pathways with economic incentives include contract farming, community seed banks, and small seed pack distribution. Contract farming appears to perform well in value chains for cash crops such as tea, tobacco, CGIAR and cotton, mainly because they have ready markets that demand certain quality and standards (Minot and Sawyer, 2016). In the same way, out-grower schemes provide farmers with inputs and guaranteed markets to lower transaction costs. Minot and Sawyer, (2016) found that farmers engaged in contract farming increased their income by 25% to 75% in East and Southern Africa. As such, contract farming can be characterized as a purely market-based scaling approach. In addition to contract farming, community seed banks and small seed packs also helped smallholders conserve local varieties and supply seed during climate shocks in Uganda, Ethiopia, and Zimbabwe (Adokorach et al., 2019). Community seed banks operate through revolving seed loan schemes, where farmers receive seeds before the planting season and repay after harvesting, but they also function as market-based institutions by offering seed access on a commercial basis through sales or structured loan arrangements (Tione et al., 2025; Vernooy et al., 2014). This arrangement lowers transaction costs, helps conserve genetic resources, improves access to diverse local crops, and strengthens both seed and food sovereignty. In this sense, community seed banks are market-smart and market-based. Seed packs increased the demand for climate resilient bean grains in Burundi, DRC, Zambia, and Zimbabwe because contracted farmers accessed training, loans and markets which improved their productivity (Rubyogo et al., 2020). Mulungu et al., (Forthcoming) found that seed packs increased farmers technical knowledge of biofortified iron beans, cowpeas and improved traditional African vegetables in Tanzania and Zambia. Authors found that access to seed packs increased the probability of adopting legumes by 10-16 percentage points (pp), 9-13 pp for cereals and 5-9 pp for vegetables in Tanzania. In Zambia, seed packs increased the likelihood of adopting legumes by 9-11 pp, and about 12 pp for drought tolerant maize (DTM). The impact pathways follow the logic that seed packs allow farmers the opportunity to try new crop varieties on a small portion before making the decisions to buy the seeds in large quantities. Alternative business models driven by technology can also transform agriculture in developing countries by connecting smallholders to regional and global value chains, thereby allowing them to access buyers from wider regions. These technology-driven models have proved to reach wider distribution in Kenya, where they were used by Hello Tractor and WeFarm (Aker et al., 2016). Digital and ICT-mediated pathways: The use of digital and information and communication technologies (ICT) pathways has enabled the widespread scaling of various innovations via direct dissemination. ICTs provide avenues for sharing agricultural information, especially in areas where extension services are limited. ICTs such as mobile-based services, SMS alerts, interactive voice response (IVR), and other digital platforms have good potential to raise awareness and increase adoption (Fabregas et al., 2019). For example, in Zambia, Viamo uses its IVR platform to disseminate various agricultural advisories related to agricultural production and weather conditions. This IVR platform is accessible on feature phones without any internet connectivity. In Uganda, Engotoit et al., (2016) found that mobile services and community sensitization improved farmers' adoption of agricultural technologies. Policy-led or vertical scaling pathways: Policy-led scaling pathways use government policies and extension services to spread agricultural practices. A notable example of this policy-led scaling approach is the Pfumvudza program in Zimbabwe. Pfumvudza is a government-led conditional subsidy initiative that promotes minimum tillage, mulching, crop rotation, and the use of organic matter on 1/16 of a hectare to boost crop production. The approach is simple; every farmer who intends to access subsidized inputs should demonstrate that they have implemented the minimum required Pfumvudza plot. Tying input subsidies to CA practices has increased instantaneous participation to more than 75% (Mujere, 2022; Tufa et al., 2023). Over 2.5 million farmers have adopted Pfumvudza in the country, and the World Bank's Climate-Smart Agriculture Investment Plan (CSAIP) identifies Pfumvudza as a scalable climate adaptation concept that can stabilize yield and improve food security (World Bank, 2019). Another example is from Ethiopia, where the (1 to 5) farmer training doubled crop yields and incomes and increased adoption of improved agronomic practices (Spielman et al., 2015). Scaling for Impact | 11 CGIAR Scaling for Impact | 12 2.1 Scaling drivers The scaling of agricultural innovations is driven by a combination of enabling policies, partnerships, and context- sensitive approaches (Rivera et al., 2025). Favorable policy frameworks and financial mechanisms are key to adoption. Capacity building, extension services, and multi-stakeholder platforms facilitate learning and collaboration. For example, the widespread adoption of the System of Rice Intensification (SRI) was facilitated largely through NGO involvement and farmer networks. SRI improves water saving but does not increase yields beyond standard recommended rice management practices when nutrient inputs are kept constant (Krupnik et al., 2012). Similarly, push-pull technologies have been effectively implemented in maize-based systems in East Africa to manage pests and weeds, with its spread largely attributed to field-based, experiential learning approaches (Khan et al., 2011). Climate-Smart Agriculture (CSA) has scaled best in contexts where strong extension systems provide both technical and financial support, with adoption shaped by institutional support, gender dynamics, and farmers’ risk perceptions (Alam et al., 2024; Feder and Anderson, 2004; Ishtiaque et al., 2024). Likewise, agroforestry efforts in Tanzania, particularly under the Vi Agroforestry project, further demonstrate the importance of multilevel collaboration, achieving adoption rates ranging from 10% to 90% across different villages (Johansson et al., 2013). Scaling is further driven by affordable technologies and market incentives that sustain adoption and align innovations with farmers’ needs and local contexts. Access to reliable input and output markets and competitive pricing create incentives for both farmers and private sector actors to invest in innovation. For example, improved market linkages in Zambia have facilitated the uptake of drought-tolerant maize and CA practices (Mulungu et al., Forthcoming). Similarly, digital platforms in Kenya have enhanced market transparency and reduced transaction costs, encouraging wider adoption of improved seed and fertilizer technologies (World Bank, 2020b). These different outcomes emphasize that scaling transcends technological solutions but is embedded in economic dynamics, driven by both supply-side push factors such as R&D, technology subsidies and by demand-side factors such as consumer subsides, and economic incentives for system-level adoption. In Zimbabwe, investments in irrigation infrastructure and market reforms have enabled smallholders to adopt improved maize and legume varieties where markets are reliable and input supply chains are functional (Mhembwe et al., 2019). In Malawi, digital platforms and coordinated value chain interventions enhanced farmer access to inputs and market information, which accelerated the adoption of improved seed systems and the use of Purdue Improved Crop Storage (PICS) bags (IFPRI, 2018). These examples reinforce the importance of market fundamentals in scaling agricultural innovations. The cases reviewed above point to several successful scaling examples. However, there are cases where delivery mechanisms worked less effectively for some innovations. For example, the limited uptake of improved cookstove technologies in the Indian Himalayas illustrates how promising technologies can stall when the chosen delivery mechanisms do not align with local realities. In this case, poorly developed distribution channels, and entrenched cultural preferences for traditional stoves weakened the scaling pathway (Pattanayak et al., 2019). Similarly, Malawi’s Farm Input Subsidy Program initially showed success between 2005 and 2008, but subsequent attempts to scale it further did not translate into proportional gains in productivity due to weak targeting systems, political interference, and insufficient complementary services (Chirwa and Dorward, 2013; Jayne et al., 2018). This is the case with most input support programs in SSA where errors of omission and commission, and elite capture led to only marginal improvements in productivity and non-significant effects on poverty reduction (Jayne et al., 2018). It is noteworthy that different scaling approaches will work differently for specific innovations in given contexts. The challenge, therefore, is to identify scaling pathways and approaches that are feasible and potent for specific innovations in a given context. CGIAR Scaling for Impact | 13 3. Targeting scaling approaches to agricultural innovations: A conceptual framework This section develops a conceptual framework for scaling. The goal is to identify the scaling approaches that are suitable for different innovations within specific scaling pathways and socioeconomic contexts. This is important given that there are multiple scaling approaches with different delivery mechanisms, which necessitate guidance on what options are better suited for specific agricultural technologies and innovations. Such guidance can be gleaned from rigorous impact assessments and cost-benefit analyses to identify returns to different extension approaches. Without such analyses, we develop a conceptual framework to provide guidance. We use a two-by- two matrix measuring complexity on one axis and knowledge intensity of given agricultural innovations on the other axis (Figure 2). This conceptualization is influenced by Jayne et al., (2019) and Ngoma et al., (2021) who apply the induced innovation and the Boserup hypotheses to identify conditions under which given integrated soil fertility management options and CA are likely to be adopted. In the spirit of Rogers, (1962), complexity measures the degree to which a given technology/practice/innovation is difficult to understand and use by farmers. A technology is complex if it involves several other components, such as cropping system configurations like intercropping or rotations, specialized spacing and tillage options, and crop residue management. Knowledge intensity refers to the degree to which a given agricultural technology is cognitively heavy and, therefore, requires farmers to invest time (and resources) to learn it. Knowledge intensity as used here is related to the perceived compatibility and trialability innovation characteristics in Rogers, (1962). This framing results in four quadrants, each showing possible scaling approaches for given innovations based on how complex and knowledge-intensive such innovations are. Complex but not knowledge-intensive innovations, which are embodied, such as improved seeds and fertilizers, are captured by quadrant A. Suitable scaling approaches for such innovations include road shows, television, videos, print media (pamphlets, brochures, booklets), demonstrations, field days, and infrequent extension contacts. Quadrant B is for complex and knowledge-intensive agricultural innovations such as CA, sustainable intensification, artificial insemination, sustainable land management, etc. Such knowledge-intensive and complex innovations require comprehensive and participatory in-person scaling approaches such as farmer field schools, mother and baby trial approaches, lead farmer models, living labs, digital agro-advisories such as interactive voice response (IVR), and regular extension visits. The feasibility and scale of these different scaling approaches could be undermined by associated costs. This expose assumes costs are not a barrier to technology/innovation use. Less complex and less knowledge-intensive agricultural innovations, e.g., for mechanical weeding, basin digging, shelling/threshing and conventional tillage in quadrant C, can rely on radio messages (in local languages), SMS reminders, USSD platforms, and other bulk SMS products. Less complex but knowledge-intensive options such as agroforestry, intensive grazing, and cow fattening are represented in quadrant D, where the most appropriate scaling approaches include training and visit (T&V), regular training and follow-ups, and intensive demonstrations using mega demonstrations, for example. This framework is context specific and not cast in concrete across different scenarios. For example, for a technology we consider to be in quadrant A in ESA, it is possible that in a different setting, this technology could be in quadrant B as farmers may not be familiar with it or the baseline knowledge may be low such that it is characterized as knowledge intensive. It could also be the adaptation of technology that makes it change characteristics. For example, in some places, intercropping configurations that involve several crops and spacings, timed weeding and harvesting may become knowledge intensive and complex while not so, if involving only two crops that grow and mature at the same time. Technology attributes and how they interact with the farmers and local context matters (Adesina and Zinnah, 1993). CGIAR Figure 2: Potential scaling approaches for given agricultural innovations based on complexity and knowledge intensity Notes: CSA – climate smart agriculture; CA – conservation agriculture, SIPs – sustainable intensification practices; ISFM – integrated soil fertility management; SLM – sustainable land management; AI – artificial insemination; IVR - Interactive Voice Response Scaling for Impact | 14 More complex agricultural innovations (B) Complex and knowledge intensive agricultural innovations, e.g., CSA, CA, SIPs, ISFM, SLM, AI, processing Scaling approaches: • Farmer field schools • Mother and baby trial approaches • Innovation systems, living labs • Digital agro-advisories, e.g., IVR, videos • Regular extension visits • Lead farmer model • Mega demonstrations Less complex agricultural innovations M o re k n o w le d g e i n te n s iv e L e s s k n o w le d g e i n te n s iv e (A) Complex and less knowledge intensive agricultural innovations, e.g., e.g., seed variety(ies), mechanized planting, aflasafe Scaling approaches: • Television/mobile vans/road shows • Roadside demonstrations • Print media – pamphlets, brochures • Field days • Infrequent extension visits • Trial seed packs (D) Less complex and high knowledge intensive agricultural innovations, e.g., agroforestry, intensive grazing, cow fattening, farmer managed natural regeneration, mechanized CA tillage, value addition Scaling approaches: • Lead farmer model • Training and visit (T&V) • Regular training and follow up visits • Intensive demonstrations (C) Less complex and less knowledge intensive agricultural innovations, e.g., mechanized weeder, basin digger, shelling/threshing, and conventional tillage Scaling approaches: • Radio campaigns in local languages • SMS reminders, USSD • Once off training meetings/workshops/demonstrations • Road shows/fairs CGIAR Figure 3: Flowchart for understanding what scaling approach to use to achieve widespread adoption of an innovation or technology The conceptual framework proposes that the nature of an agricultural innovation defined by its complexity and knowledge intensity should guide the selection of appropriate scaling approaches. This section aligns the framework's innovation typologies with the strategic dimensions of scaling — scaling out, scaling up, and scaling deep — to connect the nature of a technology with the ultimate purpose of the scaling effort. A more strategic analysis involves aligning innovation types with the overarching dimensions of scaling: scaling out (geographical spread), scaling up (institutional and policy change), and scaling deep (transforming practices, values, and beliefs) (Moore et al., 2015; Omann et al., 2020). This alignment provides a strategic compass, helping practitioners determine the goal of their scaling effort based on the nature of the technology itself. We show a flowchart or tree diagram in Figure 3. Ideally, one would begin by categorizing the technologies into the four quadrants, based on their characteristics and the local context. This is a two-step process, first determining if the innovation is complex, then determining the knowledge intensity and the scaling pathways as indicated. We summarize these more succinctly in Table 1 and propose primary suitability for each quadrant, linking the characteristics of an innovation to the most logical scaling ambition. Scaling for Impact | 15 4.Scaling dimensions and innovation characteristics: Putting the framework into practice CGIAR The strategic alignment in Table 1 demonstrates two further critical considerations for scaling practitioners. First, the three scaling dimensions are not merely distinct options or sequential stages but components of a dynamic, interacting system. While the table suggests a primary suitability for each quadrant, the empirical evidence shows these dimensions are deeply intertwined. Policy action (scaling up) can create an entry point for adoption, but its long-term success and ability to drive genuine geographic spread (scaling out) is contingent on achieving a critical mass of transformative change at the user level (scaling deep) and the perceived legitimacy of government agency implementing the policy and cross sectoral coordination among government agencies (Marenya et al., 2025). The case of the CIMMYT and ILRI-supported crop-livestock integration in Zimbabwe provides a compelling illustration. The project initially promoted forage legumes to improve livestock nutrition, an intervention requiring a change in farmer practice and knowledge (scaling deep). While adoption was steady, it was transformed when a private seed company, Klein Karoo, offered contracts for lablab seed production . This is an example of how institutional market linkages can boost adoption. This is a form of scaling up. This powerful financial incentive dramatically reinforced the value of the innovation, leading to near-complete adoption (scaling out) in some communities. This reveals a causal loop: scaling deep (building farmer knowledge and demonstrating value) created a foundation that was amplified when scaling up (institutional market linkage) provided clear economic incentives, which in turn fueled widespread scaling out (geographic spread). This dynamic model is far more realistic than a simple linear progression from one stage to the next. Second, the primary focus of the scaling effort varies significantly by quadrant. For innovations in Quadrants A and C, the primary effort is external to the farmer. Scaling improved seeds or simple tools involves building supply chains, demand creation through mass media campaigns, and ensuring market functionality. The key actors are institutions—research centers, seed companies, governments, and media houses (CIMMYT, 2014). The farmer, in this model, is largely a recipient or consumer of a finished product or a simple message. In contrast, for innovations in Quadrants B and D, the primary scaling effort is internal to the farmer and their community. The scaling efforts are co-created by farmers. An example is the successful scaling of Farmer Managed Natural Regeneration (FMNR) that began in West Africa and has spread across countries to East Africa. Scaling FMNR or agroforestry is fundamentally about changing knowledge, beliefs, and long-term land management practices. The key actor is the farmer, supported by facilitators who guide a process of discovery and peer-to-peer learning (Lohbeck et al., 2020). External actors are catalysts for a social process, not deliverers of a physical product. This distinction has profound implications for project design, funding, and evaluation. Scaling Quadrant A innovations requires investment in physical and market infrastructure (Figure 2). Scaling Quadrant D innovations require investment in social capital, facilitation, and learning networks. Donors and governments often prefer investment in physical and market infrastructure because the outputs are more tangible and help secure votes (e.g., tons of seed distributed) and the timeline appears faster. However, the latter (investments in social capital and networks), as the FMNR case proves, can be more sustainable, transformative, and ultimately more cost-effective at achieving impact on a large scale. The dimensions of scaling up, scaling out, and scaling deep are not independent strategies but are part of a reinforcing system. The long-term success of any scaling effort depends on achieving a dynamic equilibrium between them (Figure 3). The contrasting case studies of Zimbabwe's Pfumvudza program and Niger's FMNR movement illustrate two different entry points into this systemic loop. Pfumvudza represents a "push" model that begins with scaling up. The government used a powerful policy lever – making access to subsidized inputs conditional on the adoption of CA principles – to drive rapid and widespread geographical adoption (scaling out) among more than 2.5 million farmers. This approach is fast and can achieve impressive reach in the short term. However, its long-term sustainability is contingent on whether this incentivized (forced) adoption leads to genuine understanding and integration of the principles into farming practice (scaling deep). If farmers are merely complying to receive farming inputs, the practice is likely to be abandoned once the subsidy is removed, a common outcome in subsidy-driven programs or those associated with freebies and poorly designed adoption nudges (Ngoma and Mulenga, 2016). Scaling for Impact | 16 CGIAR Scaling for Impact | 17 Complexity/knowledge Matrix Quadrant Innovation characteristics Primary suited scaling dimension Rationale for suitability Illustrative example from literature A: Complex, less knowledge-intensive (e.g., Improved Seeds, Fertilizers) Embodied, discrete, low learning curve for users, but requires a complex support ecosystem (markets, supply chains). Scaling out (primary) and scaling up (essential enabler) The goal is widespread geographical adoption facilitated by the replicable nature of the technologies. However, scaling out is impossible without scaling up — policies and institutional arrangements that ensure functional seed systems and input markets. The Drought Tolerant Maize for Africa Seed Scaling (DTMASS) project aimed for massive scaling out across six countries. Its success was contingent on scaling up efforts to strengthen national seed systems and partnerships with private companies (CIMMYT, 2014). C: Less complex, less knowledge-intensive (e.g., Mechanical Weeder, Basic Crop Rotation) Simple, discrete, low learning curve, low systemic requirements. Often addresses a single, specific problem like labor drudgery. Scaling out These innovations are easily replicable and can spread rapidly once awareness is created and local supply is available. The primary barrier is information and access, not a need for deep behavioral change or major policy shifts. The promotion of the ring hoe weeder across West Africa. Once farmers see its effectiveness, the main challenge becomes local manufacturing and distribution to achieve wide scaling out (Johnson et al., 2019). A similar trend is emerging for mechanized basin diggers in Zimbabwe. B: Complex, knowledge-intensive (e.g., CA/SI, digital mechanization services) Systemic, multi- component, high learning curve, requires significant management skill and often a change in farming philosophy. Scaling deep (foundation)à Scaling out (expansion)à Scaling up (institutionalization) Sustained adoption requires a fundamental change in farmer practice and understanding (scaling deep). Without this, adoption is partial or temporary. Once deeply embedded, it can spread horizontally (scaling out), often requiring policy support (scaling up) like conditional subsidies to overcome initial barriers (e.g., labor, capital). The Pfumvudza program in Zimbabwe used scaling Up (policy) to drive rapid scaling out (Mavesere and Dzawanda, 2023). However, the long-term sustainability depends on achieving scaling deep, as evidenced by the persistent challenges of full CA adoption in Zambia and Malawi (Tufa et al., 2023). D: Less Complex, high knowledge-intensive (e.g., Agroforestry, farmer managed natural restoration (FMNR)) Physically simple but conceptually transformative. Requires a long-term perspective and a shift in mindset regarding land management and ecological principles. Benefits are often delayed. Scaling deep (primary driver) and scaling out (consequence) The core challenge is changing hearts and minds—convincing farmers of the long-term value and ecological principles. This is scaling deep. Once a critical mass of farmers is convinced, the practice spreads virally through farmer-to-farmer networks, leading to massive scaling out. How to keep farmers motivated in the initial stages before benefits begin to accrue remains a barrier to uptake. FMNR in Niger is the archetypal example. The innovation was the idea of the "underground forest." Once this was deeply understood by farmers (scaling deep), it led to a self-propelled movement that re-greened millions of hectares (scaling out) with minimal external policy support (Lohbeck et al., 2020; Reij and Garrity, 2016). https://hdl.handle.net/10568/159862 Table 1: Innovation characteristics and scaling dimensions https://hdl.handle.net/10568/159862 CGIAR Scaling for Impact | 18 The Pfumvudza model is therefore a high-stakes gamble that policy can catalyze a deeper behavioral shift through a “push” model. In stark contrast, FMNR in Niger represents a "pull" model that begins with scaling deep. The innovation was not physical but a paradigm shift, fueled by a deep understanding of the intrinsic value of forests and the importance of regenerating them. The initial scaling effort focused on changing the hearts and minds of a small number of farmers through intensive awareness-raising and peer-to-peer demonstration. Once this concept was deeply understood and its benefits were visible, it spread organically and at an accelerated pace through farmer-led networks, resulting in massive scaling out (the restoration of an estimated 5 million hectares) with minimal top-down policy intervention or external funding. This model is slow to start but proves to be resilient, cost-effective, and sustainable because the motivation for adoption is internal to the community. These cases demonstrate that while the starting point may differ, long-term success requires all three scaling dimensions. Photo: CGIAR CGIAR 5. Case study: a synthesis of scaling approaches from CIMMYT Table 2: Overview of scaling approaches, associated projects and geographical focus across selected agricultural innovation initiatives CIMMYT and partners have applied diverse approaches and pathways to scale agricultural innovations across Africa. The Ukama Ustawi (UU) Initiative (2023–2025) and the R4 Rural Resilience Initiative (2018–2025) used the mother-and-baby trial approach in Ethiopia, Kenya, Malawi, Zambia, and Zimbabwe. Mother trials, managed by researchers and extension agents, act as demonstration plots, while baby trials allow volunteer farmers to test and adapt selected technologies on their own fields, encouraging wider diffusion. In Zambia, the Sustainable Intensification of Smallholder Farming Systems in Zambia (SIFAZ) project applied standard demonstrations, while the Accelerated Innovation Delivery Initiative (AID-I) project scaled innovations through mega demonstrations, trial seed packs and digital advisories in Zambia, Malawi, and Tanzania, engaging large farmer groups (Table 2). AID-I distributed small trial seed packs in Malawi, Zambia and Tanzania, providing farmers with an opportunity to try new, improved seeds. Each approach offers distinct advantages: mega demonstrations maximize visibility but are resource-intensive, small seed packs are cost-effective and market-based, while mother-and-baby trials provide the strongest participatory learning platform but are expensive to implement. Taken together, these approaches show how combining strategies can expand participation, accelerate adoption, and strengthen resilience. All the scaling efforts discussed here focused on scaling out – expanding the geographical reach of various innovations. The next section uses data collected from countries where the above-mentioned projects were implemented to demonstrate how reach varies by scaling approach. Scaling for Impact | 19 Scaling approach Data availability and years/project Geographical focus Standard demonstrations SIFAZ project (2019-2026) Zambia Mega demonstrations (larger in size, with multiple crops and varieties per crop) AID-I (2022-2025) Zambia, Malawi, Tanzania Seed packs AID-I (2022-2025) Zambia, Tanzania Mother and baby trials/demos Ukama Ustawi Initiative (2023- 2025) R4 Rural Resilience Initiative (2018-2025) SIFAZ project (2019-2026) Ethiopia, Kenya, Malawi, Zambia and Zimbabwe Zimbabwe Zambia CGIAR 5.1 Reach Table 3: Number of people reached under the Ukama Ustawi regional initiative The section draws heavily from Ngoma et al. (2024b) who measured the extent of reach and use of agricultural innovations promoted under UU across the five project countries using a robust quantitative methodology combining both primary and secondary data sources. Primary data were collected through structured questionnaires administered via a Computer-Assisted Telephone Interviewing (CATI) platform facilitated by GeoPoll. Secondary data, specifically rural population figures for the study districts, were sourced from national statistics agencies in Kenya, Ethiopia, Malawi, Zambia, and Zimbabwe. Sampling was guided by Cochran’s formula for large populations, targeting a 95% confidence level and a 5% margin of error. Sample sizes were calculated based on the rural population of operational districts and then allocated across 27 districts. The final target sample size was 6,500, with a 99% completion rate (6,445 responses), yielding minimal margins of error across countries. This approach ensured statistically reliable data, outperforming typical mobile and personal interviewing benchmarks. Readers are referred to Ngoma et al. (2024b) for more details. Scaling for Impact | 20 In this section, we present the adoption statistics for innovations implemented under the UU regional initiative across Ethiopia, Kenya, Malawi, Zambia, and Zimbabwe and reported in Ngoma et al. (2024b). The initiative used the mother and baby trial approach to facilitate the adoption of minimum tillage, intercropping, and mechanization technologies, with varying uptake levels by country and practice. While other extension and dissemination approaches may have been employed within the project areas, the beneficiary reach and adoption figures presented here are primarily attributable to the mother-and-baby trial approach, as it was the principal methodology used in the UU project intervention districts. Through various interventions and across in the five target countries, a total of 164,363 people benefited from UU through using various innovations promoted by UU as of September 2024 (Table 3), including 59,457 people through agronomy, 16,743 through mechanization and irrigation, 48,134 through nutrition, and 40,029 through livestock. For detailed statistics on the number of people reached and using different innovations under Ukama Ustawi, refer to the technical report in Ngoma et al., (2024b). 5.1.1. Reach through mother and baby trial approaches Technology Total used Std Error [95% conf. interval] Agronomy 59,457 11,115 37,668 81,246 Mechanization and irrigation 16,743 5,590 5,784 27,702 Nutrition 48,134 10,010 28,512 67,756 Livestock 40,029 9,013 22,360 57,698 Total 164,363 35,728 94,324 234,402 Source: Ukama Ustawi survey data 2024 Ethiopia showed relatively low adoption of minimum tillage methods, with only 1–3% of the population using practices such as dibble sticks, direct seeders, ripping, and planting basins. The adoption of intercropping varied, with 17% practicing 2-row strip cropping, 10% implementing cereal-legume Gliricidia intercropping, and 5% using 4-row strip cropping. Mechanization uptake was limited: 5% used the multicrop threshers, 1% the groundnut sheller, and no recorded use of the roasters. The mother-and-baby trials reached 1,499 participants with mechanization technologies and 1,458 with agronomic practices. The total number of people reached by this approach in Ethiopia was 7,413. CGIAR In Kenya, ripping was adopted by 12% (approximately 610,000 individuals in the study areas), while basins, dibble sticks, and direct seeders each saw 2–3% adoption. Ripping refers to the use of a ripper tine to open narrow furrows in the soil surface for planting while leaving most of the soil surface undisturbed. Direct seeding involves planting seeds into uncultivated soils using a pointed wooden stick (dibble stick) or a jab planter which minimizes soil disturbance by only creating a planting hole where seeds are placed. Similarly, planting basins are small permanent or semi-permanent holes dug into the soil for fertilizer and/or manure application and seeding (Ngwira et al., 2013; Thierfelder et al., 2016). The multi-crop thresher had a significantly higher uptake at 12% (around 571,000 users), with the groundnut sheller and roaster used by only 1%. The project reached 13,279 beneficiaries through agronomy and 5,456 through mechanization. The total number of beneficiaries reached by the approach in Kenya was 37,470. Malawi demonstrated higher adoption of minimum tillage methods, with planting basins adopted by 23% (930,144 people), dibble sticks by 18%, and ripping by 13%; direct seeders were less common at 2%. Intercropping methods had moderate uptake, with 4–5% using 2-row and 4-row strip cropping or cereal-legume Gliricidia intercropping, respectively. Mechanization was moderately adopted, with 14% using the multicrop thresher. Agronomic practices benefitted 24,906 individuals, while mechanization reached 3,112 people. The UU initiative reached a total of 66,014 people in Malawi. In Zambia, 24% of the population (327,439 people) in the study area adopted ripping, and traditional intercropping was used by 21%. Other technologies, such as basins, groundnut shellers, and multi-crop threshers, had adoption rates between 4–5%. The use of dibble sticks, direct seeders, 2-row and 4-row strip cropping, cereal-legume Gliricidia intercropping, and groundnut roasting was minimal (1–2% or less). The UU intervention reached 16,208 beneficiaries, with agronomic practices benefitting 7,289 individuals. The low adoption of mechanization services is mainly attributable to limited machinery availability amid high demand, a challenge that governments and partners, including CGIAR centers, are actively working to mitigate through the establishment of mechanization service providers or machinery rental centers operated by the private sector (Baudron et al., 2019; Ngoma et al., 2023; Tufa et al., 2023). Zimbabwe reported comparatively higher adoption rates for minimum tillage practices: 10% for dibble sticks, 32% for direct seeders, 55% for ripping, and 30% for planting basins. Traditional intercropping was prevalent, with a 40% adoption, alongside 7% for 2-row and 5% for 4-row strip cropping, and 3% for cereal-legume Gliricidia intercropping. Post-harvest technologies showed adoption rates of 14% for groundnut shellers, 6% for multicrop threshers, and 11% for roasters. The UU project engaged 37,258 recipients overall, with 12,525 benefiting from agronomic practices and 6,676 from mechanization. Scaling for Impact | 21 In Malawi, Tanzania, and Zambia, demonstration activities implemented under the CIMMYT’s AID-I project during the 2024/2025 agricultural season benefited several farmers by showcasing sustainable agricultural practices in real farm settings. These options have proven to be highly effective tools for learning and technology adoption among smallholder farmers. According to Subakanya et al., (2024) over 80% of farmers who attended AID-I-supported mega demonstration events reported learning something new, with drought-tolerant maize varieties emerging as the most preferred innovation. Importantly, the study found a strong correlation between farmers’ intent to adopt and actual adoption, with more than 65% of those expressing interest in drought-tolerant maize going on to implement it during the El Niño-affected 2023/2024 season. These demonstrations not only raised awareness but also facilitated behavioral change, especially when bundled with complementary innovations like trial seed packs and advisory services. The findings underscore that mega demonstrations, when strategically designed and contextually embedded, can significantly accelerate the scaling of climate-smart agricultural technologies in Southern Africa (Subakanya et al., 2024). 5.1.2. Reach through regular demonstrations 5.1.3. Reach through small trial seed packs This section summarizes outreach activities conducted through seed distribution across DR Congo, Malawi, Tanzania, and Zambia through CIMMYT’s AID-I project during the 2024/2025 agricultural season. The objective was to showcase improved crop varietal technologies while ensuring broad access through the distribution of small trial seed packs. Overall, 6,911 individuals were reached through seed distribution. CGIAR Scaling for Impact | 22 Small seed pack distribution efforts reached 253 individuals in the Democratic Republic of Congo, including 73 women and 180 men. Through similar outreach in Malawi, 554 individuals were engaged, comprising 237 women and 324 men. In Tanzania, the initiative reached 3,381 individuals, with 1,127 women and 1,654 men participating. Lastly, in Zambia, small trial seed distribution reached 2,723 individuals, of whom 1,487 were women and 1,236 were men. Following the synthesis of evidence above, it is natural to ask, to what extent did the on the ground implementation follow the predictions of the conceptual framework in section 3. The review on reach presented above involves a set of five distinct technologies: climate smart agriculture/sustainable intensification/conservation agriculture, strip cropping, intercropping, improved crop varieties, and mechanization (including irrigation). The scaling approaches used for these innovations or technology clusters in ESA show very close alignment to the predictions of the conceptual framework in Figure 2. Complex and knowledge intensive technologies like climate smart agriculture/sustainable intensification/conservation agriculture were promoted using mother and baby trials through a network on-farm trials, lead farmers, mega demonstrations and digital agricultural advisories as predicted for quadrant B of the framework. Improved crops and vegetable varieties, that are complex, but not so knowledge intensive, were promoted using different media options including pamphlets, television and radio, and field days and roadside demonstrations as predicted for quadrant A in Figure 2. Less knowledge intensive and less complex innovations such as mechanical weeding and basin digging were scaled through once-off training, road shows/fairs or demonstrations as predicted for quadrant C in Figure 2. Overall, we see a very close alignment between on the ground practices in the selected countries and the predictions by the scaling framework in Figure 2. Suffice to mention that this may only be applicable to the context under study and may be completely different in other settings. As such, the framework in Figure 2 can be adapted to different contexts. 6. Effectiveness and efficiency This section synthesizes evidence on the effectiveness and efficiency of different scaling approaches in East and Southern Africa. It starts by providing a high-level overview of how CA adoption by lead or host farmers influences adoption by follower farmers. We then discuss the effects of demonstrations and seed packs, and end with some discussions on the efficiency of demonstrations as a scaling approach. 6.1 A preview of conservation agriculture adoption Conservation agriculture adoption in ESA remains generally low, particularly when considering the full CA package (Ngoma et al., 2021). However, targeted interventions such as demonstrations, sustained provision of extension services, and social learning significantly boost uptake (Pangapanga-Phiri et al., 2024). This is also in line with (Mulungu et al., 2025), who found that high yields on CA demonstration plots significantly boost the likelihood of adoption of minimum tillage and full conservation by 4%. Follower farmers exhibit a 20% higher likelihood of adopting technologies demonstrated by host farmers beyond the demonstration sites (Mulungu et al., 2025). The speed of CA adoption from exposure/training in southern Africa was estimated at approximately four years (Ngoma et al., 2024). Furthermore, Mulungu et al., (2025) found an inverted U-shaped relationship between yield gaps and adoption, suggesting that moderate performance gaps between demonstration plot hosts and follower farmers can drive observational learning, whereas gaps that are too large may seem unattainable, and those that are too small may reduce perceived relevance or credibility. 6.2 Demonstration effects, social learning, and gendered influence Empirical studies confirm that lead/mother farmer adoption is highly predictive for followers. Social learning, neighbor effects, and demonstration plot success are repeatedly found to increase the probability of neighbor/follower adoption compared to conventional “blanket” extension (Takahashi et al., 2020; Umar, 2021). The power of peer influence and visible success on demonstration plots translates into measurable gains, studies report adoption increases of 17–20 percentage points for minimum tillage and 11–12 percentage points for full CA (Mulungu et al., 2025; Kassie et al., 2018; Marenya et al., 2021). These effects are further reinforced CGIAR Scaling for Impact | 23 by community-based extension activities, where relative risk ratios for adopting conservation farming range from 10 to 14, underscoring the effectiveness of localized, socially embedded learning channels (Marenya et al., 2021). Pangapanga et al., (2024) also reported that 57% of farmers in treatment areas with host farmer contact adopted full CA compared to only 7% in control areas in the sentinel site where CA had been promoted for than 15 years in Malawi. Mulungu et al., (2025) show a more modest 11-12 percentage point effect for the full CA package, while Tufa et al., (2023) showed significant gaps between awareness and adoption rates across Malawi, Zambia, and Zimbabwe. Adoption trends in Mozambique and Zimbabwe mirror Zambia’s, where 45– 55% of farmers adopt at least one CA component, but full-package adoption remains rare (Chichongue et al., 2020; Mujeyi et al., 2021). Thus, while individual practices like minimum tillage strongly benefit from peer learning, complex integrated packages face additional adoption barriers that limit demonstration effects. Gender patterns across different studies in Africa on CA reveal variations in adoption trends. Moreover, gender dynamics shape peer influence: male lead farmers often exhibit greater impact on peer adoption due to their typically higher access to resources, decision-making autonomy, and land (Kondylis 2017). This gendered influence is particularly relevant for practices such as rotation and intercropping, where resource constraints can disproportionately affect women’s ability to demonstrate or replicate (Mulungu et al., 2025). Men’s greater access to resources and decision-making autonomy often translates to higher “influence weight” within these farmer-to- farmer networks. 6.3 Seed packs effects Small seed packs act as an effective mechanism for stimulating demand, significantly enhancing farmer knowledge and the adoption of improved seed varieties. In Tanzania, seed packs enhanced farmers’ knowledge of promoted varieties by as much as two pp and led to a 9–13 pp increase in the adoption of cereals (Mulungu et al., Forthcoming). Receiving iron bean seed packs increased farmers knowledge of promoted seed by 2 pp in Zambia, while receiving cowpea seed packs increased it by 1 pp. In Tanzania, farmers were 10–16% more likely to adopt legumes (biofortified iron beans, common beans, groundnuts, and cowpeas) after receiving a seed pack, whereas seed packs enhanced legume uptake by 9–11% in Zambia (Mulungu et al., Forthcoming). The comparatively low cost and ease of distribution render seed packs a viable option for introducing new varieties, especially in resource-constrained smallholder farming systems where risk aversion hinders experimentation with new seeds/technologies. The findings highlight the effectiveness of seed packs in encouraging smallholder farmers to adopt better germplasm, emphasizing their significance as a practical and scalable approach for promoting adoption. In addition, seed packs' availability significantly enhanced farmer awareness and fostered more favorable attitudes toward the adoption of new seed varieties (Asfaw et al., 2012; Wanyama et al., 2023). The targeted seed interventions prove to be an effective strategy for shaping farmer decision-making and promoting the uptake of improved agricultural technologies, although their effectiveness may vary depending on crop type and local conditions (Asfaw et al., 2012). The effectiveness of seed packs is rooted in their ability to enhance knowledge and promote adoption. Another result from the literature highlights that direct access to seed packs significantly accelerates the adoption of improved seed varieties, as demonstrated by Foster and Rosenzweig, (1995). Their findings emphasize that when farmers are provided with tangible inputs, such as seed packs, it reduces uncertainty and encourages experimentation with new technologies. 6.4 The costs of scaling agricultural technologies: A focus on mother host farmers Comprehensive data on the costs of scaling different agricultural technologies remains scarce in the literature, making it difficult to conduct systematic comparisons across methodologies (Sperling et al., 2004). While demonstration plots have been widely documented as extension tools, limited research has focused specifically on their implementation costs. These costs vary extensively and are dependent on what is being demonstrated, costs incurred by the community and the project or organization promoting the technology, and the locality. On the other hand, mega demonstrations appear to require substantial infrastructure investments. The West Africa Seed Alliance implemented a five-year, $6.1 million project across five countries to modernize seed distribution systems, though per-farmer costs were not disaggregated (https://cnfa.org/program/west-africa-seed-alliance/). Recent evidence from Tanzania provides rare quantitative data on small seed pack distribution costs: trial packs (150 packs of 400 g seeds per village) increased implementation costs by 37.5 times compared to demonstration plots alone, requiring an additional 60 kg of seeds and 75 g of treatment chemicals per village, yet showed no statistically significant impact on technology adoption (Maredia et al., 2025). These fragmented cost estimates highlight the urgent need for more systematic documentation of scaling approach expenses across different methodologies. CGIAR Scaling for Impact | 24 The SIFAZ project provides unique insights into the actual costs of mother-baby farmer host approaches through detailed field-level data collected from 105 farmers across nine districts in Zambia between 2020 and 2023. Table 5 presents the comprehensive cost breakdown for different CA practices, revealing total implementation costs ranging from about $600 to $730 per hectare for different CA systems. Notably, community contributions (primarily labor costs) account for approximately 51-53% of total costs across most treatments, with absolute community labor contributions ranging from $317-$374 per hectare annually. The data captures not only input costs (seeds, fertilizers, pesticides) but also detailed labor requirements across the entire agricultural calendar, from land clearing through post-harvest processing, providing a comprehensive picture of the true resource requirements for technology adoption at the farm level. Note: CP = Conventional Practice; MT = Minimum Tillage; PP = Pigeon Pea. Community costs represent labor contributions valued at prevailing market wage rates. Practice Total Costs ($/ha) Community Labor Costs ($/ha) Community Share of costs (%) Labor Activities Conventional ploughing (CP) 610 317 52% Land clearing, preparation, planting, weeding, harvest Minimum tillage (MT) maize-soybean intercropping 733 374 51% Enhanced planting operations, multiple crop management CP maize-soybean intercropping 673 343 51% Dual crop planting, coordinated weeding, split harvesting MT maize-groundnut intercropping 693 354 51% Intercrop management, specialized harvesting MT maize-common beans rotation and PP 689 352 51% Rotation planning, pigeon pea establishment, and maintenance Table 4: Annual costs and community contributions for selected innovation in the SIFAZ mother farmer trials These cost figures from SIFAZ trials provide a valuable benchmark for understanding the resource intensity of farmer-hosted technology demonstrations. The community labor contribution of $317-$374 per hectare annually represents substantial farmer investment, particularly when considering that participating farmers also provided land and bore production risks. The SIFAZ approach appears more cost-efficient on a per-farmer basis, leveraging existing farmer networks and knowledge systems. However, the 51-53% community contribution requirement suggests that the successful scaling of this approach depends heavily on farmers' ability and willingness to provide significant labor inputs, which may limit adoption among resource-constrained households. This finding aligns with broader concerns in the scaling literature about the sustainability of approaches that require substantial farmer contributions yet have no immediate guaranteed returns. CGIAR Scaling for Impact | 25 7. Ranking of scaling approaches We ranked the different scaling approaches used for improved seeds, CA, sustainable intensification and mechanization. The ranking was based on the actual and potential reach, implementation costs, and alignment to market-based approaches of different scaling approaches. These rankings were validated during a stakeholder workshop attended by 57 participants at the SIFAZ annual planning and review meeting held on 23 September 2025 in Lusaka, Zambia. The tool was designed and deployed using the Survey Monkey platform. An additional six agricultural experts completed the survey online, bringing the total sample to 63 respondents. One-third (33%) of respondents held a bachelor’s degree, 32% had a master’s degree, and 13% possessed a PhD. The majority (57%) had more than nine years of experience, and most reported using farmer-centric extension approaches such as the lead farmer model, mother-and-baby demonstrations, and farmer field schools. Over 60% were from the public sector, with smaller proportions from CGIAR centers (17%) and other research or development organizations. Results suggest that small trial seed packs have the highest reach capacity and are the most market-oriented, hence they were ranked first, with the highest scores for reach and market orientation, and second best for cost of implementation (Table 6). Roadside demonstrations followed in second place, performing moderately well across all three criteria. Mega demonstrations ranked third overall, while road shows and fairs ranked fourth, showing moderate performance in reach, implementation costs, and market orientation. In contrast, DAAS were ranked fifth with varied scores across these three criteria. Farmer field schools were ranked sixth, and mother and baby hosts took the seventh spot. Conversely, the lead farmer approach ranked the lowest in eighth place. These results show that in-person and participatory approaches that are critical for complex and knowledge- intensive innovations, as per the framework in Figure 2, are the most expensive and have the lowest reach and are the least market-oriented. This highlights that there will be tradeoffs not only based on the innovation characteristics, but also based on the intended results, cost of implementation and market orientation. While the findings from the ranking exercise underscore prioritizing small trial seed packs as the primary scaling approach, this will vary depending on the innovation type. Small trial seed packs have the highest actual or potential reach capacity, the second lowest cost of implementation, and strong market orientation, offering the most efficient and scalable way to accelerate adoption of crop varietal innovations as per quadrant A in Figure 2. Rank 1 Rank 2 Rank 3 Scaling approach Actual or potential reach (1 = highest) Least cost of implementation (1 = least expensive) Market- based (1 = most market- based) Overall rank Small trial seed packs (SP) 1 2 1 1 Roadside demonstrations (RD) 4 1 3 2 Mega demonstrations (DE) 5 4 2 3 Road shows/fairs (RS) 3 4 4 4 Digital agricultural advisories (DAA) 2 6 5 5 Farmer field school (FFS) 7 3 7 6 Mother and baby hosts (MBH) 6 7 6 7 Lead farmer approach (LF) 8 8 8 8 Table 5: Ranking of scaling approaches CGIAR Scaling for Impact | 26 8. Conclusion and implications Expanding the reach and use of agricultural innovations is crucial to transform food systems, improve rural livelihoods, and support resilience building. How to do so is often not clear. Scaling is complex and costly. We developed a conceptual framework to identify scaling approaches feasible for different types of innovations based on complexity and knowledge-intensity. We then used case studies to assess and compare the reach and effectiveness of various scaling approaches for CA, mechanization and improved seeds in Ethiopia, Kenya, Malawi, Tanzania, Zambia and Zimbabwe and compare the extent to which these different approaches align to our conceptual framework. Lastly, we then ranked the scaling approaches based on reach, cost of implementation and market alignment. There are five main findings: First, feasible scaling approaches for complex but not knowledge-intensive innovations (e.g., where the innovation is embodied, such as improved seeds and fertilizers) include road shows, television, videos, print media (pamphlets, brochures, booklets), demonstrations, field days, and infrequent extension contacts. Second, complex and knowledge-intensive agricultural innovations such as CA, sustainable intensification, artificial insemination, sustainable land management etc. may require comprehensive and participatory in-person scaling approaches such as farmer field schools, mother and baby trial approaches, lead farmer models, living labs, digital agricultural advisories such as interactive voice response (IVR), and regular extension visits. The cost implications of these in person approaches make them infeasible for scaling. Third, less complex and less knowledge-intensive agricultural innovations can use radio messages (in local languages), SMS reminders, USSD platforms, and other bulk SMS products. Fourth, less complex but knowledge-intensive options such as agroforestry, farmer-managed natural regeneration, intensive grazing, and cow fattening would use training and visit, regular training and follow-ups, and intensive demonstrations using mega demonstrations, for example. Lastly, an ordinal ranking based on reach, cost of implementation, and alignment to market-based strategies suggests that small seed packs, roadside demonstrations and mega demonstrations as the top three options for crop varieties, sustainable intensification and mechanization innovations. The scaling approaches used for the innovations studied show very close alignment to the predictions of the conceptual framework. For example, case studies from East and Southern Africa show that complex and knowledge intensive technologies like climate smart agriculture/sustainable intensification/conservation agriculture were promoted using mother and baby trials through a network on farm trials, lead farmers, mega demonstrations and digital agricultural advisories as predicted for quadrant B of the framework. Improved crops and vegetable varieties, that are complex, but not so knowledge intensive, were promoted using different media options including pamphlets, television and radio, and field days and roadside demonstrations as predicted for quadrant A in Figure 2. Less knowledge intensive and less complex innovations such as mechanical weeding and basin digging were scaled through once-off training or demonstrations as predicted for quadrant C in Figure 2. Overall, we see a very close alignment between on the ground practices in the selected countries and the predictions by the scaling framework in Figure 2. Suffice to mention that the applicability of this framework may be completely different in other settings and require adaptation. In relation to scaling dimensions, innovations characterized by low complexity and embodied knowledge, such as improved seed varieties, are prime candidates for scaling out. Success in this dimension can be achieved through widespread, low-touch methods like mass media campaigns and strengthening agro-dealer networks, which focus on broad dissemination. In stark contrast, highly complex and knowledge-intensive innovations, like CA, demand a multi-faceted approach. Their successful adoption is contingent on scaling deep—changing farmer mindsets, practices, and beliefs through intensive, participatory methods like farmer field schools and innovation platforms. Simultaneously, achieving scale requires scaling up—actively working to embed the innovation within national policies, institutional standards, and market structures to create a supportive ecosystem. Misaligning the approach with the technology type leads to wasted resources and, ultimately, failed adoption. CGIAR Scaling for Impact |27 In summary, the choice of scaling approaches for given scaling pathways should be guided by the innovation's complexity and knowledge intensity while considering the potential reach and impact, cost of implementation, and market alignment. The choices will be context-specific, emphasizing that there are no one-size-fits-all- scaling approaches. A scaling approach feasible in one area will not be feasible everywhere. Looking forward, this framework offers a valuable diagnostic and strategic planning tool. It compels governments, NGOs, and research institutions to first characterize an innovation before designing its scaling strategy. However, its use will require refinement, and adaptation to different settings and contexts. Future research should aim to further refine this model by integrating socio-economic factors, market dynamics, and the political economy of policy change. Additionally, exploring how digital tools can create hybrid scaling models—perhaps using targeted videos to scale out complex knowledge or online platforms to facilitate scaling up—could unlock new efficiencies. By respecting the unique character of each innovation, we can build more effective bridges between research and the millions of farmers who stand to benefit. Photo: CGIAR CGIAR Scaling for Impact | 28 References 1. Adesina, A.A., Zinnah, M.M., 1993. Technology characteristics, farmers’ perceptions and adoption decisions: A Tobit model application in Sierra Leone. Agric. Econ. 9, 297–311. https://doi.org/10.1016/0169-5150(93)90019-9 2. Adokorach, J., Vernooy, R., Kakeeto, R., 2019. Scaling Community Seedbanks and Farmer Seed Enterprises in East and Southern Africa: Workshop Highlights, 2-4 October 2019, Entebbe, Uganda. The Alliance of Bioversity International and CIAT, Rome, Italy. 3. Aker, J.C., Ghosh, I., Burrell, J., 2016. The promise (and pitfalls) of ICT for agriculture initiatives. Agric. Econ. 47, 35–48. https://doi.org/10.1111/agec.12301 4. Alam, M.J., Sarma, P.K., Begum, I.A., Connor, J., Crase, L., Sayem, S.M., McKenzie, A.M., 2024. Agricultural extension service, technology adoption, and production risk nexus: Evidence from Bangladesh. Heliyon 10, e34226. https://doi.org/10.1016/j.heliyon.2024.e34226 5. Asfaw, S., Kassie, M., Simtowe, F., Lipper, L., 2012. Poverty Reduction Effects of Agricultural Technology Adoption: A Micro-evidence from Rural Tanzania. J. Dev. Stud. 48, 1288–1305. https://doi.org/10.1080/00220388.2012.671475 6. Baudron, F., Misiko, M., Getnet, B., Nazare, R., Sariah, J., Kaumbutho, P., 2019. A farm-level assessment of labor and mechanization in Eastern and Southern Africa. Agron. Sustain. Dev. 39, 17. https://doi.org/10.1007/s13593-019-0563-5 7. Braun, A., Duveskog, D., 2011. The Farmer Field School Approach – History, Global Assessment and Success Stories (Background Paper for the IFAD Rural Poverty). IFAD. 8. Brooks, M., Lybbert, J.T., Walker, M., 2025. Does An Innovation’s Reach Reveal Anything About Its Impact? Under The Right Conditions: Possibly (Technical Note), N.12. Rome. SPIA. 9. Chichongue, O., Pelser, A., Tol, J.V., Du Preez, C., Ceronio, G., 2020. Factors influencing the adoption of conservation agriculture practices among smallholder farmers in Mozambique. Int. J. Agric. Ext. 7, 277– 290. https://doi.org/10.33687/ijae.007.03.3049 10. Chirwa, E., Dorward, A., 2013. Agricultural Input Subsidies: The Recent Malawi Experience, 1st ed. Oxford University PressOxford. https://doi.org/10.1093/acprof:oso/9780199683529.001.0001 11. CIMMYT, 2014. Drought Tolerant Maize for Africa Seed Scaling (DTMASS). CIMMYT. URL https://www.cimmyt.org/projects/drought-tolerant-maize-for-africa-seed-scaling-dtmass/ (accessed 9.10.25). 12. De Roo, N., Andersson, J.A., Krupnik, T.J., 2019. ON-FARM TRIALS FOR DEVELOPMENT IMPACT? THE ORGANISATION OF RESEARCH AND THE SCALING OF AGRICULTURAL TECHNOLOGIES. Exp. Agric. 55, 163–184. https://doi.org/10.1017/S0014479717000382 13. Douthwaite, B., Kuby, T., Van De Fliert, E., Schulz, S., 2003. Impact pathway evaluation: an approach for achieving and attributing impact in complex systems. Agric. Syst. 78, 243–265. https://doi.org/10.1016/S0308-521X(03)00128-8 14. Engotoit, B., Moya, M.B., Mayoka, K.G., Bonface, A., 2016. A Mobile-Based Communication Adoption Model for agricultural market information dissemination in Uganda. Technol. Vol 5. 15. Evangelista, G.K., Samoy-Pascual, K., Cabangon, R.J., Regalado, M.J., Enriquez, Y., Lampayan, R., Rala, A., Yadav, S., 2026. Why AWD isn’t taking off: Understanding barriers and pathways for scaling in gravity-fed irrigation systems in rice landscape. Agric. Syst. 231, 104491. https://doi.org/10.1016/j.agsy.2025.104491 16. Fabregas, R., Kremer, M., Schilbach, F., 2019. Realizing the potential of digital development: The case of agricultural advice. Science 366, eaay3038. https://doi.org/10.1126/science.aay3038 17. FAO, 2014. The State of Food and Agriculture 2014: Innovation in Family Farming Food and Agriculture Organization of the United Nations. Rome, Italy. CGIAR Scaling for Impact | 29 18. Feder, G., Anderson, J., 2004. Agricultural Extension: Good Intentions and Hard Realities. World Bank Res. Obs. 19, 41–60. https://doi.org/10.1093/wbro/lkh013 19. Foster, A.D., Rosenzweig, M.R., 1995. Learning by Doing and Learning from Others: Human Capital and Technical Change in Agriculture. J. Polit. Econ. https://doi.org/10.1086/601447 20. Gebreyes, M., Mekonnen, K., Thorne, P., Derseh, M., Adie, A., Mulema, A., Kemal, S.A., Tamene, L., Amede, T., Haileslassie, A., Gebrekirstos, A., Mupangwa, W.T., Ebrahim, M., Alene, T., Asfaw, A., Dubale, W., Yasabu, S., 2021. Overcoming constraints of scaling: Critical and empirical perspectives on agricultural innovation scaling. PLOS ONE 16, e0251958. https://doi.org/10.1371/journal.pone.0251958 21. GRZ, 2024. National Agricultural Mechanisation Strategy 2024 - 2028. Government of the Republic of Zambia, Ministry of Agriculture, Lusaka, Zambia. 22. Gwada, R.O., 2021. EFFECT OF INTEGRATED PEST MANAGEMENT TECHNOLOGY ON THE LIVELIHOODS OF SMALL-SCALE MAIZE PRODUCERS. Rev. Agric. Appl. Econ. 24, 37–55. https://doi.org/10.15414/raae.2021.24.01.37-55 23. Hartmann, A., Linn, J.F., 2008. Scaling Up: A Framework and Lessons for Development Effectiveness from Literature and Practice. SSRN Electron. J. https://doi.org/10.2139/ssrn.1301625 24. IFPRI, 2018. Scaling up agriculture technologies in Malawi: A study of approaches used by agriculture research and development organizations [WWW Document]. URL https://massp.ifpri.info/2018/10/09/scaling-up-agriculture-technologies-in-malawi-a-study-of-approaches- used-by-agriculture-research-and-development-organizations/ (accessed 9.4.25). 25. Ishtiaque, A., Krupnik, T.J., Krishna, V., Uddin, Md.N., Aryal, J.P., Srivastava, A.K., Kumar, S., Shahzad, M.F., Bhatt, R., Gardezi, M., Bahinipati, C.S., Nazu, S.B., Ghimire, R., Anik, A.R., Sapkota, T.B., Ghosh, M., Subedi, R., Sardar, A., Uddin, K.M.Z., Khatri-Chhetri, A., Rahman, Md.S., Balwinder-Singh, Jain, M., 2024. Overcoming barriers to climate-smart agriculture in South Asia. Nat. Clim. Change 14, 111–113. https://doi.org/10.1038/s41558-023-01905-z 26. Jayne, T.S., Mason, N.M., Burke, W.J., Ariga, J., 2018. Review: Taking stock of Africa’s second- generation agricultural input subsidy programs. Food Policy 75, 1–14. https://doi.org/10.1016/j.foodpol.2018.01.003 27. Jayne, T.S., Snapp, S., Place, F., Sitko, N., 2019. Sustainable agricultural intensification in an era of rural transformation in Africa. Glob. Food Secur. 20, 105–113. https://doi.org/10.1016/j.gfs.2019.01.008 28. Johansson, K.-E., Axelsson, R., Kimanzu, N., Sassi, S., Bwana, E., Otsyina, R., 2013. The Pattern and Process of Adoption and Scaling up: Variation in Project Outcome Reveals the Importance of Multilevel Collaboration in Agroforestry Development. Sustainability 5, 5195–5224. https://doi.org/10.3390/su5125195 29. Johnson, J.-M., Rodenburg, J., Tanaka, A., Senthilkumar, K., Ahouanton, K., Dieng, I., Klotoe, A., Akakpo, C., Segda, Z., Yameogo, L.P., Gbakatchetche, H., Acheampong, G.K., Bam, R.K., Bakare, O.S., Kalisa, A., Gasore, E.R., Ani, S., Ablede, K., Saito, K., 2019. FARMERS’ PERCEPTIONS ON MECHANICAL WEEDERS FOR RICE PRODUCTION IN SUB-SAHARAN AFRICA. Exp. Agric. 55, 117– 131. https://doi.org/10.1017/S001447971700059X 30. Khan, Z., Midega, C., Pittchar, J., Pickett, J., Bruce, T., 2011. Push-pull technology: a conservation agriculture approach for integrated management of insect pests, weeds and soil health in Africa UK government’s Foresight Food and Farming Futures project. Int. J. Agric. Sustain. 9, 162–170. https://doi.org/10.3763/ijas.2010.0558 31. Kohl, R.D., 2023. Key factors for advancing innovations to scale: Evidence from multiple country case studies of agricultural innovations. Front. Sustain. Food Syst. 7, 1053152. https://doi.org/10.3389/fsufs.2023.1053152 32. Kondylis, F., Mueller, V., Zhu, J., 2017. Seeing is believing? Evidence from an extension network experiment. J. Dev. Econ. 125, 1–20. https://doi.org/10.1016/j.jdeveco.2016.10.004 CGIAR Scaling for Impact | 30 33. Krupnik, T.J., Shennan, C., Rodenburg, J., 2012. Yield, water productivity and nutrient balances under the System of Rice Intensification and Recommended Management Practices in the Sahel. Field Crops Res. 130, 155–167. https://doi.org/10.1016/j.fcr.2012.02.003 34. Lohbeck, M., Albers, P., Boels, L.E., Bongers, F., Morel, S., Sinclair, F., Takoutsing, B., Vågen, T.-G., Winowiecki, L.A., Smith-Dumont, E., 2020. Drivers of farmer-managed natural regeneration in the Sahel. Lessons for restoration. Sci. Rep. 10, 15038. https://doi.org/10.1038/s41598-020-70746-z 35. Maertens, A., Michelson, H., Nourani, V., 2021. How Do Farmers Learn from Extension Services? Evidence from Malawi. Am. J. Agric. Econ. 103, 569–595. https://doi.org/10.1111/ajae.12135 36. Makate, C., 2019. Effective scaling of climate smart agriculture innovations in African smallholder agriculture: A review of approaches, policy and institutional strategy needs. Environ. Sci. Policy 96, 37– 51. https://doi.org/10.1016/j.envsci.2019.01.014 37. Maredia, M.K., Farris, J., Mason, N.M., Morgan, S., 2025. Does Providing Free Trial Packs With Demonstration Plots Increase the Adoption of Agricultural Technologies? Cost Effectiveness Evidence From Tanzania. Agric. Econ. 00, 1–16. https://doi.org/10.1111/agec.70033 38. Marenya, P., Chiduwa, M., Ngoma, H., Chipindu, L., Nohayi, N., Chimonyo, V., 2025. A one government approach for agroecology policy coordination in Zimbabwe. Agroecol. Sustain. Food Syst. 1–29. https://doi.org/10.1080/21683565.2025.2541375 39. Mavesere, F., Dzawanda, B., 2023. Effectiveness of Pfumvudza as a resilient strategy against drought impacts in rural communities of Zimbabwe. GeoJournal 88, 3455–3470. https://doi.org/10.1007/s10708- 022-10812-3 40. Mhembwe, S., Chiunya, N., Dube, E., 2019. The contribution of small-scale rural irrigation schemes towards food security of smallholder farmers in Zimbabwe. Jàmbá J. Disaster Risk Stud. 11, 11. 41. Minot, N., Sawyer, B., 2016. Contract farming in developing countries: Theory, practice, and policy implications. International Food Policy Research Institute, Washington, DC. https://doi.org/10.2499/9780896292130_04 42. Moore, M.-L., Riddell, D., Vocisano, D., 2015. Scaling Out, Scaling Up, Scaling Deep *: Strategies of Non-profits in Advancing Systemic Social Innovation, in: Large Systems Change: An Emerging Field of Transformation and Transitions. Routledge. 43. Mujere, 2022. An Assessment of the Contribution of The Pfumvudza Concept Towards Climate Smart Agriculture in Zimbabwe: A Review. J. Agric. Hortic. Res. 5. https://doi.org/10.33140/JAHR.05.02.01 44. Mujeyi, A., Mudhara, M., Mutenje, M., 2021. The impact of climate smart agriculture on household welfare in smallholder integrated crop–livestock farming systems: evidence from Zimbabwe. Agric. Food Secur. 10, 4. https://doi.org/10.1186/s40066-020-00277-3 45. Mulungu, K., Pangapanga‐Phiri, I., Ngoma, H., 2025. Impacts of Fall Armyworm, Groundnut Rosette, and Soybean Rust Diseases on Smallholder Welfare and the Effectiveness of Control Strategies. Food Energy Secur. 14, e70078. https://doi.org/10.1002/fes3.70078 46. Mulungu, K., Subakanya, M., Peter, S., Essegbemon, A., Chivasa, W., Gethi, J., Chiduwa, M., Sigalla, J., Mvugi, H.-H., Ngoma, H., Forthcoming. Small Seed Packs, Big Potential? Effect of Seed Packs on Knowledge and Adoption of Improved Crop Varieties. 47. Ngoma, H., Angelsen, A., Jayne, T.S., Chapoto, A., 2021. Understanding Adoption and Impacts of Conservation Agriculture in Eastern and Southern Africa: A Review. Front. Agron. 3, 671690. https://doi.org/10.3389/fagro.2021.671690 48. Ngoma, H., Marenya, P., Tufa, A., Alene, A., Chipindu, L., Matin, M.A., Thierfelder, C., Chikoye, D., 2023. Smallholder farmers’ willingness to pay for two-wheel tractor-based mechanisation services in Zambia and Zimbabwe. J. Int. Dev. 35, 2107–2128. https://doi.org/10.1002/jid.3767 CGIAR Scaling for Impact | 31 49. Ngoma, H., Marenya, P., Tufa, A., Alene, A., Matin, M.A., Thierfelder, C., Chikoye, D., 2024a. Too fast or too slow: The speed and persistence of adoption of conservation agriculture in southern Africa. Technol. Forecast. Soc. Change 208, 123689. https://doi.org/10.1016/j.techfore.2024.123689 50. Ngoma, H., Mulungu, K., Manyanga, M., Mhlanga, B., Hanke-Louw, N., Waswa, B., Mata, J.I., Givertz, E., hierfelder, C., 2024b. Many Partners, Big Numbers? Estimating the extent of reach and use of innovations under the CGIAR Research Initiative – Diversification in East and Southern Africa. CIMMYT/CGIAR, Harare, Zimbabwe. 51. Ngoma, H., Mulenga, B.P., 2016. Minimum tillage uptake and uptake intensity by smallholder farmers in Zambia 11, 14. 52. Ngwira, A.R., Thierfelder, C., Lambert, D.M., 2013. Conservation agriculture systems for Malawian smallholder farmers: long-term effects on crop productivity, profitability and soil quality. Renew. Agric. Food Syst. 28, 350–363. https://doi.org/10.1017/S1742170512000257 53. Omann, I., Kammerlander, M., Jäger, J., Bisaro, A., Tàbara, J.D., 2020. Assessing opportunities for scaling out, up and deep of win-win solutions for a sustainable world. Clim. Change 160, 753–767. https://doi.org/10.1007/s10584-019-02503-9 54. Pangapanga-Phiri, I., Ngoma, H., Thierfelder, C., 2024. Understanding sustained adoption of conservation agriculture among smallholder farmers: insights from a sentinel site in Malawi. Renew. Agric. Food Syst. 39, e10. https://doi.org/10.1017/S1742170524000061 55. Pattanayak, S.K., Jeuland, M., Lewis, J.J., Usmani, F., Brooks, N., Bhojvaid, V., Kar, A., Lipinski, L., Morrison, L., Patange, O., Ramanathan, N., Rehman, I.H., Thadani, R., Vora, M., Ramanathan, V., 2019. Experimental evidence on promotion of electric and improved biomass cookstoves. Proc. Natl. Acad. Sci. 116, 13282–13287. https://doi.org/10.1073/pnas.1808827116 56. Reij, C., Garrity, D., 2016. Scaling up farmer-managed natural regeneration in Africa to restore degraded landscapes. Biotropica 48, 834–843. https://doi.org/10.1111/btp.12390 57. Ricciardi, V., Ramankutty, N., Mehrabi, Z., Jarvis, L., Chookolingo, B., 2018. How much of the world’s food do smallholders produce? Glob. Food Secur. 17, 64–72. https://doi.org/10.1016/j.gfs.2018.05.002 58. Rivera, M., Fieldsend, A., Muñoz-Rojas, J., Martin, S., Van Dijk, L., 2025a. Drivers of success when scaling innovations: insights from European agricultural and forestry co-innovation processes. Agron. Sustain. Dev. 45, 32. https://doi.org/10.1007/s13593-025-01025-2 59. Rivera, M., Fieldsend, A., Muñoz-Rojas, J., Martin, S., Van Dijk, L., 2025b. Drivers of success when scaling innovations: insights from European agricultural and forestry co-innovation processes. Agron. Sustain. Dev. 45, 32. https://doi.org/10.1007/s13593-025-01025-2 60. Rogers, E.M., 1962. Diffusion of innovations, 1. ed. ed. Free Press. A Division of Macmillan Publishing Co., Inc., New York, NY. 61. Rubyogo, J.C., Aseete, P., Kalemera, S., 2020. Characterization of Bean Seed Producers, Seed Production and Marketing in Eastern and Southern Africa. CGIAR. 62. Sperling, L., Remington, T., Haugen, J., Nagoda, S., 2004. Addressing seed security in disaster response: linking relief with development, overview. 63. Spielman, D., Mekonnen, D., Alemu, D., Headquarters, I., 2015. Seed, Fertilizer, and Agricultural Extension in Ethiopia. 64. Subakanya, M., Mulungu, K., Ngoma, H., 2024. Mega demonstrations as a tool for experiential learning: Insights from Malawi and Zambia. 65. Takahashi, K., Muraoka, R., Otsuka, K., 2020. Technology adoption, impact, and extension in developing countries’ agriculture: A review of the recent literature. Agric. Econ. 51, 31–45. https://doi.org/10.1111/agec.12539 66. Thierfelder, C., Matemba-Mutasa, R., Bunderson, W.T., M