Splitwise, redesigned

Research

People wanted tips, not just totals.

We surveyed people who track expenses about how often they track, which apps they use (Splitwise, Plum, Monzo, Revolut), what they like, what annoys them, and the one thing they wish their app could do.

Tagging the answers in Dovetail and grouping them in an affinity map gave one clear theme: users want tips to reduce spending and a forecast of what’s coming up.

A side lesson: people answered closed questions far more willingly than open ones, which shaped how I write surveys now.

How might we motivate people to save more money and build good habits with it?

Audit

Where the current app gets in the way.

• Cramped navigation. Too many options, a “More” button in the tab bar, and group settings in the top bar.

• Inconsistent UI. Button styles, heading colours and placement change from screen to screen.

• Copy that works against the app. “Don’t invite Ivan” in an app about splitting costs. “Invite Ivan later” says the same thing without the friction.

• Missing feedback. No confirmation after creating a group, and onboarding with no welcome.

Design

Adding an expense, one decision per screen.

The new flow breaks a crowded form into short steps: photo or receipt, amount and currency, description, category, date and repeat, then how to split it.

• Categories with favourites. Icons grouped by Entertainment, Food and drink and Home, with a heart to pin the ones you use most.

• Repeat frequency. Once, weekly, fortnightly, monthly or yearly, so rent and bills forecast themselves.

• Clear split. Paid by and split between, each person with an amount you can see and remove.

Forecasting

See where it goes. See what’s coming.

When you add an expense, the app compares it with your average for that category. Analytics screens show monthly spend per category, weekly, monthly or yearly trends, and upcoming expenses based on what repeats.

If an expense is larger than usual, a weekly review nudges you, so the next choice is a more deliberate one. You can also set a budget per category, like rent.

Reflection

My first project, and where I learned to tag before I design.

Not every tag is useful, and some only appear once you’re deep in the responses. Setting up a tagging taxonomy before synthesis made the affinity map faster and more honest.

Looking back, I’d test the forecasting screens with real numbers from participants, since the value of a forecast depends on whether people trust it.