Tripify
Every untracked mile is 72.5¢ left on the road.
A concept study of a mileage-tracking app for US self-employed drivers: the full process on record, from research and flows to a token-bound design system and 21 screens.
Role
Product Designer, solo end-to-end
Platform
iOS · concept
Timeline
4 weeks
Scope
Research · UX · DS · 21 screens
Екран Routes у девайс-фреймі на брендовому синьому, по боках картка Unclassified і екран Reports.
cases/mileage/img_mileage_01.jpg@2x 3360×2100per mile: the 2026 IRS deduction rate every untracked trip leaves behind
72.5¢
screens from wireframe to hi-fi, all bound to tokens and styles
21
research → UX → UI → validation plan, on a fixed timeline
4 wks
TL;DR · what I did
- Solo, end-to-end, on record: persona, journey map, user flows, wireframes, design system, 21 screens.
- Core insight: classification, not tracking, is the real job. GPS logs trips by itself; deciding business vs personal is where users fail an audit.
- Built the design system Light with Dark ready: 110 variables in 4 collections, 16 components, zero drift.
Overview
A concept study that plays by real rules.
This is a concept study. It is not affiliated with any company, and it invents no product metrics: every market number here is sourced, from the IRS 2026 standard mileage rate (Notice 2026-10) to the scale incumbents claim for themselves, 100B+ miles logged and 1M+ users at MileIQ.
The interesting problem is not drawing a tracker. It is that mileage deduction is a records problem wearing a maps costume: the IRS does not ask where you drove, it asks for a dated, classified, documented log. That reframe drives every screen.
Routes: лєджер поїздок із сумами вирахувань
cases/mileage/img_mileage_02.jpg@2x 3360×1890Unclassified trip: одна картка, одне свайп-рішення
cases/mileage/img_mileage_03.jpg@2x 1600×2133Mileage Reports: спершу долари, IRS-експорт в один тап
cases/mileage/img_mileage_04.jpg@2x 1600×2133Challenge
Problems & solutions.
Three failure points define the category: people forget to track, avoid classifying, and panic at tax time. Each one became a core flow.
Problem
Tracking is a chore
Self-employed drivers forget to log trips because logging competes with driving. Every forgotten trip is a real deduction lost at 72.5¢ per mile.
Solution
Zero-touch detection
Trips record themselves from drive detection. The user’s only job moves from “remember to start” to “confirm what happened”, a task that tolerates being done days later.
Problem
The unclassified pile
Automatic logging creates a new problem: a backlog of trips with no answer to the only question the IRS asks. Business or personal?
Solution
One swipe, one decision
Each unclassified trip is a card with one gesture: swipe one way for business, the other for personal. The pile that felt like homework clears at inbox speed.
Problem
Audit panic
IRS Publication 463 wants dated records with mileage, destination and business purpose. A map history is not a record; drivers rebuild logs from memory in April.
Solution
Reports built IRS-first
The report screen is generated from Pub 463’s required fields, not from what the app happens to store. Export produces the document an accountant actually asks for.
Research
Melissa drives 2,000 miles a month and logs none of them.
The persona, a real-estate agent juggling showings across the metro area, came from studying how incumbents pitch and how their users complain. Her journey has five stages, and the emotional low is not on the road: it is in April, reconstructing months of driving from a calendar.
Journey mapping surfaced the reframe the whole concept stands on: capture is a solved problem, classification is the abandoned one.
Аркуш персони + journey map на 5 стадій з кривою емоцій: квітневий обрив позначений як ціль дизайну.
cases/mileage/img_mileage_05.jpg@2x 2176×1224Insight
The product is a decision, not a map.
Competitors race on detection accuracy. But detection only fills a pile; the pile still needs answers. The concept bets its core loop on making the answer effortless.
GPS can log a trip on its own. What it cannot do is decide: business or personal? That decision is the product.
User flow: три доріжки (capture / classify / report) з чипами дата-шару: що кожен крок пише в запис.
cases/mileage/img_mileage_06.jpg@2x 2176×1224Wireframes
Structure first, chrome second.
Each key screen exists twice: a grayscale wireframe locking the hierarchy, and the hi-fi pass layering the visual system on top. The pairs read side by side, so every styling decision stays accountable to a structural one.
Пари wireframe vs hi-fi: Routes, Unclassified trip і Reports, вирівняні поруч.
cases/mileage/img_mileage_07.jpg@2x 3360×1890Design system
Small system, fully bound.
110 variables in 4 collections: primitives, semantic theme (Light shipped, Dark ready), a 4-grid scale and typography. 16 components from Button to Trip Card and Number Keypad, each with a hand-off description.
All 21 screens are bound to tokens and styles with zero residual drift, checked by script rather than by eye.
Огляд foundations: колекції змінних, тип-рампа і сет із 16 компонентів на одній дошці.
cases/mileage/img_mileage_08.jpg@2x 3360×1890Key screens
Three screens carry the whole loop.
Routes
The home list is a ledger, not a map app: every auto-detected trip with distance, deduction value and classification state at a glance.
Routes: автовизначені поїздки із сумами вирахувань
cases/mileage/img_mileage_09.jpg@2x 2176×2901Classify by swipe
The unclassified trip card leads with the two facts that drive the decision, route and time, and accepts exactly one gesture in each direction.
Unclassified trip: свайп вправо business, вліво personal
cases/mileage/img_mileage_10.jpg@2x 2176×2901Reports
Monthly totals framed in dollars first, miles second, because the deduction is the product. One tap exports the Pub 463 record set.
Mileage Reports: доларові підсумки + IRS-експорт
cases/mileage/img_mileage_11.jpg@2x 2176×2901Validation plan
Three hypotheses, each with a test attached.
A concept cannot claim outcomes, so it states its bets. Each hypothesis names the metric that would prove it wrong, which is exactly what a first live sprint would run.
H1 · Detection lifts capture
Automatic drive detection raises the share of logged trips versus manual start-stop logging. Validation: compare trip counts against odometer deltas in a 2-week diary study.
H2 · Swipe clears the backlog
Single-gesture classification keeps the unclassified pile near zero. Validation: time-to-empty-inbox and abandonment rate in a moderated usability round.
H3 · The report is the retention moment
Value lands when the export meets an accountant, so retention should spike around tax deadlines. Validation: cohort activity against filing dates.
Reflection
What this project taught me.
Honest framing is a feature
This case claims no invented users or revenue. Sourced market numbers plus a full process turned out to be a stronger story than fabricated “results” could ever be.
Concept is not a shortcut
Working without a client removes constraints, so I re-imposed them: a fixed 4-week timeline, IRS documentation as the requirements spec, and a token-bound system with zero drift.