Women in low- and middle-income countries (LMICs) have limited access to long-acting contraceptives. Access to long-acting contraceptives, such as subcutaneous contraceptive implants, could be increased by task-shifting implant administration from advanced to minimally trained healthcare providers. The objective of this study was to investigate the usability of a task-shifting device for administering subcutaneous contraceptive implants. Healthcare providers (n = 128) from multiple health centers in Ethiopia were trained to administer implants on an arm simulator with the traditional method and a method using the device. Participants were observed while inserting implants into the arm simulator, and procedural error rates were calculated. Observations were analyzed using an iterative inductive coding methodology. For the device-assisted method, minimally trained healthcare providers had larger procedural error rates than other professions (p = 0.002). For the traditional method, physicians had larger procedural error rates than nurses and midwives (p = 0.03). Several procedural errors were identified such as participants inserting and removing the trocar and plunger completely or inserting and/or removing the trocar too far or not enough. These findings reinforce the importance of performing formative usability testing during the early phases of a medical device design process, considering users' mental models, and avoiding assumptions about healthcare providers' abilities.