What do smallholders hire Fall Armyworm control to do? A jobs-based, segmented design brief from maize systems in Siaya, Kenya

Abstract Background : Fall Armyworm (FAW), Spodoptera frugiperda , threatens food security and maize-based livelihoods across sub-Saharan Africa. Adoption research treats control technologies as fixed options whose uptake must be explained. This study asks a prior design question: what job do farmers hire FAW control to do, is that job uniform across farmers, and what must a solution deliver to be retained? Methods : We conducted an abductive thematic analysis of semi-structured interviews with 88 purposively selected smallholder maize farmers in six wards of Siaya County, Kenya. Jobs-to-Be-Done (JTBD) and the Value Proposition Canvas (VPC) were applied retrospectively, because the interviews were not designed as JTBD hiring-timeline interviews. Two researchers coded all records independently against deposited rules (agreement 97.3%; kappa 0.94). Pains and gains were compared across segments defined a priori by market orientation and land constraint, using 95% confidence intervals and Fisher’s exact tests, and read as exploratory. Results : Farmers’ accounts indicated that control was hired to protect the food, income, and stored-grain (buffer) security functions of maize, to reduce anxiety and restore confidence, and to make the control decision defensible through trusted advice and proof observed on farms like the farmer’s own. Interpreted through JTBD, movement among controls is consistent with a stable job pursued through imperfect means. The most prominent pains were uncertain or failed control (98%), cost (89%), risk to human health (75%), and crop damage by the product (75%); the most prominent gains were proof before commitment (100%), value for money (72%), and trusted advice (68%). Income-oriented growers (maize grown wholly or partly for income) more often raised trusted advice, value for money, and ease of use than consumption-only growers, who more often raised restored confidence. Farmers cultivating under one acre more often sought fast, visible control. Only two contrasts survived correction for multiple comparisons, and both segments were confounded with ward, advisory contact, and response length. Eight pains and eight gains were translated into design hypotheses for co-design and testing. Conclusions : Heterogeneous control practices can be read, through a jobs lens, as coherent farmer evaluation, and this evaluation appears to vary with farmer circumstance, although these exploratory patterns need confirmation in larger, multi-site studies. This jobs-first account complements determinant, innovation-systems, and participatory approaches by locating part of the adoption problem in product-service fit, and offers a testable basis for co-designing biopesticides and integrated pest-management packages.