Penni

— Generative research end-to-end: screener, consent, interview guide, and facilitation that builds safety before dollars — because shame is a research confound, not a “soft skill.”

— Curriculum + content IA: which concepts unlock the next (50/30/20 before compound interest), jargon protocol (define before use), and the interactive compound slider that became the highest-engagement moment in V2.

— RITE usability: prioritized fixes between P3 and P4, validated in-session behavior shifts instead of treating usability as a report card.

The market optimizes for people who already speak “APR.” Students get the bill first.

Young adults show the lowest measured financial literacy of any cohort — yet colleges assume fluency in tuition, credit, rent, and debt servicing. Budgeting apps assume habit formation; textbooks assume attention. The gap is a learning surface that meets students in emotional reality (anxiety, jargon fog) and still respects their intelligence.

My lens is not hypothetical: three years as a revenue accountant at Lionsgate ($175M quarterly contract scope) taught me how opaque systems feel from the outside — and how much clarity reduces costly mistakes. Penni is the translation layer I wished existed before I sat in those spreadsheets.

A study plan is a design deliverable — it’s how you pre-commit to valid evidence.

Before interviews, the team locked a Final Study Plan: research questions tied to product risks, recruitment criteria (literacy extremes to sharpen contrast), session structure, and how notes would map to design decisions. That plan is what lets a hiring manager trace “why these seven people” instead of seeing a convenience sample dressed in quotes.

The screener + short survey established a literacy baseline; interviews were segmented so we could compare language barriers vs. motivation barriers without conflating them. Warm-up questions were intentional — finance interviews fail when you jump straight to numbers before trust exists.

Jargon isn’t a copy problem — it’s a trust signal that the product wasn’t built for you.

Seven generative interviews plus twelve survey responses. Sharp pattern: even self-described “literate” participants hit undefined terms and confidence collapsed. People also described learning finance through expensive mistakes — overdrafts, compounding card balances — because no safe rehearsal space existed.

V1 proved the lesson architecture; V2 proved we could teach without hiding interactivity.

V1 (see V1.pdf) validated topic flow and mascot presence but surfaced discoverability issues: buttons read as decoration, progress felt invisible, and the emotional “win” at lesson end was weak.
V2 (see Penni_App (YC).pdf) introduced clearer affordances, explicit practice beats (“Let’s practice!”), coin celebration on completion, and the compound-interest slider interaction that participants physically played with in the lab.

Penni pig (original illustration) is not decoration — it marks safe completion states and gives the brand a face when topics feel intimidating.

Content design is UX: sequence, metaphor, and cognitive load are interface decisions.

Lesson 1 anchors budgeting with 50/30/20 in plain language plus a pie visualization — users need a map of money before compounding means anything emotionally. Lesson 2 tackles compound interest with metaphors (“planting seeds”) before naming the term, then an interactive slider manipulating years and contribution so the curve isn’t abstract algebra.

Fix between sessions; use the next participant as a regression test.

Six college students (18–25). Participants 1–3 on V1 → prioritized fixes → participants 4–6 on V2. RITE rewards ruthless triage: if a defect blocks learning, it ships ahead of nice-to-have visual polish.

Success looks like demand for harder lessons — not just smiles on exit.

V2 participants asked for taxes, investing, credit, and loans unprompted — the behavioral signal that fundamentals landed. Multiple students said they wished they’d had the compound slider in high school. That’s the kind of qualitative “metric” that predicts retention better than a five-star mood alone.

Next study: watch people fail in the wild, not only in the interview.

Generative interviews nailed emotional barriers, but I’d supplement with contextual inquiry (e.g., observing someone try to dispute a fee or read a loan estimate) to capture task-based literacy gaps. I’d also prototype the coin onboarding earlier — invisible game mechanics are still mechanics.