On-Farm Experimentation (OFE) begins with connecting people and sharing ideas. The question to be asked by OFE emerges from and should be defined by a shared un- derstanding of the problem or issue at hand, accounting for what is published in the knowledge base and any experience garnered either formally or informally. Progress is generally quicker by working together, enabling comparisons across environments and conditions, sharing data on metrics that matter, converging around questions to test and co-ordinate experiments across multiple farms. We are developing a platform (Farm-PEP: www.farmpep.net) to support farmers, advisors, industry and researchers in a ‘Learning by Sharing’ approach, providing a space to find, reconcile and share, ideas and projects; tools to enable interactive and dynamic benchmarking of metrics across fields and farms; tools putting the robust design and analysis of on-farm trials into the hands of farmers and practitioners; and a place to report findings and distil messages to inform the knowledge base and resolve actionable guidance for farmers. The Problems & Opportunities Variation in performance on-farm is huge, at all scales and for all metrics The cause of Amost of Athis variability Ais unknown and under-explored. Some farms are A achieving much more than others but we often do not know Awhy. For Athe industry Aas a A whole to improve, we must share and distil knowledge to inform best practice. A Whether we are interested in profitability, efficiency, productivity, sustainability, GHG-intensity, biodiversity, resource use, natural capital, soil health or carbon ba- lance, the variability that exists within and between fields and farms is enormous. Yet this variability is largely unquantified and not understood, so the best routes to improvement are not being identified or prioritized. For example, wheat yields achie- ved on-farm and in trials in the UK consistently range from <6 to >16 t/ha, whilst UK national average yields are st