Predictive banking for freelancers with variable income

The problem is structural, not behavioural

Freelancers with variable income rely on banking interfaces that display a total account balance but do not reflect future obligations — taxes, recurring expenses, the buffer they are trying to protect.

Current banking systems visualise present money but hide future commitments. That creates a decision gap between what someone sees and what they actually have available.

It is tempting to read this as a discipline problem — people should just budget better. The research said otherwise. The disciplined freelancer I spoke to was already doing the work manually, in spreadsheets, and still did not trust the answer.

Who this is for

Empathy map — The empathy map. What kept recurring was not carelessness — it was the mental effort of doing the bank's arithmetic by hand, and never being sure it was right.

Two personas, built from the research, sitting at opposite ends of the same problem.

The two personas — Elena manages her money carefully and still feels uncertain. Marco reads the balance and assumes it is his. They need the same information for opposite reasons.

Elena Müller, 34, a freelance UX designer in Berlin, five years self-employed. She reserves part of every payment for tax and still hesitates before a large purchase:

I try to stay disciplined, but I'm never completely sure what's actually safe to spend.

Marco Rossi, 29, an independent marketing consultant in Milan, two years self-employed, no structured tracking at all:

If the money is in my account, I assume I can use it — but sometimes that backfires.

Elena's problem is cognitive load. Marco's is visibility. A design that solves only one of them fails the other.

The journey map — Mapping the decision as it actually happens: identify a need, check the balance, validate the maths somewhere else, decide, then monitor the consequence. The uncertainty never resolves — it just moves.

Define — the statements the design had to answer

Problem statement, Elena. A structured freelancer with variable income who actively manages her finances needs a clear way to confidently determine how much money is safe to spend, because she currently relies on manual calculations and separate tools which create mental effort and ongoing uncertainty about future tax obligations.

Problem statement, Marco. A reactive freelancer with irregular income who relies on his visible account balance needs a clear way to understand how much money is truly available, because his balance does not reflect upcoming tax and recurring obligations — which leads to uncertainty and financial risk.

The hypothesis. If a freelancer is shown clear visibility of committed funds and future obligations alongside the current balance, then they will make more informed spending decisions and reduce the likelihood of shortfalls.

The value proposition, deliberately written twice, because the same feature has to be sold to two different mindsets:

For structured freelancers who track taxes manually — transparent visibility into committed funds and safe-to-spend amounts, reducing cognitive load and increasing confidence without removing control.
For freelancers who rely on the visible balance — a clear line between total liquidity and committed obligations, preventing avoidable shortfalls.

Ideate — where the market actually leaves a gap

Competitive audit — Deutsche Bank, N26, Revolut Business and Kontist, scored across desktop UX, mobile UX, accessibility and user flow.

The audit's conclusion was narrow and useful:

Every competitor prioritises current balance.
Future obligations are either hidden or pushed into an external tool.
Tax tools exist — Kontist, for instance — but they are narrow, and they sit beside the bank rather than inside the decision.
No product integrates obligation visibility into the moment money is spent.

That last line is the whole opportunity, and it is why the design work concentrated on the transfer flow rather than the dashboard.

Crazy 8 concepts — Eight concepts in eight minutes: dual-balance dashboard, dynamic tax bucket, timeline-first view, transaction impact modal, financial health indicator, quarterly projection, smart allocation toggle, spending simulation.

Four of the eight survived into the design, chosen because together they solve the cognitive gap, restructure the decision logic and increase transparency — while keeping the user in control and avoiding feature overload.

Design — architecture before screens

Site map — The information architecture. Committed Funds is a first-class destination rather than a settings page, because the breakdown is the product's actual argument.

User flow — The transfer flow. The impact preview sits between entering an amount and confirming it — the only place an intervention can change a decision rather than report on one.

Low-fidelity wireframes — Lo-fi: dashboard, send money, review transaction, committed funds.

The low-fidelity prototype, running. This is the build that went into testing — every finding below came from watching five people use it.

Testing changed the design twice

Five participants, tested on the low-fidelity prototype. Two changes came directly out of it, and both are visible as before-and-after below.

Dashboard iteration — Before and after on the dashboard. "Risk: Low" became a state-based buffer health signal, and the next obligation — quarterly tax, due in 12 days — was surfaced on the home screen instead of living inside a menu.

Reframing risk. The first version labelled the state "Risk: Low", which tested badly — participants could not tell whether that was a judgement about them or about the account. Restating it as a buffer health signal removed the ambiguity.

Surfacing the next obligation. Putting the upcoming tax deadline on the dashboard was the single change that most reduced surprise. Visible deadlines turned an abstract worry into a date.

Transfer review iteration — Before and after on the review step. The 30-day cash outlook moved into the confirmation screen — the moment where seeing a projected dip can still change the decision.

The 30-day cash outlook is the piece I would keep if I could keep only one. In testing, 4 of 5 participants reviewed the projection before confirming, and 3 of 5 changed their transaction after seeing the dip near the tax deadline.

The visualisation turned an abstract calculation into a visible consequence.

Hi-fi annotations — Two hi-fi states that came out of testing: an explicit transaction-successful screen, and the detailed cash outlook. 5 of 5 participants expected visible confirmation after sending money — the concept did not have one until they said so.

The solution

The finished interface end to end — dashboard, committed funds, the transfer flow and the intervention state.

The dashboard — Safe-to-Spend as the primary metric, with total balance demoted to a supporting line.

Safe-to-Spend as the primary metric. Instead of centring the account balance, the interface leads with what is genuinely available once tax, recurring costs and buffer are set aside. The total balance is still there — it is just no longer the number that makes the decision.

Transfer flow — The transfer flow: enter an amount, see the impact before confirming, get explicit closure afterwards.

Intervention and transparency — A contextual shortfall alert, and the Committed Funds breakdown behind it. The tax rate is adjustable, so Safe-to-Spend can be understood rather than accepted as a black box.

Intervention without restriction. When a transfer would push someone below an upcoming obligation, the system says so — with the number and the date — and then offers Adjust Amount or Continue Anyway. It never blocks the transaction. Autonomy was a constraint, not a nice-to-have: a banking app that refuses your own money is a banking app you stop trusting.

Structural transparency. The Committed Funds screen breaks the total into Safe-to-Spend, tax allocation, recurring expenses and buffer, with an adjustable tax rate. Users trusted the system more when they could see why a number changed.

The design system

Design system foundations — Four principles — trust first, clarity over complexity, proactive guidance, supportive rather than alarmist — and the colour system built on them.

Component library — Buttons across four styles and four states, inputs, navigation, toggles, modals, toasts and the card system.

The principle that did the most work was supportive, not alarmist. A product that tells a freelancer they are about to be short of money is one bad tone away from being deleted. Every warning state uses encouraging language and soft visual cues; none of them shame.

What I took away

Financial products are not about displaying data — they are about shaping decisions. Safe-to-Spend only became credible once it was tied to time, obligations and consequences.

Transparency builds trust more reliably than simplification. Breaking down committed funds and exposing the tax logic increased perceived credibility more than any amount of visual tidying. Clarity in financial products is not about showing less; it is about showing the structure.

Projection reduces anxiety. When people saw future dips before they happened, the surprise went out of it. Foresight turned uncertainty into confidence.

Scope, honestly

This is a self-initiated concept. It integrates no real financial data, it has had no long-term validation, and the usability study had five participants — which is why every finding above is written as a fraction rather than a percentage.

What it demonstrates is the reasoning: system thinking, behavioural insight, and a design argument that survived contact with people who were not me.

The visual presentation of this project is also on Behance.