Subseasonal to seasonal (S2S) forecasts offer valuable opportunities for the agricultural sector by bridging the gap between short-term weather forecasts and long-term climate projections. However, their adoption faces significant challenges, including high uncertainty, coarse spatial and temporal resolution, and difficulties in communicating probabilistic information effectively. Establishing a meaningful dialogue between forecast producers and users is crucial to ensure these forecasts address user needs. This study explores the use of a serious game methodology to facilitate such dialogue and guide the development of S2S forecast services tailored to the Norwegian agricultural sector. Through game sessions involving farmers and agricultural advisors, we identified strategies users employ to interpret S2S forecasts and uncovered diverse use approaches for these products. Our findings highlight key recommendations for service development, including ensuring seamless integration of forecasts across different lead times, providing narrative explanations of probabilistic data, supporting user familiarization with the tools, and exploring the use of recent years' conditions as reference points instead of historical averages. The serious game approach proved effective in fostering meaningful dialogue and mutual learning among agricultural users, forecasters, and service developers, ultimately supporting forecast services' co-production.