BACKGROUND: Approximately one third of the global population cannot access essential health services. Access is strongly determined by sociodemographic group membership, with marginalised groups often experiencing the highest health needs but the worst access to care. I aimed to develop a continuous improvement approach to identify and address inequitable barriers to care, and then test this approach in the context of a community-based eye screening programme in Kenya, where half of all people do not receive the eye care they need. METHODS: I conducted evidence reviews to inform the development of an overall approach (dubbed ‘IM-SEEN’), and then implemented the three stages in Meru county, Kenya: 1) a cross-sectional sociodemographic analysis of access to community-based eye care clinics, 2) interviews, a survey, and a multistakeholder workshop to identify barriers and potential service modifications to improve equitable access to care, 3) setting up an embedded randomised controlled trial (RCT) to test the most promising service modification within the ongoing screening programme. FINDINGS: After analysing data from 4,240 people referred to local eye clinics, I found that only 46% reached care. Younger age, male gender, and sales/services/manual occupation were the strongest predictors of nonattendance (p<0.001). During interviews with 67 people aged 18-44 who had not received care, 21 different barriers and 25 potential solutions were suggested. I asked a further 401 members of the same group to rank the solutions and took the results to a multistakeholder workshop. Lay representatives, programme partners, and public health experts identified enhanced information provision as the most promising solution. I set up an embedded, pragmatic, adaptive platform trial. In the near future this will be used to test whether enhanced information provision – and other interventions that arise from further iterations of the IM-SEEN cycle - improve access to care. CONCLUSIONS: The IM-SEEN approach grounds continuous service improvement in engagement with groups who face the greatest barriers to care. The approach can be used to rapidly generate and test service modifications intended to improve equitable access to care.