In-Store Wayfinder
During my first month at Instacart, I went shopping to experience the product firsthand and discovered a major opportunity. Shoppers were spending an extra 40% of time in-store trying to find items for customers. I designed an in-store wayfinder that guides shoppers through unfamiliar stores, mapping item locations to store layouts and optimizing pick routes.
Product Design, UX Research
An in-aisle wayfinder that cut Shopper item-search time about 30% across an 80-store pilot, with no beacons or live planogram feed.
- Role
- Senior Product Designer, Shopper Experience
- Team
- 1 PM, 2 iOS engineers, 1 researcher, 2 retail-ops leads
- Timeframe
- 5 months, prototype to in-market pilot
- What I owned
- Research synthesis, prototype, retailer alignment, ship spec.
The problem
Shoppers were spending 40% of in-store time looking for items, not picking them.
Shoppers are paid by completed order, so every minute lost to search is unpaid. In unfamiliar stores, search-and-walk ate 38 to 44% of shop time before a single item hit the cart. Retailers would not fund beacons, shelf wiring, or a live planogram feed, so we inferred the layout from what we already held: thousands of past batches per store, each a record of where an item was actually found.
What shipped
Three surfaces. One lightweight map.
Aisle, side, position: location in the words Shoppers already use, not coordinates. One item ahead: a peek at the next pick turns a flat list into a route preview. Map on demand: one tap from the card, never pushed, because a Shopper mid-aisle needs the next token, not a floorplan.
Inline aisle hints grouped by section.
Location tokens and route peek.
Opt-in map sheet, one tap from the card.
What we killed
AR demoed well. The pilot saved us from shipping it.
Cut from the build · AR aisle overlay
What we tried. Camera-on AR floating aisle markers onto the live shelf feed. It demoed cleanly with four experienced Shoppers in a quiet back room.
Why it lost. The store killed it, not the demo. Retail lighting glare washed the markers out. Holding a phone up while steering a cart was the safety problem Shoppers named first. Indoor relocalization drifted wider than an aisle in most test stores, pointing the marker at the wrong shelf. We shipped the map sheet instead.
Pilot results
Faster batches, fewer support pings.
Exploration set
From inline aisle hints to a full store map.
TL;DR
Three directions, one shipped. The readout shows the time-to-find curve shifting left.
Next pass
I would have shipped the route view sooner.
Two more months
I cut a drag-reorder route view, scoped at three sprints, to land the pilot on time. I would now trade the tutorial cards Shoppers swiped past for that route view, and pull retailer alignment a week earlier. We learned late that no live planogram was coming, the one fact that forced a redesign I could have started from.
- Tidjane
Shipped with
Notes
- 1
Pilot readout, internal Instacart program review. Directional pilot-window measurements, not steady-state production figures.
More work
Other projects worth a look.
