Redesigning Fig's loan origination flow — Qinglan Zheng
qinglan.zheng
← all work About
21.4% → 44.1% conversion FIG · LOAN ORIGINATION · 2026

Redesigning Fig's loan origination flow

Doubling pre-qual conversion by rebuilding a desktop-era application into a mobile-first, step-by-step journey.

ROLE
Senior product designer — end-to-end owner
PLATFORM
Responsive web, mobile-first
SCOPE
Full origination journey — pre-qual to booking
21.4% → 44.1%
Pre-qual completion — the funnel's biggest leak
79.9% → 91.1%
Flinks bank-connection completion after the splash redesign
80–90%
Of applicants on mobile, now the flow's real primary surface
The redesigned origination flow across four steps: pre-qualified offer, customize offer, confirm payment schedule, and booking confirmation
01 · Context

The flow is the entire business

Fig, a Fairstone Bank company, offers fully digital personal loans up to $35,000 in Canada. The origination flow — 14 steps from "how much do you want to borrow?" to "your loan is booked," gated by a soft credit check at the offer and a hard check at booking — is the product. It had grown organically over time and was showing its age. My mandate was to redesign it end-to-end.

02 · The problem

A desktop-era form, on a mobile-only user base

A scary consent page came first. Before users saw a single form field, a dense biometric/data-sharing consent screen asked for agreement — 20% dropped off right there.

Then a wall of a form. One long page asked for everything at once — name, DOB, phone, email, income, SIN, address, housing costs — on a desktop two-column layout with no progress indicator, pinched through on mobile by 80–90% of users. Combined, only 21.4% of the 4,608 who viewed it completed it.

BEFORE · ONE LONG FORM
Original pre-qual form: a scary consent page upfront, then a single long scrolling page with Personal details, Residential address, and Loan details sections all on one screen
20% drop-off at consent, 80% drop-off on the long form that followed
03 · Research & discovery

Finding where the funnel leaked, and why

The redesign goal was to increase conversion. To understand where and why the funnel was leaking, we ran a mixed-method discovery phase:

01
Team discussions — mapped the funnel end-to-end with product, engineering, risk, and CX.
02
User testing — moderated sessions on the existing flow to watch where people hesitated, misread, or quit.
03
Funnel data analysis — step-by-step performance of the current funnel to pinpoint where drop-off concentrated.
04
Competitor review — how other digital lenders structure application, offer, and verification steps.
05
Secondary research — the form-design literature consistently shows multi-step forms outperform long single-page forms.
Team whiteboarding session mapping the funnel
Team discussion around the funnel redesign
04 · Design principles

Five rules that shaped every screen

{{ $index }}
{{ p.title }} — {{ p.body }}
05 · Key decisions — 01

One long form becomes five focused steps

The single form became five mobile-first steps with a progress bar and thumb-sized inputs. Email and phone moved earlier, unlocking drop-off recovery outreach for most of the funnel.

Consent moved from one dense page upfront into smaller asks in context. Biometric consent now appears only right before identity verification, and only for Quebec users, where it's legally required.

Every screen reassures "This will not impact your credit score." Employment and loan purpose moved into pre-qual, so ineligible users are declined in the first two minutes instead of after identity verification.

AFTER · FIVE MOBILE STEPS
New pre-qual flow — five mobile steps
21.4% → 44.1%
Pre-qual completion, roughly doubled
1 of 5
Decisions per screen, down from one long form
Step 2
Earliest point email is captured — unlocking outreach
2m 16s → 2m 5s
Median time to convert — despite four added fields for risk
3.7% → 0.6%
KYC PII mismatch, after splitting date of birth into three fields
2 min
Time to decline ineligible users, down from after ID verification
06 · Key decisions — across the flow

Four more moments, redesigned to reduce doubt

01

An offer page that closes the loop

Aggregator traffic arrives cold, comparing offers, having never heard of Fig. So the page introduces the company, leads with the numbers users weigh, and pre-answers the questions flooding CX. It isn't a receipt — it's the pitch.

Offer page redesign — before and after
02

Customize your offer, payment-first

The old screen led with term and buried the number that drives the decision. Now monthly payment is the hero, updating live as the sliders move, with APR locked from pre-qual — closing a trust bug where the quoted rate could drift between screens.

Customize offer redesign — before and after
03

A bank connection worth trusting

Connecting a bank account is the scariest ask in the flow. The splash was rebuilt around the two real objections — credentials never touch Fig, and Flinks is used by 500k+ people monthly — plus a plain answer to "why do we need this?"

Bank connection splash page redesign — before and after, 79.9% to 91.1% completion
04

A decline that doesn't punish

Previously, a decline showed a generic error, and users assumed the worst about their credit. The redesign leads with "your credit score is unaffected," then explains what happened and gives one clear recovery action.

Decline page redesign — before and after
07 · Outcomes

A flow built for the phone it's actually used on

21.4% → 44.1%
Pre-qual completion rate
79.9% → 91.1%
Flinks bank connection completion
2m 16s → 2m 5s
Median time to convert
Fig flow book rate — monthly trend
October 2025 – June 2026, approved applications
Affiliate Embedded
35% 40% 45% 50% 55% Oct Nov Dec Jan Feb Mar Apr May Jun

Book rate climbed on both channels across the nine months following launch — roughly a 30% relative improvement on each. The lift held through every month rather than spiking once, which is what separates a flow change from a traffic change.

Splitting the form roughly doubled pre-qual conversion — the funnel's single biggest leak. Clarifying the offer, selection, connection, and error moments removed reasons to doubt, call, or quit, on the surface where most applicants actually were. Because pre-qual conversion compounds downstream, the lift carries through to bookings, not just completions.

08 · What I learned

Trust is a conversion lever

The single highest-leverage copy in the flow is one repeated line: "This will not impact your credit score." Reassurance at the moment of anxiety outperformed any layout change.

Sequencing is design. Moving employment, purpose, email, and phone capture earlier changed business outcomes without changing a single field's UI.

CX tickets are a free research pipeline — every top theme became an in-context answer on the screen where it came up.

09 · What's next

Still refining, screen by screen

The flow keeps evolving from what SMS, support tickets, and live chat surface next. Using the Design Playground, I can prototype and ship refinements against production data without waiting on a full release cycle. Click through both prototypes below, step by step, to compare the current flow against the work in progress.

CURRENT · LIVE FLOW
{{ v1Label }}
{{ v1Label }}
IN PROGRESS · REDESIGN 2.0
{{ v2Label }}
{{ v2Label }}
{{ lightboxAlt }}