Abstract Background It is important to evaluate fidelity of delivery and engagement during feasibility trials. However, there is little guidance on how to systematically develop strategies to improve implementation if problems arise. We aimed to: 1) Assess fidelity of delivery and engagement, 2) Identify factors influencing fidelity of delivery and engagement, and 3) Develop strategies to improve fidelity of delivery of, and engagement with, a complex intervention to improve independence in dementia, within a feasibility trial. Methods A mixed methods evaluation of an intervention that aimed to improve independence in dementia. To assess fidelity of delivery and engagement, observation and self-report methods were used: 60% of audio-recorded intervention sessions were transcribed and reliably rated for fidelity. Providers (n = 12) and people with dementia/supporters (n = 34) were asked to complete checklists after each session. Descriptive statistics were used to analyse the data. To identify factors influencing fidelity and engagement, one-to-one semi-structured interviews were conducted with providers (n = 8), people with dementia (n = 7) and supporters (n = 7). Thematic analysis and content analysis were used to analyse data. To develop strategies, we followed four steps proposed by the authors of the Behaviour Change Wheel (1. Understanding the behaviour, 2. Identifying intervention functions, 3. Specifying intervention content, 4. Identifying mode of delivery). Results Researcher ratings indicated moderate fidelity and provider/participant ratings indicated high fidelity of delivery. Knowledge, providers’ attributes, ease of adaptation of the intervention in relation to participants’ needs and logistical considerations influenced fidelity. We developed four strategies to improve fidelity of delivery of PRIDE: 1) showing a video, 2) giving an instruction sheet, 3) giving time to practice and 4) providing continued support. Participants reported high levels of engagement. Participants’ attributes, capability and opportunity influenced engagement. We developed four strategies to improve engagement with PRIDE: 1) a session summary document, 2) clear instructions, 3) time to practice activity and 4) providing regular compulsory telephone support. Conclusion Fidelity of delivery and engagement are complex behaviours. This manuscript provides an example of how the Behaviour Change Wheel can be used during a feasibility trial to systematically develop strategies to improve implementation of complex interventions.