Unassisted visits to unfamiliar buildings can be an insurmountable barrier for people with visual impairments and blindness. Depending on their individual needs and preferences, people with visual impairments and blindness need specific information regarding indoor barriers, accessibility features, landmarks, temporal relations, etc. to find their way. While research in the field of accessible indoor navigation has been going on for several years, proposed systems need a huge amount of localization beacons, rely on proprietary map data, and mostly follow a classical turn-by-turn navigation approach, resulting in high installation and maintenance costs and presumably do not lead to accurate cognitive maps in users. This paper introduces the concept of app-based, accessible indoor orientation support through automatically generated landmark-based route descriptions utilizing open-source map data. A user study is conducted with 4 blind participants, interacting with both a working app-based high-fidelity prototype for indoor orientation support and a classical turn-by-turn indoor navigation system using Wizard-of-Oz to traverse through an unfamiliar building, followed by backtracking to a starting point. Cognitive maps, performance during backtracking, perceived usability, and acceptance are measured after execution of the route-finding. Additionally, an expert interview is conducted with a blind guide dog user. Results indicate a good acceptance and usability of the system. Furthermore, measured cognitive maps are closer to ground truth when using the proposed system in comparison to the control condition.