Case Study — Fintech
Rhythm
A personal finance tool for gig workers that replaces the monthly budget with a single daily-updating "Pulse" — designed from my own lived experience of zero-hour hospitality.
The Problem
Over 1.1 million people in the UK work on zero-hours contracts, and hospitality leans on them harder than any other sector — close to a third of its workforce (ONS Labour Force Survey, 2024–25). Their income swings from week to week, yet mainstream budgeting apps still assume a stable monthly salary — punishing them with red alerts during lean weeks and ignoring the feast weeks that keep them afloat.
The "Monthly" Myth
Gig workers are paid daily or weekly. A monthly reset is meaningless when income fluctuates constantly.
The Guilt Cycle
Apps use red "overspending" warnings during famine weeks, ignoring the feast week that just passed.
The Tax Trap
The self-employed workers I spoke with described spending gross income, then scrambling for a Self-Assessment bill each January.
Discovery & Research
I ran semi-structured interviews with four variable-income hospitality workers in London — a bartender, an event supervisor, an agency waitress, and an event-setup worker — recruited through my own hospitality network. As someone who has worked zero-hour hospitality myself, I also drew on that lived experience throughout, keeping it visible as both a source of insight and a source of bias. This is generative research: the goal was to frame the problem, not to validate it statistically.
"It uses a monthly average, so it tells me I've overspent when I haven't. It makes me feel broke when I'm really just waiting to get paid."
"January came and my Self-Assessment bill was £1,400. I had nothing left. I'm still on a payment plan. It's humiliating."
What the Research Surfaced
Six themes emerged across the conversations. Three are captured above; the four below reframed the design space and ruled several "obvious" solutions out.
They already run the algorithm
Workers mentally assign specific shifts to specific bills — "Tuesday's shift is the phone bill." The product's job is to carry that load, not teach a new habit.
Cash income is invisible
Cash tips and last-minute shifts don't reach the bank feed in time. Any tool relying on Open Banking alone misses a real slice of what people earn.
Timing, not just amount
Sometimes the crisis isn't too little money overall — it's money arriving after rent is due. The tool has to model the timeline, not only the total.
Distrust of bank linking
For some, manual entry isn't a fallback — it's a requirement. Forcing a bank connection would exclude exactly the people this is for.
Who I Designed For
Two composite personas — built from the interviews, not single individuals — kept the work focused.
Composite persona
The Reshuffler
- Snapshot
- Urban, zero-hour or agency hospitality. Mentally maps shifts to bills; when a shift is cancelled, re-plans the whole week in her head.
- Pain
- Budgeting apps blame her for "overspending" in lean weeks because they can't see next week's booked shifts.
- Need
- One current answer to "what is actually safe to spend right now, given what I know is coming?"
- Emotional driver
- Relief from being judged by her own banking app.
Composite persona
The Caught-Out Earner
- Snapshot
- Event and gig work with lumpy, feast-and-famine income. Earns well in peaks, gets caught in the troughs.
- Pain
- Spent gross income, hit by a large January tax bill; also faces payday-vs-bill-date timing crises.
- Need
- Forced, invisible setting-aside for tax, and a clear view of money available between now and next payday.
- Emotional driver
- Avoiding the humiliation of being caught short again.
The Jobs To Be Done
I framed the problem as job stories rather than feature requests, so the design stayed anchored to the moment of need.
- When a booked shift is cancelled at short notice, I want my "safe to spend" figure to update immediately, so I can re-plan without doing the maths in my head.
- When I earn cash or pick up a last-minute shift, I want to log it instantly, so my available balance reflects what I actually have.
- When I get paid, I want a portion automatically set aside for tax before I can spend it, so January doesn't bankrupt me.
- When rent is due before my next payday, I want to see whether I'll have enough in time, so I can act before I'm in crisis.
- When I have a lean week but shifts booked next week, I want the app to connect the two, so I'm not made to feel broke when I'm just waiting to get paid.
- When I don't trust connecting my bank, I want to manage everything by manual entry, so I can use the tool on my own terms.
A Week Without Rhythm
Following The Reshuffler through a week with one cancelled shift. The emotional low is Wednesday — the moment the app actively misinforms and blames. That became the design target.
| Stage | What happens | Feeling | Opportunity |
|---|---|---|---|
| Mon — week planned | Assigns three booked shifts to bills in her head | In control | Capture this mental model in the app |
| Tue 8pm — shift cancelled by text | £100 of expected income vanishes | Anxiety, scramble | Instant re-calculation of safe-to-spend |
| Wed — checks banking app | Still shows "on track" on a monthly average, then later flags "overspent" | Confused, then blamed | Show a real forward-looking position, not a monthly average |
| Thu — lean, but shifts booked next week | App issues red overspend alerts | Panic, despite knowing more work is coming | Connect this week's gap to next week's booked income |
| Fri — borrows from a friend | Covers the gap socially | Embarrassed | Earlier visibility could have prevented this |
The Core Innovation: The "Pulse"
The core design decision was to kill the monthly budget. The Pulse is a single number — your safe-to-spend amount over the next 14 days — recalculated daily from your balance, upcoming shifts, pending bills, and an automatic tax stash.
Live Pulse Demo — adjust the variables
The Pulse is your safe-to-spend amount. Green means you're okay. Red means plan ahead.
From Research to Features
Every feature traces back to a specific finding. Nothing in the build exists because it was a nice idea — it's there because the research demanded it.
| Finding | Feature response |
|---|---|
| The monthly model fails | The Pulse — a rolling 14-day safe-to-spend figure, recalculated daily |
| Manual shift-to-bill allocation | Forecast view placing shifts and bills on one timeline |
| The tax trap | Auto Tax Stash — a set % held from every logged shift |
| Manual income is essential | Quick-Add Shift — one-tap manual income logging |
| Timing crises | Forecast shows the run-up to each bill date, not just monthly totals |
| Distrust of bank connection | Manual-first design; bank connection optional, never required |
Key Design Decisions
Three calls did the most to shape Rhythm. Each one was a trade-off, and each came directly from what the research said — not from what was easiest to build.
Manual-first, not bank-first
Why
Cash tips and last-minute shifts are invisible to Open Banking, and some workers actively distrust bank linking.
Decision
A one-tap Quick-Add Shift makes manual logging the default path; bank connection is optional, never required.
Trade-off
More effort up front, in exchange for accuracy and including the people automation would otherwise lock out.
A forced, invisible tax stash
Why
The tax shock carried the heaviest emotional weight in the interviews, and the ask was specific: money set aside before it's ever seen.
Decision
A set percentage is held from every logged shift, by default, into a separate Tax Stash.
Trade-off
Deliberately paternalistic. I kept it opt-out rather than opt-in, because the whole point was not having to decide each time.
Model the timeline, not the total
Why
Timing crises are distinct from amount crises — pay can arrive after a bill is already due.
Decision
The Forecast shows the run-up to each bill date, so a shortfall is visible before it bites — not buried in a single monthly figure.
Trade-off
A richer view than one number. Mitigated by leading with the Pulse and keeping the timeline one tap away.
Method & Limitations
Being clear about what this research can and can't claim matters more to me than overstating it:
- Four participants. This is generative, problem-framing research — the findings are directional, not statistically representative.
- Recruited through my own hospitality network (convenience sampling), so the group may share more than a random sample would.
- I have lived experience of the problem, which helped rapport and interpretation but is also a source of bias — flagged deliberately rather than hidden.
- Participants are anonymised; quotes are attributed by neutral codes only.
Next Steps
The next phase would involve a closed alpha with 15-20 gig workers, measuring weekly financial stress via a simple self-report scale, and iterating on the Quick-Add flow and tax-stash defaults before wider release.