Case Study
Budget Buddy
An SMS-based budgeting service.
Role
Focus
Tools
Too long; didn't read
The Problem
80.5% of single-parent households in the US are headed by women. 31% live below the poverty line.
Despite having the most to gain from a personal budget, this group is the least served by the $1.5B personal finance app market. The market is built around assumptions of smartphone access, baseline financial literacy, and discretionary time to learn new software.
The budgeting space is saturated, but saturation isn't the same as solved. Every major app shares the same design premise: the user already knows what a budget is, already owns a capable device, and just needs a better interface. For single parents living paycheck to paycheck, none of those premises hold.
What was missing wasn't a better budgeting app.
It was a service that met users where they actually were.
The Hypotheses
1. Users lack financial literacy — the solution is education and guidance.
2. Users lack a system that fits their real lives — the solution is removing friction, not adding content.
The distinction between the two mattered because it would determine whether I was building a teaching tool or a service tool. These require fundamentally different design approaches.
The Interviews
I conducted 6 semi-structured interviews with people who did not currently maintain a budget. I needed to understand the non-starters, not the converted.
Every participant understood what a budget was; not one cited lack of knowledge as the reason they hadn't started. Hypothesis 1 was wrong. What emerged instead were three interlocking themes.
Fear of failure
The paycheck-to-paycheck trap
Accountability without judgement
These aren't usability problems, they're emotional and logistical ones.
The design response had to address both.
The Solution
The most important insight from research wasn't what users wanted from a budgeting tool, it was that the interface itself was the barrier. Apps require learning. They require opening. They require remembering. For users already stretched thin, that overhead was enough to kill adoption before it started.
The solution that checked every box wasn't a better app. It was no app at all.
Budget Buddy is a personal budgeting service delivered entirely over SMS. After a one-time web sign-up, Budget Buddy guides users through building their first budget, prompts them to categorize transactions, sends check-ins, and celebrates milestones — all through standard text messages, on any phone.
Why SMS specifically
Works on any mobile phone, including non-smartphones — critical for a demographic where 31% live below the poverty line.
Meets users in an interface they already use dozens of times per day.
Eliminates the complex UI that 60% of interviewed users cited as a past barrier to budgeting apps.
Conversation Design
Giving Budget Buddy a voice
Without the right personality, the service wouldn't work. Financial conversations carry real emotional weight, especially for users who've experienced financial stress or failure. The personality had to be warm without being patronizing, expert without being intimidating, and consistent enough to feel trustworthy over weeks of interaction.
Designing for conversation repair
One of the core challenges in SMS-based interfaces is that conversation can break in ways a graphical UI never encounters. A user can type anything. They can ignore a prompt. They can respond hours later. The system has to be resilient to all of it.
I mapped repair patterns — the mechanisms Budget Buddy uses when a conversation goes off the expected path — as a first-class design deliverable alongside the happy path flows. These included:
Testing
Wizard of Oz
I used Wizard of Oz testing to simulate the live service. Participants were instructed to "think out loud" as they signed up for and interacted with what they believed was an automated system while I manually populated each "system" response in real time based on the conversation design flows.
This method gave me direct observational data on how users interpreted Budget Buddy's prompts — not just what they said, but what they thought, where they paused, hesitated, or typed a response and then deleted it.
