Breadcrumbs

The problem

Every navigation app answers 'how do I get there'. Almost none answer 'how do I get back', which is the question that matters on an unlit street, a ridge line, or a run you improvised. The ones that come close want an account, a signal and your location history. Breadcrumbs refuses all three and still has to be genuinely useful — that constraint is the product.

What I did

• Owned product direction and the backlog: what got built next, what got upgraded, and what was cut to keep the promise simple enough to state in one sentence.
• Held the line on scope. An offline, account-free app attracts feature requests that quietly require a server; each one got measured against the covenant before it entered the backlog.
• Specified the return itself: turns derived from the recorded route's geometry rather than the compass, so guidance never flips left and right on a stationary phone.
• Defined the safety layer as separate escalating pieces — a check-in timer, safe zones that notice when you leave, a buddy that speaks up when a walk goes quiet, and an SOS with a five-second cancel so a pocket press cannot arm it.
• Scoped per-person sharing levels inside a circle (live location, status only, or last known position), so sharing is a dial rather than an on/off switch.
• Specified the watch companion around a single question — which way home and how far — with SOS one tap away, leaving the heavy work on the phone.
• Specified the SOS packet for the people who would actually be searching: accuracy circle, heading and speed, the last fifteen minutes of path in every ping, battery, and whether you were walking, driving or still — with 20, 10 and 5% battery each forcing a fresh location out.
• Kept every optional signal off until switched on, and write-only where it touches other apps: place learning, the guardian and Discover start off, and the Health integration ships with an empty read set.
• Ran user research on the return journey itself, watching where people lose confidence in a route they have already walked, and fed it straight into how guidance is phrased and timed.
• Ran QA and structured testing across the four ways people actually use it — walking, running, hiking and cycling — and reviewed each build against the spec before sign-off.
• Turned heuristic evaluation and tester feedback into a prioritised fix list rather than a bug pile, so polish and real gaps got sequenced against each other.

Where it landed

Stack & tools