Designing trust when GPS accuracy is unreliable
Mapping technology constraints to user mental models
Problem
Most pet geo-fencing apps assume ideal conditions — but real users spend most of their time indoors, in courtyards, parks, or wooded areas — where accuracy degrades. When this happens, interfaces continue to behave as if data were precise, creating false alarms, anxiety, and loss of trust.
Project goal
To design a system that's honest about its own limitations, so users can trust it even when it's wrong.
Transforming complexity into Product Vision
Research on what users exeprience
Interviews with dog owners to find out how people think and act during stressful events (e.g. losing a pet); what they expect the system to do when something goes wrong. Key users groups. Personas. Empathy and journey mapping.
Learning how GPS signal quality works depending on context (indoors, trees, buildings, sky visibility) and how it causes mismatch between users' mental models and reality.
Inventory of pain points as experienced by users of existing products.
Research synthesis
Most situations fall under these three scenarios (see below matrix):
Pet indoors / courtyard (poor GPS)
Pet outdoors with obstructed sky view
Real escape events requiring action.
Defining the expected user experience
Defined the user story solving each pain point (persona, context, behaviour, need, action).
Drew flowchart logic of the interaction design based on each user story.
Illustrating the visions for each scenario.
Building & testing the prototype
Key findings
Users did not want technical explanations.
Users expected the system to learn and stop repeating false alarms.
Trust increased when uncertainty was acknowledged instead of hidden.
Each iteration reduced friction, clarified intent, and simplified decision paths.
UI
Learning points
Good interaction design aligns mental models with real-world constraints.
Users don’t need explanations — they need clarity.
Usability testing needs measurable goals.
Design process
Full research notes, synthesis artifacts, and iteration history [View as PDF]