Synthesis — Wayfinding
Cross-participant synthesis, n=8 (2 deep write-ups: Interview — P04, Interview — P07; 6 in field notes; ~90 usable diary entries of a ~112 target — decent for a two-week light-touch study).
Method: affinity mapping, bottom-up. Started with ~240 tagged observations, clustered on the wall, collapsed to the six themes below. I've kept observed behavior separate from my inference throughout — where I'm extrapolating, I say so.
One-line finding: People don't navigate space. They navigate a story made of landmarks, told in the body, and the more you outsource that story to the app, the less of the place you actually own.
Theme 1 — People navigate by story and landmark, not by direction
Nobody in the sample navigated by cardinal directions. Not one. They navigated by named things and remembered images — "turn at the tiled saint," "it's near the fountain," "past the busker." Street names were nearly absent even after weeks in place (P04 retained zero after six weeks; P07, three years in, thinks in ~40 landmark nodes and "the strings between them").
"It's not a map. It's a feeling with landmarks stuck in it." — P07
Implication: A wayfinding app built on street names and compass directions is speaking a language humans don't navigate in. Directions should be landmark-anchored ("continue toward the yellow church"), not cardinal or street-name-based.
Theme 2 — Getting lost productively (and who's allowed to)
The single sharpest divide in the data is the emotional valence of getting lost. Experienced navigators treat it as generative — the mechanism by which you earn the good routes ("you can't be given the good routes, you earn them by being wrong first," P07). Newcomers treat it as threat and personal failure (P04: panic, self-blame).
Implication: The app's job is not to eliminate getting-lost — that would deskill people, see Theme 6 — but to change its stakes. Make wandering feel safe and recoverable (a reliable "how do I get un-lost" fallback) so exploration stops being frightening. Protect the productive detour; remove the panic.
Theme 3 — The anxiety of the blue dot
The GPS pin is simultaneously the primary crutch and a primary source of dread. In dense/waterside streets the dot drifts and spins, and for reliant newcomers this is a genuine acute-anxiety event, not a minor glitch (P04 named "the spinning blue dot" as her panic trigger, unprompted, three times). The more dependent the user, the more catastrophic the dot's failure feels — dependency and dread scale together.
Implication: Precision-comms matter emotionally, not just functionally. Show uncertainty honestly (a confidence halo, not a false-precision dot). And critically: don't deepen dependence, because dependence is what makes the failure mode devastating.
Theme 4 — The body knows the hill before the map does
Physical effort is navigational data the app throws away. Gradient, sun, wind, time-of-day. P07's whole routing logic minimizes climbing ("the map doesn't know the hills"); he routes around late-afternoon sun-walls and the 3pm school crowd. Diary entries echoed it: shortcuts were described in terms of effort and comfort ("the flat way," "the shady way"), almost never distance.
Implication: Shortest is not best. Wend should route on human cost — elevation, exposure, crowd, surface — with walker/cyclist-specific weighting. This is a genuine differentiator; the incumbents are all still optimizing flat-map distance.
Theme 5 — Landmarks are sensory and social, not geographic
The landmarks people actually use are multisensory and often human: the smell of a bakery, a fish shop, a busker who's "a waypoint now," the school letting out, a particular quality of light. These don't exist on any map layer, yet they're the actual anchors of the mental model.
Implication: Consider user-generated, personal landmarks — let people pin their own sensory/social anchors ("the good bakery," "the loud corner"). The city's official POI layer is not the layer people navigate by.
Theme 6 — Confidence is a route you earn; the app can shortcut past it
The throughline under everything. Repetition converts space into place — abstract coordinates into somewhere you own. Turn-by-turn lets you arrive without ever building that model, so you stay permanently dependent (P04: "I outsourced it before I ever learned it... I skipped the part where you learn a place"). The app can either replace the learning or scaffold it. Right now every incumbent replaces it.
"Give me my map back better. Don't give me the map." — P07
Implication — the strategic one: Wend's opportunity isn't to be a better turn-by-turn. It's to be the wayfinding tool that builds the user's own spatial model instead of substituting for it — fading its own guidance as your competence grows, teaching you the landmark-chains, celebrating the unassisted arrival (P04's happiest moment in the whole study). Design for graduation, not dependence.
The model, one page
We navigate by story, in the body, spot to spot. The map is flat; the city has hills and light and smells and a saint on a wall. Getting lost is how locals learn and how newcomers suffer — same event, opposite meaning. The blue dot is a crutch that becomes a source of dread in proportion to how much you lean on it. The winning wayfinding tool doesn't perfect the dot; it hands you back your own map, better — and then gets quieter as you learn.
Open questions / for the readout
- P05, P08 write-ups still pending — do they hold Theme 2's valence split? (early read: yes)
- The graduation/fading-guidance idea is the big swing. Product will resist it (retention math cuts against "make the user need us less"). But it's the honest finding, and it's the only one that isn't just a nicer turn-by-turn.
- Note for me: this is genuinely good work and I'm about to walk away from it. See Quitting Wend.