SPOTTER turns a walk into a patrol. Find the mural, the fire escape, the rooftop view — photograph it, claim it, move on. A concept app for people who bounce off step counters, designed against what the published research already says about why they bounce.
A number that only goes up is a scoreboard for a game nobody agreed to play.
Movement apps are good at counting and bad at making you want to go outside. They reward volume — more steps, more minutes, more streak — which turns walking into an unpaid job with a quota. The retention numbers show how that ends.
SPOTTER starts from a different place. Give the walk a destination and a reason to look up, and the distance takes care of itself. You're not hitting 10,000 steps. You're getting the shot.
This is a weekend concept, not a funded project — so instead of inventing participants, I built it on published work. Pokémon GO has been studied properly, streak mechanics have been studied properly, and fitness app churn is well documented. Between them there's a clear design brief sitting in plain sight.
A Microsoft Research and Stanford team studied 32,000 wearable users over three months, and estimated the game added 144 billion steps to US physical activity. Heavily engaged players walked 1,473 more steps a day — over 25% above their prior activity. The finding that matters for design: it reached low-activity populations, while all four leading mobile health apps in the same study largely drew from an already very active population.
Althoff, White & Horvitz · J Med Internet Res, 2016Read the study
A Harvard-led cohort study in the BMJ followed 1,182 US adults aged 18–35 using automatically recorded iPhone step data. Steps rose after installing the game — but the association was moderate, and by six weeks it had disappeared entirely. A separate study of 167 players measured a 34.8% jump in daily steps, from about 5,700 to about 7,650, with more in-game progression corresponding to more walking.
Howe et al. · BMJ, 2016 · Xian et al. · JAHA, 2017Read the BMJ study
Health and fitness apps are widely reported to retain only around 3–4% of users by day 30. A scoping review of lifestyle and mental-health apps found a median of 70% of users discontinuing within 100 days, with the sharpest drop-off in the first two weeks. Novelty decay isn't a Pokémon GO problem. It's the category's default.
Aggregated retention reporting & app-abandonment scoping review, 2024–26See the summary
Duolingo describes streaks as its single most effective retention lever, and reports that learners who reach a seven-day streak are markedly more likely to return the next day. That's the upside. The mechanic runs on loss aversion, which is also why it hurts: in Weinstein's qualitative work on adolescents and Snapchat, teens said streaks motivated them and gave them social validation, while also describing the experience as stressful, obligatory, and a source of anxiety about letting someone down.
Weinstein, 2018 · Duolingo retention reportingOn Snapchat streaks
Duolingo's answer to streak anxiety is the streak freeze — and the interesting detail is that it's applied silently, on the user's behalf, while they're absent. A freeze that required you to open the app and confirm it would fail exactly the people it exists to catch. Giving new learners freezes up front improved retention; unlimited forgiveness would hollow the mechanic out.
Shuttleworth (Duolingo Retention) via Lenny's Podcast, 2024Mechanic breakdown
Put those five together and the brief writes itself. A play frame gets sedentary people moving in a way health apps don't. That frame decays in about six weeks. Streaks would patch the decay, but they buy retention with anxiety. So the design problem is: how do you renew novelty on a schedule, without using loss to do it?
Daily streak. Step goal. Don't break the chain.
Seasonal issues that refresh the novelty. Totals that only ever go up.
Spots expire daily, so there's still urgency in the moment. But nothing you've already earned can ever be taken away, and a new themed issue lands before the old one goes stale. Urgency without debt. That's the whole idea, and every screen below is just that idea drawn out.
A comic-and-street visual language built for a product used at night, outdoors, one-handed. Dark canvas as the stage. Red for action, blue for support, yellow for surprise — never all three at full volume.
Hierarchy runs dark canvas → white type → red action → blue support → yellow surprise. Yellow is capped at roughly one moment per screen.
Five faces, five jobs. The contrast between condensed display and neutral body type is what carries the personality — not decoration.
Sharp editorial panels mixed with rounded controls. The mixed shape language is deliberate — it's what keeps it street-poster instead of generic SaaS.
44px min height, 2px border, hard offset shadow on hover. No thin outlines.
One rotated element per composition. Rotation is a garnish, not a grid.
48px height. Focus is a 2px blue outline, never colour-only.
Colour the number, keep the label neutral.
Yellow default, red for time-limited. Never sits over body text.
A node at the break. Used at section seams only — sparingly.
Patrol, route, capture, collect, learn. Tap the phone to move through it — everything else is settings.
The map opens on what's within walking distance right now — never a total, never a streak. Blue draws the suggested route; red pins mark live spots, each carrying its category glyph so you can read what's nearby without tapping. The card sits thumb-height so the whole primary action lives in the lower third.
Spots expire daily. That's the urgency the streak used to supply, but it resets rather than accumulating into a debt you can default on.
One maneuver at the top, sized to be read at a glance while moving. Turn-by-turn steps below, and hand-off to Apple Maps or Google Maps for anyone who'd rather navigate in what they already trust.
Handing off raises an obvious question — does the spot survive leaving the app? The line under the buttons answers it: tracking continues. Route guidance is haptic-first, so the screen isn't required between spots.
Within 30 metres the viewfinder unlocks. The corner-bracket reticle frames the subject and confirms the match without a modal — the system's web-line motif doing real work rather than decoration.
One control, 68px, centre-bottom, reachable with either thumb. Everything secondary sits clear of the safe zone so a moving hand can't fire it by accident.
Captures fill an issue — twelve spots, themed. Six categories run through it: murals, food trucks, hydrants, rooftops, architecture, ghost signs. Broadening past street art wasn't only a licensing decision; hydrants and fire escapes exist on almost every block, while murals cluster in a handful of neighbourhoods.
A season structure gives lapsed users somewhere to return to, because a half-finished issue is an invitation rather than a broken streak. Locked covers stay visible, so you can always see what you're walking toward.
A badge stops being interesting around week six — that's what the Pokémon GO data shows. Learning something true about the building you just photographed doesn't decay the same way. So every claimed spot opens into its story: who made it, when, and what stood there before.
Artist credit sits above the facts, not beneath them, and mural spots are listed with permission. On a screen built around someone else's work, attribution is the first thing rather than a footnote. This screen deliberately breaks the one-paragraph rule the others follow — it's read standing still, after the capture, not mid-walk.
Rockaway Ave mural
Painted over eleven days in the summer of 2019 on the side of a former knitting mill. The building supplied trim to Manhattan garment factories until 1961 — the sawtooth roofline above the mural is original to it.
Bushwick had 26 breweries before Prohibition. Three of the buildings on this block were part of one.
TAP THE PHONE TO CYCLE SCREENS
This app is used outdoors, one-handed, in motion, often in bad light. That's an ergonomics problem before it's a visual one — and it's where a loud design system is most likely to fail.
A dark UI is the riskier choice outdoors, so I treated glare as the design constraint rather than an afterthought. Every text pair was checked against the system's minimum contrast, not its average.
→ Body text floor raised to 15px; no text under 4.5:1; halftone textures removed from behind any type.
An app you use while walking is a one-handed app. The comfortable thumb arc on a 6.1″ phone covers roughly the lower two-thirds of the screen and misses the top corners entirely.
→ All primary actions live in the bottom third; the top bar is display-only, never interactive.
Nobody stops walking to read. I designed each screen against a two-second glance: if the next action wasn't obvious in that window, it got cut or promoted.
→ Max one paragraph per screen; distance and time in mono so they scan as data, not prose.
A system built on red and blue can't use hue to carry meaning. Live, locked, and claimed states each carry a shape and a word as well as a colour.
→ Locked spots get a padlock plus reduced opacity; claimed spots get a ribbon label.
Pokémon GO generated real case reports of distraction injuries. An app that makes people stare at a phone while walking is a safety failure, whatever the engagement numbers say.
→ Route guidance is haptic-first; the app never requires the screen between spots.
Stickers rotate in, cards snap, the route draws. All of it is decoration on a screen used while the ground is moving underneath you.
→ Reduced-motion removes all transform animation and keeps only opacity fades.
The honest part. This is desk research, not primary research — I'm reasoning from other people's participants. These are the gaps that leaves, ranked by how much damage they'd do if I'm wrong.
Every number in this case study comes from published work, not from participants I didn't recruit. Where reporting is aggregated rather than peer-reviewed, it's labelled as such.
SPOTTER is an unaffiliated concept. Pokémon GO, Snapchat and Duolingo are referenced only as subjects of published research. No assets, marks or interfaces belonging to any third party are used in this project.