the case for

A Fin-Lit Education App

Personalized learning, powered by AI

Built on real data from the National Financial Capability Study and the Survey of Consumer Finances. Every claim here comes from what households report.

01 / The case for fin-lit

The need is big and growing

Most U.S. adults are under real financial strain, and the data shows it getting worse, not better. The financial education they do get is generic, the same lessons for everyone, so it rarely fits the person in front of it. Targeted financial literacy, tuned to a person's income range, state, community, language, family, and immigration status, barely exists today. This is the case for building it, education grounded in what people report about their own lives, and shaped to actually reach them.

55%
struggled to pay their bills in 2024
52%
have no 3 month emergency fund, up since 2021
4 in 5
have never even taken a financial literacy course

Source: NFCS, weighted to the U.S. population.

The bigger picture

These figures come from outside our own analysis. In a 2024 survey, U.S. adults said that not understanding money cost them about $1,015 each that year, roughly $243 billion across the country (National Financial Educators Council). At the same time, money apps are one of the fastest growing corners of tech, one forecast puts the market near $208 billion by 2026, growing about 25% a year (Business Research Insights). A large and worsening need, meeting a fast growing market.

02 / The problem

One size fits all keeps failing

Most money apps teach everyone the same way, no matter their income, habits, or where they live. But a lesson that ignores your situation is easy to ignore back. Three points of failure emerge:

Generic content

The same lessons for everyone, ignoring income, habits, and goals.

User fatigue

Too much complexity. People tune out and decide worse.

Context gap

Lessons ignore where someone lives and what it costs.

03 / The Data

Segments the data can prove

This runs on 15 years of respondent data from the National Financial Capability Study (NFCS), over 50,000 rows cleaned and normalized in Python and Pandas. That evidence points to real groups we can identify, by income, community, and place.

Income

Lower income families

Under the most financial pressure, and their outlook has swung the hardest since 2021.

See 04, The strain
Life stage and place

Recent graduates in high strain states

Young, degree-holding adults living in states where money troubles are most common.

See 06, The targeting
Community

Minority communities

Non-White households get offered financial education more often, and say yes to it more too.

See 05, The appetite
Frontier

New immigrant families

There's no public data on this group yet. They'd be the first we add once there is.

See 07, Where this goes next
04 / The strain

Satisfaction down, vulnerability up, worst at the bottom

People felt best about their finances in 2021, thanks to pandemic relief, then that feeling dropped off by 2024. Financial vulnerability moved the opposite way: down in 2021, back up by 2024. Both scores start at the same point (100) in 2009, so they're easy to compare on one chart. Solid lines (satisfaction) trend down, dashed lines (vulnerability) trend up. The lower income band swings the most.

Strain Index Chart
Lower income Middle income Higher income satisfaction vulnerability

Time trends charted with ECharts.

05 / The appetite

Non-White households want it more

Everyone falls into one of three groups: offered financial education and took it, offered it and skipped it, or never offered it at all. Non-White households are offered financial education more often, and say yes to it more often too; they're more engaged and less often left out. The one place White households come out ahead is among people who were already offered it, more of them actually took it, about 74% versus 66%. But far fewer White households are offered it in the first place, so overall, the appetite for it is still stronger in minority communities. That's the audience a targeted platform should serve first. This is shown for 2024, but the same pattern holds all the way back to 2012.

Appetite Demographics Chart
Took it Offered, skipped Never offered

Stacked bars charted with ECharts.

06 / The targeting

Where vulnerability lives

Each state is a 3D block. Color is financial vulnerability (darker is worse, averaging five hardships: bills, emergency fund, overspending, overdrafts, predatory borrowing). Height is average income (taller is higher). A tall dark block means high income but high strain; a short dark block is the poorer, struggling pattern. Drag to rotate, scroll to zoom, hover for the detail.

Vulnerability Map
Income band Color by 2024 Try dragging the timeline
Hover over states to see the breakdown

Utah, Alaska, and North Dakota all out-earn the national average, and land above it on vulnerability too. Income and financial safety don't move together the way you'd expect.

3D map built with Deck.gl.

07 / Where this goes next

Beyond broad groups, into a real product

The race categories here are broad. A single "Hispanic" or "Asian" label can hide very different communities inside it, each with different income and homeownership patterns.

  • Hispanic Mexican-origin, Puerto Rican, Cuban, Central American, South American, Dominican
  • Asian Chinese, Indian, Filipino, Vietnamese, Korean, other Asian origins
  • Black U.S.-born multi-generational, Caribbean immigrant, African immigrant
  • White Multi-generational U.S.-born, recent European immigrant, MENA origin
The data gap

How recently someone immigrated shapes money habits more than a broad race label does. The catch: our survey never asks where someone was born. Pulling it from the Census or the Fed means merging datasets whose geographic boundaries and survey weights don't line up.

The fix is to ask directly. Onboarding already builds a profile, so it can collect country of origin and years in the U.S. first-hand. Until that pool is large enough, Census figures guide what we build, but they don't back claims about any specific group.

The spread inside one label can be extreme, not just varied. In NYC, Asian American households with less than a high school education report a starkly lower median income than those with a graduate degree. This is a gap wide enough that the group's own average sits between two real populations, representing neither. The lower end gets less support than a Black or Hispanic household at the same income, because it's statistically absorbed into a higher group average. Any segment in this model risks the same failure if it resolves to the higher all-encompassing label instead of the sub-groups underneath it. Which is why onboarding collects income, education, and place directly rather than inferring from a demographic checkbox.

What we'd build

The product turns the evidence into a path for each user. No generic videos, a profile built from real signals.

How AI powers this

Every step above runs on the same model. Quiz answers and later signals, like a new job, a move, a maxed card, become a profile; the model matches that profile against segment patterns to assemble a roadmap and writes it in plain language instead of policy jargon.

A rules engine can sort someone into a bucket. It can't read "$400 emergency, no savings, picking up more shifts" and know that means both an emergency-fund lesson and a gig-income tax lesson. That's the personalization a static decision tree can't do, which AI helps with.

1
Onboarding
2
Profile
3
Roadmap
4
Adaptive lessons

Click a step to walk through it

MVP Scope

Gamification, credentialing, social features can wait. The first release only has to prove that a lesson tuned to someone's income, state, and life stage lands differently than generic advice.

Rather than build the loop for all four segments in 03 at once, the first release proves it on one: Lower income families. This group has the deepest NFCS data, and it's where personalized lessons should show the biggest lift over generic advice (see 04, The strain). Once that shows a real difference, the same pipeline extends to recent graduates in high-strain states and minority communities, with new immigrant families added once the onboarding data described above is large enough.

The MVP is scoped down to just the personalization loop:

  • Secure user profiles
  • Literacy and behavior assessment
  • AI personalized learning path
  • Local context by region