Case Study · Personal Project
Shift
Buddy
A shift-logging app designed for tipped and gig workers who need a faster, smarter way to track income across multiple jobs.
View Clickable Prototype →
A shift-logging app designed for tipped and gig workers who need a faster, smarter way to track income across multiple jobs.
View Clickable Prototype →
Tipped workers — bartenders, servers, baristas, rideshare drivers — have a genuinely messy income picture. They juggle multiple jobs, get paid in a mix of cash tips, card tips, and hourly wages, and have no single place where any of it lives together.
Most tools that existed were either too generic (notes apps, spreadsheets) or too rigid (payroll tools built for salaried employees). Nothing was built for the specific cognitive reality of a closing shift at 2am.
Shift Buddy started as an exploration into that gap — a logging tool designed around how tipped workers actually think about their money, not how accountants do.
"Tipped workers who hold multiple jobs lack a fast, frictionless way to log shift earnings by employer — resulting in inaccurate income tracking and a poor understanding of which job is actually worth their time."
Workers are tired, rushing out, or unprepared to log immediately after a shift. By the time they remember, the details are fuzzy.
"Which job is actually worth my time?" was a question none of the research participants could answer quickly — or at all.
Workers don't think of their night's earnings as one number. Apps that collapse cash and card into a single field lose meaningful data.
n=8 participants, 3 weeks of self-reported shift logging behavior
8 sessions observing workers in their actual end-of-shift environment
30 respondents — supplementary pressure-test of qualitative themes, directional use only
7 products evaluated across multi-job support, tip split, backward logging, speed, and projections
Findings 1–7 are grounded primarily in the diary study and contextual inquiry. The supplementary survey (n=30) provides directional reinforcement where noted — not statistical validation. Finding 8 is drawn directly from the competitive audit.
Real-time logging was functionally nonexistent across all qualitative sessions. The end of a shift is chaotic — tips are being counted, coworkers are leaving, managers are running closing tasks. There is minimal cognitive bandwidth for app interaction in the moment. The supplementary survey reinforced this directionally — the clear majority reported logging the following day or less frequently.
Quick Select date chips — Today, Yesterday, Last Friday, Last Saturday — cover the realistic logging window for the vast majority of sessions without requiring a calendar interaction.
11 of 14 participants held more than one job. When asked which employer paid better per hour including tips, 9 of those 11 said they genuinely didn't know. One participant who had worked at the same two venues for 18 months still couldn't answer without doing math on the spot. The problem wasn't motivation — it was a complete absence of per-employer income history.
Employer selection is the very first step in the log shift flow — not an optional field. Every shift entry is anchored to a job from the start, which is what makes employer-level analytics possible over time.
Without exception, every diary study participant distinguished between card and cash tips in their own mental accounting. Common language: "I made $80 on card and like $40 cash." Cash is immediate and tangible; card tips often pay out differently or get pooled. Workers use the split to informally assess the quality of a shift.
Card Tips and Cash Tips are separate sequential screens — never collapsed into one field. This mirrors how workers already think, and preserves the tip-split data that becomes meaningful in the Insights section.
During prototype testing of an earlier form-based design, participants consistently fumbled with native iOS keyboards when entering tip amounts. Keyboards appeared with a delay, covered the input field, and offered no confirmation feedback. Workers who handle cash and POS terminals all day have strong physical memory for numpad layouts — a form-based input runs against that muscle memory.
Custom 3x4 numpad with a large real-time display acts like a point-of-sale interface — familiar, fast, and self-correcting. The Clear action in the nav bar handles fat-finger recovery without breaking flow.
Participants were not unaware that tracking income was a good idea. The tools that existed were designed for freelancers, general budgeters, or employers — none were built around the mental model of a tipped worker ending a shift at 11:30pm. The survey reinforced this directionally: memory and informal notes dominated, and several respondents mentioned trying an app and giving up.
Shift Buddy's warm palette and conversational copy is a deliberate departure from the cold blues and dashboard-heavy UIs of financial apps. The goal is a tool that feels like something you'd actually open at 2am.
When shown a concept for photo-based auto-fill from a clock-out slip, 11 of 14 qualitative participants rated it their most desired feature. But behavioral data tells a more complicated story: slips are often crumpled, thrown away, or unavailable entirely. More than half of contextual inquiry participants could not produce a receipt or slip at end of shift.
Snap Fill is positioned as the hero entry point with "Fastest way to log a shift" copy — but the manual flow is equally complete. The feature signals ambition while the manual flow handles reality.
Early prototype testing of a history-only dashboard scored low on motivation to return. When Predicted Month and Predicted Year cards were added — even with placeholder data — perceived utility increased noticeably among qualitative participants. Workers responded to future-facing data more than past-facing data, consistent across both workers with savings goals and those without.
The home dashboard leads with Today's Earnings and Your Outlook (predicted month + year) above recent shift history. This makes the promise of the app legible even with only a few shifts logged.
Multi-job context switching was the biggest friction point in research. Workers didn't want to type — so the first screen surfaces saved jobs as a single-tap list. Forcing a job selection at the start means every shift entry is contextualized by employer from the beginning, making per-job analytics compound over time.
Research confirmed almost nobody logs in real time — most do it the next morning or after a day off. Putting Today, Yesterday, Last Friday, and Last Saturday front-and-center removes the calendar interaction for ~85% of real use cases. The full calendar is still there for edge cases, but it's not the default path.
Native number inputs tested poorly — slow to summon, small targets, no error feedback. A custom numpad with a large live display ($142) creates immediate familiarity for workers who are around cash registers every shift. The prominent display acts as a self-correction mechanism. A persistent Clear action handles fat-finger moments without breaking flow.
After a worker logs their shift, the confirmation screen does more than confirm the save. It shows the estimated total, a breakdown of hours and tips, and a contextual insight — how this shift compares to their typical day. The goal is to make the act of logging feel rewarding, not administrative. Workers who feel good about logging are more likely to do it again.
A real audit of seven products across the App Store revealed a consistent gap: every competitor misses at least one table-stakes feature for multi-job tipped workers.
| Product | Multi-Job | Cash / Card Split | Backward Logging | Sub-60s Entry | Projections |
|---|---|---|---|---|---|
| ServerLife | ◐ Paid | ◐ Paid | ✓ | ◐ Partial | ✗ |
| TipTracker | ◐ Paid | ◐ Partial | ✓ | ✓ | ✗ |
| Shift App | ◐ Paid | ✓ | ✓ | ✓ | ◐ Partial |
| Balance | ✓ | ✓ | ✓ | ◐ Partial | ✓ |
| TipSee | ◐ Paid | ✗ | ✓ | ✓ | ✗ |
| YNAB | ✗ | ✗ | ✓ | ✗ | ✗ |
| Square POS | ✗ | ◐ Partial | ✗ | ✗ | ✗ |
| Shift Buddy | ✓ Free | ✓ | ✓ | ✓ Target | ✓ |
◐ = Available at paid tier only · ✓ = Fully supported · ✗ = Not available
The UX bar for AI is high: it should reduce friction, not add it. Here's where it clears that bar in Shift Buddy — and where it doesn't.
Photo a clock-out slip and auto-populate hours, card tips, and pay. Compresses a 60-second flow to under 10 seconds. 11 of 14 research participants called it their most-wanted feature. No competitor offers this.
The home screen already shows predicted monthly and yearly earnings. These projections need to be real — extrapolated from logged patterns, adjusted for day-of-week and venue trends. Research showed projections motivate return visits more than history does.
"Your effective hourly rate at Martin's is 24% higher than at Rusty's. You worked 3 more hours at Rusty's this month." Workers want answers, not charts. An AI layer converts data into plain-language conclusions that most users can actually act on.
Replacing structured UI with a chat interface would be a regression. The logging flow needs to be completable at 11:30pm by a tired person. A guided flow with large tap targets beats text conversation every time for this use case.
The core logging loop is prototype-ready. The app is approximately 25–30% designed end-to-end.
The confirmation screen shows estimated total earnings, a per-shift breakdown, and a motivational insight comparing performance to the user's typical average for that day.
Log Shift flow — Select Job → Select Date → Card Tips → Cash Tips. Full guided sequence with custom numpad.
Home Dashboard — Income outlook, shift history preview, projected earnings display.
Login / Auth screens — Full auth patterns including SSO options.
Navigation system — Home, Insights, Log Shift (center-elevated), Goals, Budget.
Onboarding — No account creation, first job setup, or goal-setting flow. Most blocking gap to a testable v1.
Snap Fill / Receipt Scan — Partially designed. Camera state, auto-populated review screen, and OCR degradation path still open.
Insights, Goals, Budget — All three secondary nav sections are zero-to-one. These are what make the app worth keeping long-term.
Empty states & error states — No error states exist across any screen. First-run experience is undefined beyond the $0.00 dashboard.
Shift Buddy started as a genuine frustration with the tools that exist — nothing was designed for how tipped workers actually think about money. Building it reinforced something I keep coming back to: the most important design decision is often the order of operations. Job selection before date before amount isn't just logical — it's the sequence that matches how workers mentally reconstruct a shift after the fact.
The competitive audit confirmed that the gap is real. No product in the space nails all five dimensions simultaneously in a free tier. That's not a small detail for a worker who can't predict their income month-to-month.
The AI framing was intentional from the start: AI should earn its place at specific friction points, not be a feature announcement. Snap Fill, projections, and plain-language Insights each solve a named problem. That discipline — resisting the urge to add AI everywhere — is something I'd carry into any product team.