INTRODUCTION: Rapid advancements in genomic testing have revolutionised cancer care diagnostics and treatment. However, keeping pace with the evolving genomics knowledge is a challenge for oncologists who are not genomic experts. This detrimentally impacts on equitable patient access to related services and benefits which require training in genomics. In Australia, cancer incidence, survival, and mortality rates are significantly worse in the most socioeconomically disadvantaged areas compared to the least disadvantaged areas. Guided by implementation science methods, the research aimed to determine how to support oncologists with varying levels of genomic expertise to tailor optimal treatment decisions and deliver a high-quality service, across diverse locations. METHODS: We used a novel approach combining clinician intuition and implementation science theory to co-design service interventions (i.e. service models) and associated implementation strategies to inform operationalisation. Phenomenology and principles of co-design guided two phases of data collection with two separate cohorts of oncologists delivering care to advanced cancer patients. Phase 1 interview data was coded thematically to develop the service models, while phase 2 focus group data was used to identify implementation strategies to support service model operationalisation. The Consolidated Framework for Implementation Research (CFIR) informed phase 1 and 2 data analysis. RESULTS: Phase 1 established three overarching themes and nine subthemes: (1) access - potential for inequitable patient access by centralising genomic expertise, (2) indicators for test use - identifying suitable patients for Comprehensive Genomic Profile (CGP) testing, and (3) supporting use of results - confidence to discuss results, particularly from germline and somatic testing. Five challenges were prioritised, mapped to the CGP clinical pathway, and coded to 11 unique CFIR constructs. Across all five prioritised challenges, we recorded 19 intuitive and generated 21 theory-informed strategies. The development of three service models (i.e., centralised expert, local super user and point of care resources) arose through considering these strategies in combination with the study teams broader experiences with the iPREDICT trial. In Phase 2, we identified 11 implementation challenges, mapped to 7 CFIR constructs, and 11 intuitive and 20 theory-informed strategies for service model operationalisation. CONCLUSION: The service models generated from our study are currently being tested in a multi-centre implementation study to evaluate feasibility, effectiveness, acceptability, sustainability, and scalability in 2025.