A delivery app built around fast decisions
A food delivery app where browsing, ordering, and live order tracking are built around quick decisions and a smooth checkout.
The brief, in specifics
- Industry
- Restaurants & Hospitality
- Duration
- 14 weeks
- Cooperation model
- Time & materials
Client name withheld under NDA. Engagement details are shown to the extent our agreement permits.
Scan-and-Select Browsing
Visual food feed designed for quick decision making and immediate menu customization.
The feed is images first, with the decision-critical facts (price, time, minimum) burned into the card, so choosing doesn't require opening anything. Cards are virtualized and images prefetched two screens ahead, which keeps scrolling at sixty frames on mid-range Android. Customization opens as a sheet over the feed, with no route change, so backing out doesn't lose your place in the list.
- Decision facts on the card: no tap needed to compare
- Virtualized list with two-screen image prefetch
- Customization as a sheet, so scroll position survives
A menu with the customization sheet open over it, so backing out keeps your place in the list.
3-Tap Express Checkout
Checkout on one screen, with address, card and the usual tip already filled in before the third tap.
Streamlined checkout with saved addresses and 1-click mobile payment methods.
Checkout is three taps because the address, the payment method and the tip default are all resolved before the screen renders. Payment goes through Stripe's native sheet, so card details never touch the app. The order is created optimistically with a hold and confirmed against the restaurant's acceptance, and a rejected order refunds itself without anyone contacting support.
- Address, payment and tip resolved before render
- Native Stripe sheet: card details never touch the app
- Rejected orders self-refund without a support ticket
Live Driver Tracking
Real-time WebSocket driver location tracking with high-accuracy ETA predictions.
Driver position streams over a WebSocket at four-second intervals and is interpolated on the client, so the marker glides smoothly instead of teleporting between pings. ETAs come from routing against live traffic, not straight-line distance, and the app shows an honest range in place of a falsely precise minute. When the socket drops it degrades to polling instead of freezing the map.
- Four-second position stream, interpolated client-side
- ETAs routed against live traffic, shown as a range
- Graceful fallback to polling when the socket drops
The rider between the kitchen and the door, with the arrival shown as a range and the map updated every few seconds.
What we were brought in to do
Browsing felt cluttered and checkout dragged. We rebuilt the ordering experience for people deciding what to eat in under a minute.
A regional delivery operator competing with two national apps on the same restaurants, where the only available differentiator was the ordering experience itself. Repeat rate was the commercial problem: acquisition worked and second orders didn't follow. The engagement was scoped around the minute between opening the app and committing to an order.
Design & Build
Where the old way broke
Too many steps to order and no reliable way to see where a delivery actually was left customers frustrated and re-ordering elsewhere.
Choosing required opening a restaurant, then a menu, then a dish before the price and delivery time were visible: three navigations to answer a question the card could have answered. Then checkout re-asked for an address that was already on file. Funnel data put the two losses at browse and at the address step, and the second one was people who had already decided to buy.
We simplified browsing into a fast scan-and-select flow, streamlined checkout to a minimal number of steps, and added live order tracking customers could trust.
What we built together
- 01
Redesigned the home feed for scan-and-select ordering
The card content was agreed with restaurants, since burning price and minimum onto it exposes a comparison they had previously been shielded from.
- 02
Cut checkout to three taps for returning customers
Address, payment method and tip default are resolved before checkout renders. That's what makes three taps a fact instead of a target.
- 03
Built live driver tracking with accurate ETAs
Driver position streams every few seconds and is interpolated on the client, with ETAs routed against live traffic and shown as an honest range.
- 04
Instrumented funnel drop-off and iterated weekly
Funnel drop-off was instrumented before the redesign and reviewed weekly, so every change was answerable against the specific step it was meant to move.
Phase by phase
Phase 1: Funnel & Drop-Off Audit
Ordering Experience Analysis
Audited restaurant search, cart build, and checkout drop-off across 20,000 monthly orders to isolate order friction.
- Funnel Audit Report
- Cart Friction Map
- Mobile Delivery UX Benchmark
Phase 2: Visual Browsing UX Design
Scan-and-Select & 3-Tap Checkout
Redesigned home feed for visual dish scanning and engineered a 3-tap checkout for returning users.
- Figma Design System
- Interactive Prototype
- Checkout Micro-animations
Phase 3: Real-Time Engineering
WebSocket Driver Tracking & Payments
Engineered real-time driver tracking with WebSockets, Mapbox GL routing, and 1-click mobile payments via Stripe.
- Real-Time Tracking API
- Mapbox Navigation Layer
- Stripe Payment Bridge
Phase 4: Store Launch & Funnel Optimization
App Release & Weekly Conversion Cycles
Shipped iOS and Android apps, running weekly conversion optimization cycles that increased order conversion by 31%.
- App Store & Play Store Submissions
- Funnel Analytics Dashboard
- Release Sign-off
Dispatch with 88 live orders on the region map and the three running late called out.
Operational results after launch
+31%
Order conversion
−48%
Checkout time
+27%
Repeat orders
Conversion is orders per session for sessions that reached the feed, so a change in traffic mix doesn't move it. Checkout time is median time from cart to confirmation for returning customers. Repeat orders compares customers ordering twice within thirty days, before and after, for cohorts acquired in the same way.
Client name withheld under NDA. Figures are approximate, drawn from the engagement’s own reporting.
About our collaboration
A cross-functional team of 5 worked on a time & materials basis over 14 weeks, covering Mobile design, Ordering flow, Live tracking. We ran daily standups with an in-house lead in the room, and a demo at the end of every sprint. Scope changed twice during the engagement, and both times the change was priced and agreed before work started.
The funnel was instrumented before any redesign and reviewed weekly, so each change was answerable against the step it was meant to move. Restaurants were involved in the feed work. The decision to burn price and minimum onto the card needed their agreement, since it exposes a comparison they had previously been shielded from.
What we'd carry into the next one
- 01
Streamlining checkout to 3 taps cut checkout time by 48% and boosted repeat orders.
Most of the checkout time was never the form. It was the address step re-asking for something already on file, which reads to a customer as the app not knowing them.
- 02
Real-time WebSocket driver tracking with accurate ETAs reduced customer support inquiries.
Tracking cut support because the questions were "where is it" and "when will it arrive." A map that moves smoothly answers both without anyone typing.
- 03
Visual scan-and-select dish browsing increased cart conversion by 31%.
Putting the comparison on the card was the whole conversion gain: the decision people were failing to make was between restaurants, not within a menu.
Four apps, one dispatch layer
High-concurrency delivery architecture: smooth live tracking, WebSocket courier dispatch, and a 3-tap checkout.
The same product, seen from the customer, the rider, the kitchen and dispatch. Switch tabs, or use the arrow keys once one is focused.
Position updates
Polling vs. a live stream
Most delivery apps ask the driver's phone for its last position every fifteen seconds and hold the map still in between. Supperly's tracking screen streams positions every four seconds and interpolates the path between them, which is why its ETA can afford to be a tight, held range instead of a shrug.
Standard delivery app
Polling every 15s (sped up here for legibility)
A wide, shifting window: the last fix is already out of date by the next one.
Supperly
Positions stream every 4s, interpolated for a smooth glide
On time: the same window the tracking screen holds to for the whole trip.
Checkout, underneath
The hold, from tap to capture
Checkout already discloses this in one line. Underneath it is a Stripe pre-authorization the kitchen has a short window to act on, with a clean exit on either side of that window.
Checkout: Stripe places a pre-authorization hold
A hold is placed until the kitchen accepts. If it declines, you're refunded automatically.
The kitchen has 2 minutes to accept the ticket
Comfortable for nearly every venue on the board. The two flagged for slow acceptance are the only ones that sometimes run past it.
Accepted
The hold is captured as a real charge. Cooking starts on the same ticket the docket already shows.
Declined or timed out
The hold is voided instantly. No charge is ever made, so there's nothing to dispute and no chargeback risk.
Building a multi-sided marketplace or real-time geolocation app? Scope your build in 3 minutes.
Scope your buildNearby engagements
AI & AutomationAI phone ordering that fills every table's takeout slot
A voice agent that answers every takeout call, takes orders straight into the POS, and upsells like a trained host, even at the dinner rush.
Restaurants & Hospitality · 9 weeks
Web PlatformsHospitality POS that keeps working when the internet doesn't
A venue system built as a desktop terminal that owns its own data, so tabs stay open, dockets keep printing, and a dropped line is an inconvenience instead of a closed bar.
Restaurants & Hospitality · 26 weeks
Mobile AppsJobs, courses, housing and the paperwork: one app instead of four rumors
A React Native app for people working abroad, holding the four things they actually search for (work, training, housing and official guidance) behind one account.
Professional Services · 16 weeks
Let's talk
Running a large platform, shaping a first MVP, or getting a product ready for a funding round? Tell us where you are. We'll shape the process around it, and stay with you after launch.
