Case Study — Confer Inc.
Case Study — Confer Inc.
How I Turned 45 - Minute Mortgage Confusion Into 5-
Minute Clarity
How I Turned 45 - Minute Mortgage Confusion Into 5-
Minute Clarity
Alex, a first-time homebuyer, was about to sign the first loan offer he got. 47% of borrowers do the same thing. I designed a mobile tool that helped people like Alex compare 3 loan estimates side by side and save $3,000+ in under 5 minutes.
Alex, a first-time homebuyer, was about to sign the first loan offer he got. 47% of borrowers do the same thing. I designed a mobile tool that helped people like Alex compare 3 loan estimates side by side and save $3,000+ in under 5 minutes.
Product Designer (Solo)
Product Designer (Solo)
2022 - 2023
2022 - 2023
Mobile iOS + Android
Mobile iOS + Android
5 min
5 min
Comparison time
Comparison time
$3K+
$3K+
Avg. savings
Avg. savings
80%
80%
Form completion lift
Form completion lift
My Role
End-to-end UX design
Information architecture
Design system creation
Stakeholder presentations
User research (6 interviews)
Usability testing (4 rounds)
Interaction design
Developer handoff specs
Solo designer on a 4-person team (1 PM, 2 engineers, 1 designer). I owned the entire design process from research through handoff, reporting directly to the CEO. Collaborated with a legal advisor on mortgage disclosure compliance and a backend engineer on PDF parsing constraints.
Solo designer on a 4-person team (1 PM, 2 engineers, 1 designer). I owned the entire design process from research through handoff, reporting directly to the CEO. Collaborated with a legal advisor on mortgage disclosure compliance and a backend engineer on PDF parsing constraints.
The Solution
The Solution
One App. Three Loan Estimates. Five Minutes to Clarity.
One App. Three Loan Estimates. Five Minutes to Clarity.
Here's how Confer works, told through Alex's first session with the app. What used to take 45 confused minutes now takes 5 confident ones.
Here's how Confer works, told through Alex's first session with the app. What used to take 45 confused minutes now takes 5 confident ones.
ACT 1
Discovery: Upload and Extract
The user opens Confer after learning about comparing loan estimates. They upload 3 PDFs from different lenders.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Smart Extraction — Progressive feedback while the app reads your documents. No waiting in silence.
ACT 2
Comparison: See the Differences
The user sees all 3 estimates side by side as summary cards and taps to drill into the details of each one.
Side-by-Side Cards — 4 key metrics per lender. Rate, payment, closing costs, total. Outliers highlighted.
Detail Breakdown — Tap any card to see the full estimate with plain-language jargon explainers.
↓ User found a question about Lender fees
ACT 3
Action: Negotiate with Confidence
Borrowers message lenders directly through the app, referencing specific numbers from the comparison.
Contextual Messaging — Messages linked to specific data points. Tap any number to reference it in your message.
Decision Dashboard — All estimates, conversations, and comparisons in one place. Alex is ready to decide.
Watch Alex Compare 3 Loan Estimates in Under 60 Seconds
ACT 1
Discovery: Upload and Extract
The user opens Confer after learning about comparing loan estimates. They upload 3 PDFs from different lenders.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Smart Extraction — Progressive feedback while the app reads your documents. No waiting in silence.
ACT 2
Comparison: See the Differences
The user sees all 3 estimates side by side as summary cards and taps to drill into the details of each one.
Side-by-Side Cards — 4 key metrics per lender. Rate, payment, closing costs, total. Outliers highlighted.
Detail Breakdown — Tap any card to see the full estimate with plain-language jargon explainers.
↓ User found a question about Lender fees
ACT 3
Action: Negotiate with Confidence
Borrowers message lenders directly through the app, referencing specific numbers from the comparison.
Contextual Messaging — Messages linked to specific data points. Tap any number to reference it in your message.
Decision Dashboard — All estimates, conversations, and comparisons in one place. Alex is ready to decide.
Watch Alex Compare 3 Loan Estimates in Under 60 Seconds
ACT 1
Discovery: Upload and Extract
The user opens Confer after learning about comparing loan estimates. They upload 3 PDFs from different lenders.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
ACT 2
Comparison: See the Differences
The user sees all 3 estimates side by side as summary cards and taps to drill into the details of each one.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
↓ User found a question about Lender fees
ACT 3
Action: Negotiate with Confidence
Borrowers message lenders directly through the app, referencing specific numbers from the comparison.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Alex, 32
First-time homebuyer
Budget: $350K | Saved: $50K
Vancouver, BC
"I got one loan estimate from my bank, felt overwhelmed by the jargon, and almost signed without comparing."
Represents 4 of 6 interview participants who nearly took the first offer
Alex
FRUSTRATED
"I spent 45 minutes flipping between PDF tabs trying to figure out which lender was actually cheaper. I kept second-guessing myself because I didn't know if I was comparing the right numbers. Eventually I just picked the one my realtor recommended."
47%
Don't comparison shop
Nearly half of borrowers take the first offer (CFPB). Not because they're lazy — because comparing feels impossible.
$3,000+
Left on the table
Getting 5 rate quotes saves $3,000+ over the loan's life (Freddie Mac). Most borrowers never get past one.
15+
Line items per estimate
Each loan estimate has 15+ financial jargon line items. No standardized way to compare them, especially on mobile.
The Core Problem
In the current mortgage shopping process, Alex receives 3 loan estimates from
different lenders, but has no way to compare them side by side without manually extracting numbers from dense PDFs. This creates a confidence problem because without easy comparison, borrowers default to the first offer rather than risk making the wrong choice with money they don't understand.
How Might We
Make mortgage comparison as simple and transparent as comparing phone plans, so borrowers like Alex feel confident enough to negotiate?
Alex, 32
First-time homebuyer
Budget: $350K | Saved: $50K
Vancouver, BC
"I got one loan estimate from my bank, felt overwhelmed by the jargon, and almost signed without comparing."
Represents 4 of 6 interview participants who nearly took the first offer
Alex
FRUSTRATED
"I spent 45 minutes flipping between PDF tabs trying to figure out which lender was actually cheaper. I kept second-guessing myself because I didn't know if I was comparing the right numbers. Eventually I just picked the one my realtor recommended."
47%
Don't comparison shop
Nearly half of borrowers take the first offer (CFPB). Not because they're lazy — because comparing feels impossible.
$3,000+
Left on the table
Getting 5 rate quotes saves $3,000+ over the loan's life (Freddie Mac). Most borrowers never get past one.
15+
Line items per estimate
Each loan estimate has 15+ financial jargon line items. No standardized way to compare them, especially on mobile.
The Core Problem
In the current mortgage shopping process, Alex receives 3 loan estimates from
different lenders, but has no way to compare them side by side without manually extracting numbers from dense PDFs. This creates a confidence problem because without easy comparison, borrowers default to the first offer rather than risk making the wrong choice with money they don't understand.
How Might We
Make mortgage comparison as simple and transparent as comparing phone plans, so borrowers like Alex feel confident enough to negotiate?
Alex, 32
First-time homebuyer
Budget: $350K | Saved: $50K
Vancouver, BC
"I got one loan estimate from my bank, felt overwhelmed by the jargon, and almost signed without comparing."
Represents 4 of 6 interview participants who nearly took the first offer
Alex
FRUSTRATED
"I spent 45 minutes flipping between PDF tabs trying to figure out which lender was actually cheaper. I kept second-guessing myself because I didn't know if I was comparing the right numbers. Eventually I just picked the one my realtor recommended."
47%
Don't comparison shop
Nearly half of borrowers take the first offer (CFPB). Not because they're lazy — because comparing feels impossible.
$3,000+
Left on the table
Getting 5 rate quotes saves $3,000+ over the loan's life (Freddie Mac). Most borrowers never get past one.
15+
Line items per estimate
Each loan estimate has 15+ financial jargon line items. No standardized way to compare them, especially on mobile.
The Core Problem
In the current mortgage shopping process, Alex receives 3 loan estimates from
different lenders, but has no way to compare them side by side without manually extracting numbers from dense PDFs. This creates a confidence problem because without easy comparison, borrowers default to the first offer rather than risk making the wrong choice with money they don't understand.
How Might We
Make mortgage comparison as simple and transparent as comparing phone plans, so borrowers like Alex feel confident enough to negotiate?
Constraints
The Realities I Designed Around
Good design doesn't happen in a vacuum. These were the real engineering, legal, and business constraints that shaped every decision.
CEO — Priority alignment
PM — Sprint scoping
Legal advisor — Disclosure compliance
Backend engineer — PDF parsing limits
PDF Parsing Was ~70% Accurate
Backend extraction varied across lender formats. 3 out of 10 fields could be wrong.
→ Built a review-and-confirm step so users catch errors before comparing.
Legal Disclosures Required (TILA-RESPA)
Federal regulations mandate specific disclosures when comparing loan terms. Every label needed legal review.
→ Simplified summary cards on top, full compliant detail via tap-to-expand.
4-Month Timeline, Solo Designer
Investor demo deadline. No other designers. Cut lender marketplace and rate alerts to focus on core comparison.
→ Scoped MVP to 3 flows: upload, compare, message. 4 testing rounds before deadline.
Users Don't Trust Fintech Apps
Trust was the #1 barrier — ahead of usability. First-time buyers feared uploading financial docs to an unknown app.
→ Added encryption badges, "we never see your data" messaging, and a no- upload preview mode.
Constraints
The Realities I Designed Around
Good design doesn't happen in a vacuum. These were the real engineering, legal, and business constraints that shaped every decision.
CEO — Priority alignment
PM — Sprint scoping
Legal advisor — Disclosure compliance
Backend engineer — PDF parsing limits
PDF Parsing Was ~70% Accurate
Backend extraction varied across lender formats. 3 out of 10 fields could be wrong.
→ Built a review-and-confirm step so users catch errors before comparing.
Legal Disclosures Required (TILA-RESPA)
Federal regulations mandate specific disclosures when comparing loan terms. Every label needed legal review.
→ Simplified summary cards on top, full compliant detail via tap-to-expand.
4-Month Timeline, Solo Designer
Investor demo deadline. No other designers. Cut lender marketplace and rate alerts to focus on core comparison.
→ Scoped MVP to 3 flows: upload, compare, message. 4 testing rounds before deadline.
Users Don't Trust Fintech Apps
Trust was the #1 barrier — ahead of usability. First-time buyers feared uploading financial docs to an unknown app.
→ Added encryption badges, "we never see your data" messaging, and a no- upload preview mode.
Constraints
The Realities I Designed Around
Good design doesn't happen in a vacuum. These were the real engineering, legal, and business constraints that shaped every decision.
CEO — Priority alignment
PM — Sprint scoping
Legal advisor — Disclosure compliance
Backend engineer — PDF parsing limits
PDF Parsing Was ~70% Accurate
Backend extraction varied across lender formats. 3 out of 10 fields could be wrong.
→ Built a review-and-confirm step so users catch errors before comparing.
Legal Disclosures Required (TILA-RESPA)
Federal regulations mandate specific disclosures when comparing loan terms. Every label needed legal review.
→ Simplified summary cards on top, full compliant detail via tap-to-expand.
4-Month Timeline, Solo Designer
Investor demo deadline. No other designers. Cut lender marketplace and rate alerts to focus on core comparison.
→ Scoped MVP to 3 flows: upload, compare, message. 4 testing rounds before deadline.
Users Don't Trust Fintech Apps
Trust was the #1 barrier — ahead of usability. First-time buyers feared uploading financial docs to an unknown app.
→ Added encryption badges, "we never see your data" messaging, and a no- upload preview mode.
Research
How CFPB Data + 6 User Interviews
Shaped Every Decision
I combined primary research (6 first-time homebuyer interviews) with secondary research from the CFPB and Freddie Mac. Both pointed to the same insight: the barrier isn't actual complexity. It's perceived complexity.
4/6
"I didn't even know I could
negotiate"
Four out of six participants didn't realize mortgage terms are negotiable. They treated the first estimate as final.
Primary Research — User Interviews, 2022
12 min
The jargon barrier
Participants spent an average of 12
minutes trying to understand a single loan estimate. Most gave up and focused only on monthly payment.
Primary Research — User Interviews, 2022
77%
"Easier than I expected"
Of borrowers who applied to multiple lenders, 77% said the process was less difficult than expected. The barrier isn't the
act. It's the anticipation.
Secondary Research — CFPB Report
Alex
SKEPTICAL
"Nobody told me I could push back on these numbers. I thought the loan estimate was like a price tag — take it or leave it."
CFPB Consumer Survey — 47% of borrowers don't comparison shop. Key data point that framed the entire project.
Freddie Mac Study — $3K+ savings with 5 quotes
Affinity Map — Key themes from 6 user interviews
Journey Map — Pain points cluster at comparison and decision stages
Competitor Analysis — NerdWallet focuses on education, not comparison action
Competitor Analysis— LendingTree leads with quotes, no side-by-side
Key Insight That Shaped Every Decision
The opportunity isn't to educate borrowers about mortgages. It's to make comparison so simple that education becomes unnecessary. If Alex can see "Lender A costs $4,200 more" in 3 seconds, he doesn't need to know what APR means.
Research
How CFPB Data + 6 User Interviews
Shaped Every Decision
I combined primary research (6 first-time homebuyer interviews) with secondary research from the CFPB and Freddie Mac. Both pointed to the same insight: the barrier isn't actual complexity. It's perceived complexity.
4/6
"I didn't even know I could
negotiate"
Four out of six participants didn't realize mortgage terms are negotiable. They treated the first estimate as final.
Primary Research — User Interviews, 2022
12 min
The jargon barrier
Participants spent an average of 12
minutes trying to understand a single loan estimate. Most gave up and focused only on monthly payment.
Primary Research — User Interviews, 2022
77%
"Easier than I expected"
Of borrowers who applied to multiple lenders, 77% said the process was less difficult than expected. The barrier isn't the
act. It's the anticipation.
Secondary Research — CFPB Report
Alex
SKEPTICAL
"Nobody told me I could push back on these numbers. I thought the loan estimate was like a price tag — take it or leave it."
CFPB Consumer Survey — 47% of borrowers don't comparison shop. Key data point that framed the entire project.
Freddie Mac Study — $3K+ savings with 5 quotes
Affinity Map — Key themes from 6 user interviews
Journey Map — Pain points cluster at comparison and decision stages
Competitor Analysis — NerdWallet focuses on education, not comparison action
Competitor Analysis— LendingTree leads with quotes, no side-by-side
Key Insight That Shaped Every Decision
The opportunity isn't to educate borrowers about mortgages. It's to make comparison so simple that education becomes unnecessary. If Alex can see "Lender A costs $4,200 more" in 3 seconds, he doesn't need to know what APR means.
Research
How CFPB Data + 6 User Interviews
Shaped Every Decision
I combined primary research (6 first-time homebuyer interviews) with secondary research from the CFPB and Freddie Mac. Both pointed to the same insight: the barrier isn't actual complexity. It's perceived complexity.
4/6
"I didn't even know I could
negotiate"
Four out of six participants didn't realize mortgage terms are negotiable. They treated the first estimate as final.
Primary Research — User Interviews, 2022
12 min
The jargon barrier
Participants spent an average of 12
minutes trying to understand a single loan estimate. Most gave up and focused only on monthly payment.
Primary Research — User Interviews, 2022
77%
"Easier than I expected"
Of borrowers who applied to multiple lenders, 77% said the process was less difficult than expected. The barrier isn't the
act. It's the anticipation.
Secondary Research — CFPB Report
Alex
SKEPTICAL
"Nobody told me I could push back on these numbers. I thought the loan estimate was like a price tag — take it or leave it."
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Key Insight That Shaped Every Decision
The opportunity isn't to educate borrowers about mortgages. It's to make comparison so simple that education becomes unnecessary. If Alex can see "Lender A costs $4,200 more" in 3 seconds, he doesn't need to know what APR means.
User Journey
From Confusion to
Confidence in 5 Steps
This map shows how a first-time homebuyer moves through the mortgage comparison process, and where Confer transforms each pain point into clarity.
Phase 1
Awareness
Day 1
Phase 2
Upload
Day 1-2
Phase 3
Compare
Day 2
Phase 4
Investigate
Day 2-3
Phase 5
Negotiate
Day 3-5
Actions
Receives 3 loan estimate
PDFs from different
lenders via email
Opens Confer, uploads 3
PDFs. App extracts key
data automatically.
Views side-by-side
comparison cards: rate,
payment, closing costs,
total
Taps into individual
estimates. Reads inline
jargon explainers. Spots
outlier fees.
Messages Lender A
through the app,
referencing Lender C's
lower rate
Thinking
"These all look the same. I
don't know what APR
means."
"Is this app actually
reading my documents?"
"Wait, Lender B has way
higher closing costs?"
"Oh, that's what
origination fee means."
"I just showed them the
other rate."
Feeling
Overwhelmed
Cautious
Aha moment
Informed
Empowered
Pain Points
PDFs are dense. No way to
compare. Gives up, takes
first offer.
No tool exists. Manual
spreadsheet. Takes 45+
min.
All numbers look identical.
Can't tell which is cheaper.
Jargon barrier. Doesn't
know APR vs interest rate.
Too intimidated to call.
Email loses context.
Confer Solves
Onboarding explains value.
"Save $3,000+ by
comparing."
Upload PDF or photo. Auto-
extraction with progress
feedback. 30 sec.
Summary cards. 4 key
metrics. Outliers
highlighted. Instant clarity.
Tap any term for plain-
language tooltip. No
context switching.
Tap a number to message
lender about it. Context
included.
Opportunity
Trust onboarding
Smart extraction
Visual hierarchy
Jargon translation
Contextual messaging
Emotional Arc
How Alex Feels at Each Stage
😰
Overwhelmed
"I don't understand any of
this"
›
🤔
Skeptical
"Can an app really help
with this?"
›
🧐
Curious
"Wait, I can see them side
by side?"
›
💡
Confident
"I can see which deal is
better"
›
🎯
Empowered
"I saved $4,100 by
switching lenders"
The Shift
Without Confer vs. With Confer
Without Confer
With Confer
The borrower's current experience
1
Receives 3 PDF loan estimates via email
2
Opens PDFs in separate tabs, compares manually
3
Spends 12 min per estimate decoding jargon
4
Gives up comparing. Focuses only on monthly payment
5
Takes the first offer. Leaves $3,000+ on the table
45+ min
Time spent (still confused)
$0
Saved by negotiating
The Confer experience
1
Uploads 3 PDFs to Confer (30 seconds)
2
Sees instant side-by-side cards with 4 key metrics
3
Spots outlier: Lender B's closing costs $1,200 higher
4
Taps term for tooltip. Understands APR in 5 seconds
5
Messages Lender A with competing rate. Saves $4,100
5 min
Time to full clarity
$4,100
Saved by negotiating
Key Touchpoints
5 Moments That Define the Experience
Phase 1
First Contact with Loan
Estimates
Overwhelmed
Pain: 15+ line items of jargon.
No way to compare. 47% of
borrowers stop here.
Confer: Onboarding shows
"Save $3K+ by comparing"
Phase 2
Document Processing
Anxiety
Cautious
Pain: Silent spinner. 3/6
testers hit "back" within 8
seconds.
Confer: Step-by-step
progress. Zero abandonment.
Phase 3
The Aha Moment
Clarity
Pain: Without this, borrower
misses $1,200 fee difference.
Confer: 4 key metrics. Outliers
highlighted. Truth in 3
seconds.
Phase 4
Jargon Becomes Clear
Informed
Pain: 5/6 testers didn't know
APR vs interest rate.
Confer: Inline tooltips. 89%
usage. No context switching.
Phase 5
From Passive to
Empowered
Empowered
Pain: 4/6 avoided contacting
lenders entirely.
Confer: Tap any number to
ask about it. Saved $4,100.
User Journey
From Confusion to
Confidence in 5 Steps
This map shows how a first-time homebuyer moves through the mortgage comparison process, and where Confer transforms each pain point into clarity.
Phase 1
Awareness
Day 1
Phase 2
Upload
Day 1-2
Phase 3
Compare
Day 2
Phase 4
Investigate
Day 2-3
Phase 5
Negotiate
Day 3-5
Actions
Receives 3 loan estimate
PDFs from different
lenders via email
Opens Confer, uploads 3
PDFs. App extracts key
data automatically.
Views side-by-side
comparison cards: rate,
payment, closing costs,
total
Taps into individual
estimates. Reads inline
jargon explainers. Spots
outlier fees.
Messages Lender A
through the app,
referencing Lender C's
lower rate
Thinking
"These all look the same. I
don't know what APR
means."
"Is this app actually
reading my documents?"
"Wait, Lender B has way
higher closing costs?"
"Oh, that's what
origination fee means."
"I just showed them the
other rate."
Feeling
Overwhelmed
Cautious
Aha moment
Informed
Empowered
Pain Points
PDFs are dense. No way to
compare. Gives up, takes
first offer.
No tool exists. Manual
spreadsheet. Takes 45+
min.
All numbers look identical.
Can't tell which is cheaper.
Jargon barrier. Doesn't
know APR vs interest rate.
Too intimidated to call.
Email loses context.
Confer Solves
Onboarding explains value.
"Save $3,000+ by
comparing."
Upload PDF or photo. Auto-
extraction with progress
feedback. 30 sec.
Summary cards. 4 key
metrics. Outliers
highlighted. Instant clarity.
Tap any term for plain-
language tooltip. No
context switching.
Tap a number to message
lender about it. Context
included.
Opportunity
Trust onboarding
Smart extraction
Visual hierarchy
Jargon translation
Contextual messaging
Emotional Arc
How Alex Feels at Each Stage
😰
Overwhelmed
"I don't understand any of
this"
›
🤔
Skeptical
"Can an app really help
with this?"
›
🧐
Curious
"Wait, I can see them side
by side?"
›
💡
Confident
"I can see which deal is
better"
›
🎯
Empowered
"I saved $4,100 by
switching lenders"
The Shift
Without Confer vs. With Confer
Without Confer
With Confer
The borrower's current experience
1
Receives 3 PDF loan estimates via email
2
Opens PDFs in separate tabs, compares manually
3
Spends 12 min per estimate decoding jargon
4
Gives up comparing. Focuses only on monthly payment
5
Takes the first offer. Leaves $3,000+ on the table
45+ min
Time spent (still confused)
$0
Saved by negotiating
The Confer experience
1
Uploads 3 PDFs to Confer (30 seconds)
2
Sees instant side-by-side cards with 4 key metrics
3
Spots outlier: Lender B's closing costs $1,200 higher
4
Taps term for tooltip. Understands APR in 5 seconds
5
Messages Lender A with competing rate. Saves $4,100
5 min
Time to full clarity
$4,100
Saved by negotiating
Key Touchpoints
5 Moments That Define the Experience
Phase 1
First Contact with Loan
Estimates
Overwhelmed
Pain: 15+ line items of jargon.
No way to compare. 47% of
borrowers stop here.
Confer: Onboarding shows
"Save $3K+ by comparing"
Phase 2
Document Processing
Anxiety
Cautious
Pain: Silent spinner. 3/6
testers hit "back" within 8
seconds.
Confer: Step-by-step
progress. Zero abandonment.
Phase 3
The Aha Moment
Clarity
Pain: Without this, borrower
misses $1,200 fee difference.
Confer: 4 key metrics. Outliers
highlighted. Truth in 3
seconds.
Phase 4
Jargon Becomes Clear
Informed
Pain: 5/6 testers didn't know
APR vs interest rate.
Confer: Inline tooltips. 89%
usage. No context switching.
Phase 5
From Passive to
Empowered
Empowered
Pain: 4/6 avoided contacting
lenders entirely.
Confer: Tap any number to
ask about it. Saved $4,100.
User Journey
From Confusion to
Confidence in 5 Steps
This map shows how a first-time homebuyer moves through the mortgage comparison process, and where Confer transforms each pain point into clarity.
Phase 1
Awareness
Day 1
Phase 2
Upload
Day 1-2
Phase 3
Compare
Day 2
Phase 4
Investigate
Day 2-3
Phase 5
Negotiate
Day 3-5
Actions
Receives 3 loan estimate
PDFs from different
lenders via email
Opens Confer, uploads 3
PDFs. App extracts key
data automatically.
Views side-by-side
comparison cards: rate,
payment, closing costs,
total
Taps into individual
estimates. Reads inline
jargon explainers. Spots
outlier fees.
Messages Lender A
through the app,
referencing Lender C's
lower rate
Thinking
"These all look the same. I
don't know what APR
means."
"Is this app actually
reading my documents?"
"Wait, Lender B has way
higher closing costs?"
"Oh, that's what
origination fee means."
"I just showed them the
other rate."
Feeling
Overwhelmed
Cautious
Aha moment
Informed
Empowered
Pain Points
PDFs are dense. No way to
compare. Gives up, takes
first offer.
No tool exists. Manual
spreadsheet. Takes 45+
min.
All numbers look identical.
Can't tell which is cheaper.
Jargon barrier. Doesn't
know APR vs interest rate.
Too intimidated to call.
Email loses context.
Confer Solves
Onboarding explains value.
"Save $3,000+ by
comparing."
Upload PDF or photo. Auto-
extraction with progress
feedback. 30 sec.
Summary cards. 4 key
metrics. Outliers
highlighted. Instant clarity.
Tap any term for plain-
language tooltip. No
context switching.
Tap a number to message
lender about it. Context
included.
Opportunity
Trust onboarding
Smart extraction
Visual hierarchy
Jargon translation
Contextual messaging
Emotional Arc
How Alex Feels at Each Stage
😰
Overwhelmed
"I don't understand any of
this"
›
🤔
Skeptical
"Can an app really help
with this?"
›
🧐
Curious
"Wait, I can see them side
by side?"
›
💡
Confident
"I can see which deal is
better"
›
🎯
Empowered
"I saved $4,100 by
switching lenders"
The Shift
Without Confer vs. With Confer
Without Confer
With Confer
The borrower's current experience
1
Receives 3 PDF loan estimates via email
2
Opens PDFs in separate tabs, compares manually
3
Spends 12 min per estimate decoding jargon
4
Gives up comparing. Focuses only on monthly payment
5
Takes the first offer. Leaves $3,000+ on the table
45+ min
Time spent (still confused)
$0
Saved by negotiating
The Confer experience
1
Uploads 3 PDFs to Confer (30 seconds)
2
Sees instant side-by-side cards with 4 key metrics
3
Spots outlier: Lender B's closing costs $1,200 higher
4
Taps term for tooltip. Understands APR in 5 seconds
5
Messages Lender A with competing rate. Saves $4,100
5 min
Time to full clarity
$4,100
Saved by negotiating
Key Touchpoints
5 Moments That Define the Experience
Phase 1
First Contact with Loan
Estimates
Overwhelmed
Pain: 15+ line items of jargon.
No way to compare. 47% of
borrowers stop here.
Confer: Onboarding shows
"Save $3K+ by comparing"
Phase 2
Document Processing
Anxiety
Cautious
Pain: Silent spinner. 3/6
testers hit "back" within 8
seconds.
Confer: Step-by-step
progress. Zero abandonment.
Phase 3
The Aha Moment
Clarity
Pain: Without this, borrower
misses $1,200 fee difference.
Confer: 4 key metrics. Outliers
highlighted. Truth in 3
seconds.
Phase 4
Jargon Becomes Clear
Informed
Pain: 5/6 testers didn't know
APR vs interest rate.
Confer: Inline tooltips. 89%
usage. No context switching.
Phase 5
From Passive to
Empowered
Empowered
Pain: 4/6 avoided contacting
lenders entirely.
Confer: Tap any number to
ask about it. Saved $4,100.
Design Decisions
3 Decisions That Shaped Every
Screen
These aren't just design choices. They're the tradeoffs I reasoned through at three levels: strategic (what to build), tactical (how it works), and micro (what users see first).
STRATEGIC DECISION
Why Cards Beat Tables for Mortgage Comparison
The Context
Borrowers needed to compare 3-5 estimates at once. Standard approach: data table. But
15+ line items on mobile means horizontal scrolling and cognitive overload. I needed a format
that surfaces the 3-4 numbers that drive decisions.
Option A: Scrollable Data Table
Traditional comparison table. Familiar pattern from desktop
comparison sites like Google Shopping. Shows all data at once.
+ Users recognize the pattern
+ All data visible at once
- Requires horizontal scroll on mobile
- 15+ rows creates cognitive overload
THE TRAP
Option B: Card-Based Summary
Each lender as a card with 4 key metrics (rate, payment, closing costs,
total cost). Tap to expand the full breakdown with inline jargon
explainers.
+ No horizontal scrolling
+ Surfaces what matters first
+ Progressive disclosure for details
+ Outlier values visually highlighted
CHOSEN
Verdict — Backed by Testing
Option B won. In usability testing, participants made faster decisions (avg 4.2 min vs 8+ min with table) and reported higher confidence. The
card format let people compare the 3-4 numbers that actually matter, then drill down when they needed detail. Progressive disclosure solved
the "too much vs. too little" tradeoff.
Tradeoff I Accepted
Cards hide data by default, which means power users (financial advisors, repeat borrowers) need an extra tap to see full details. I chose to optimize for the
primary user — first-time homebuyers who are overwhelmed by data — knowing it adds one interaction for experts. If we had built a "pro mode" toggle, I
would have exposed the table view there. That feature was cut for MVP scope.
TACTICAL DECISION
Why In-App Messaging Replaced Email Chains
The Context
4/6 participants wanted to negotiate but found email chains confusing and phone calls
intimidating. They needed a way to ask about specific numbers without losing context.
Option A: Email Integration
Pull in email threads with lenders. Familiar communication channel. But
emails lose context quickly: which estimate was the lender referencing?
+ No new behavior to learn
+ Users already have email
- Context gets lost across threads
- Can't reference specific data points
THE TRAP
Option B: Contextual In-App Messaging
Built-in chat linked to each loan estimate. Tap any data point to
reference it in a message. The lender sees exactly which number Alex is
asking about.
+ Messages tied to specific data points
+ Reduces intimidation (text vs call)
+ Lender sees exactly what you're asking about
+ Full conversation history per estimate
CHOSEN
Verdict — Backed by User Research
Option B won. Our research showed 4/6 participants avoided contacting lenders entirely because they "didn't know what to ask." Linking
messages to specific data points removed this barrier. In testing, participants sent 2-3x more messages to lenders when they could tap a
number and ask about it directly, compared to a generic "compose" button.
Tradeoff I Accepted
Building in-app messaging meant we couldn't launch without a backend messaging infrastructure. The engineering team estimated 3 extra weeks. I
pushed for it because the research signal was too strong — users who can't communicate with lenders don't negotiate, and borrowers who don't negotiate
overpay by thousands. The PM and I aligned that messaging was core to the value proposition, not a nice-to-have. We cut the rate alerts feature to stay on
timeline.
MICRO DECISION
Why Comparison Lives Before Individual Details
The Context
The app served two modes: comparison (estimates uploaded) and exploration (still
shopping). The IA question: what does Alex see first after uploading 3 estimates?
Option A: Single Feed
One scrolling feed mixing lender discovery, saved estimates, and
comparison tools. Similar to a social feed pattern. Everything in one
view.
+ Simple mental model (one feed)
+ Discoverable features
- Mixing modes confuses intent
- Comparison buried in feed
Option B: Tab-Based Navigation
Separate tabs: Explore (find lenders), Compare (side-by-side),
Messages (lender communication). Each tab matches a mental model
of the mortgage process.
+ Maps to user's mental model
+ Comparison is always one tap away
+ Clear intent per screen
+ Matches mortgage process phases
CHOSEN
Verdict — Backed by Usability Testing
Option B won. Usability testing showed participants expected distinct spaces for browsing vs. comparing. When presented with a single feed,
5/6 participants asked "where's the comparison?" The tab model matched their mental model: mortgage shopping has distinct phases
(research, compare, decide) and the navigation should reflect that.
Tradeoff I Accepted
Tabs create a harder onboarding challenge — new users need to understand 3 sections instead of one scrollable feed. I mitigated this with an onboarding
walkthrough that highlights each tab's purpose on first launch. The alternative (removing the walkthrough and relying on discovery) felt risky for a financial
tool where users arrive anxious and unfamiliar. I chose guided clarity over organic exploration.
UI Showcase
40+ Screens Built on a Modular System
Clean, high-contrast UI designed for financial clarity. Every component built in Figma with auto-layout for
responsive scaling.
Color System
Components
Upload Flow
Detail View
Messaging
Onboarding
Patterns
Comparison View
Confer Prototype
Confer Prototype
Confer Prototype
Design Decisions
3 Decisions That Shaped Every
Screen
These aren't just design choices. They're the tradeoffs I reasoned through at three levels: strategic (what to build), tactical (how it works), and micro (what users see first).
STRATEGIC DECISION
Why Cards Beat Tables for Mortgage Comparison
The Context
Borrowers needed to compare 3-5 estimates at once. Standard approach: data table. But
15+ line items on mobile means horizontal scrolling and cognitive overload. I needed a format
that surfaces the 3-4 numbers that drive decisions.
Option A: Scrollable Data Table
Traditional comparison table. Familiar pattern from desktop
comparison sites like Google Shopping. Shows all data at once.
+ Users recognize the pattern
+ All data visible at once
- Requires horizontal scroll on mobile
- 15+ rows creates cognitive overload
THE TRAP
Option B: Card-Based Summary
Each lender as a card with 4 key metrics (rate, payment, closing costs,
total cost). Tap to expand the full breakdown with inline jargon
explainers.
+ No horizontal scrolling
+ Surfaces what matters first
+ Progressive disclosure for details
+ Outlier values visually highlighted
CHOSEN
Verdict — Backed by Testing
Option B won. In usability testing, participants made faster decisions (avg 4.2 min vs 8+ min with table) and reported higher confidence. The
card format let people compare the 3-4 numbers that actually matter, then drill down when they needed detail. Progressive disclosure solved
the "too much vs. too little" tradeoff.
Tradeoff I Accepted
Cards hide data by default, which means power users (financial advisors, repeat borrowers) need an extra tap to see full details. I chose to optimize for the
primary user — first-time homebuyers who are overwhelmed by data — knowing it adds one interaction for experts. If we had built a "pro mode" toggle, I
would have exposed the table view there. That feature was cut for MVP scope.
TACTICAL DECISION
Why In-App Messaging Replaced Email Chains
The Context
4/6 participants wanted to negotiate but found email chains confusing and phone calls
intimidating. They needed a way to ask about specific numbers without losing context.
Option A: Email Integration
Pull in email threads with lenders. Familiar communication channel. But
emails lose context quickly: which estimate was the lender referencing?
+ No new behavior to learn
+ Users already have email
- Context gets lost across threads
- Can't reference specific data points
THE TRAP
Option B: Contextual In-App Messaging
Built-in chat linked to each loan estimate. Tap any data point to
reference it in a message. The lender sees exactly which number Alex is
asking about.
+ Messages tied to specific data points
+ Reduces intimidation (text vs call)
+ Lender sees exactly what you're asking about
+ Full conversation history per estimate
CHOSEN
Verdict — Backed by User Research
Option B won. Our research showed 4/6 participants avoided contacting lenders entirely because they "didn't know what to ask." Linking
messages to specific data points removed this barrier. In testing, participants sent 2-3x more messages to lenders when they could tap a
number and ask about it directly, compared to a generic "compose" button.
Tradeoff I Accepted
Building in-app messaging meant we couldn't launch without a backend messaging infrastructure. The engineering team estimated 3 extra weeks. I
pushed for it because the research signal was too strong — users who can't communicate with lenders don't negotiate, and borrowers who don't negotiate
overpay by thousands. The PM and I aligned that messaging was core to the value proposition, not a nice-to-have. We cut the rate alerts feature to stay on
timeline.
MICRO DECISION
Why Comparison Lives Before Individual Details
The Context
The app served two modes: comparison (estimates uploaded) and exploration (still
shopping). The IA question: what does Alex see first after uploading 3 estimates?
Option A: Single Feed
One scrolling feed mixing lender discovery, saved estimates, and
comparison tools. Similar to a social feed pattern. Everything in one
view.
+ Simple mental model (one feed)
+ Discoverable features
- Mixing modes confuses intent
- Comparison buried in feed
Option B: Tab-Based Navigation
Separate tabs: Explore (find lenders), Compare (side-by-side),
Messages (lender communication). Each tab matches a mental model
of the mortgage process.
+ Maps to user's mental model
+ Comparison is always one tap away
+ Clear intent per screen
+ Matches mortgage process phases
CHOSEN
Verdict — Backed by Usability Testing
Option B won. Usability testing showed participants expected distinct spaces for browsing vs. comparing. When presented with a single feed,
5/6 participants asked "where's the comparison?" The tab model matched their mental model: mortgage shopping has distinct phases
(research, compare, decide) and the navigation should reflect that.
Tradeoff I Accepted
Tabs create a harder onboarding challenge — new users need to understand 3 sections instead of one scrollable feed. I mitigated this with an onboarding
walkthrough that highlights each tab's purpose on first launch. The alternative (removing the walkthrough and relying on discovery) felt risky for a financial
tool where users arrive anxious and unfamiliar. I chose guided clarity over organic exploration.
UI Showcase
40+ Screens Built on a Modular System
Clean, high-contrast UI designed for financial clarity. Every component built in Figma with auto-layout for
responsive scaling.
Color System
Components
Upload Flow
Detail View
Messaging
Onboarding
Patterns
Comparison View
Confer Prototype
Confer Prototype
Confer Prototype
Design Decisions
3 Decisions That Shaped Every
Screen
These aren't just design choices. They're the tradeoffs I reasoned through at three levels: strategic (what to build), tactical (how it works), and micro (what users see first).
STRATEGIC DECISION
Why Cards Beat Tables for Mortgage Comparison
The Context
Borrowers needed to compare 3-5 estimates at once. Standard approach: data table. But
15+ line items on mobile means horizontal scrolling and cognitive overload. I needed a format
that surfaces the 3-4 numbers that drive decisions.
Option A: Scrollable Data Table
Traditional comparison table. Familiar pattern from desktop
comparison sites like Google Shopping. Shows all data at once.
+ Users recognize the pattern
+ All data visible at once
- Requires horizontal scroll on mobile
- 15+ rows creates cognitive overload
THE TRAP
Option B: Card-Based Summary
Each lender as a card with 4 key metrics (rate, payment, closing costs,
total cost). Tap to expand the full breakdown with inline jargon
explainers.
+ No horizontal scrolling
+ Surfaces what matters first
+ Progressive disclosure for details
+ Outlier values visually highlighted
CHOSEN
Verdict — Backed by Testing
Option B won. In usability testing, participants made faster decisions (avg 4.2 min vs 8+ min with table) and reported higher confidence. The
card format let people compare the 3-4 numbers that actually matter, then drill down when they needed detail. Progressive disclosure solved
the "too much vs. too little" tradeoff.
Tradeoff I Accepted
Cards hide data by default, which means power users (financial advisors, repeat borrowers) need an extra tap to see full details. I chose to optimize for the
primary user — first-time homebuyers who are overwhelmed by data — knowing it adds one interaction for experts. If we had built a "pro mode" toggle, I
would have exposed the table view there. That feature was cut for MVP scope.
TACTICAL DECISION
Why In-App Messaging Replaced Email Chains
The Context
4/6 participants wanted to negotiate but found email chains confusing and phone calls
intimidating. They needed a way to ask about specific numbers without losing context.
Option A: Email Integration
Pull in email threads with lenders. Familiar communication channel. But
emails lose context quickly: which estimate was the lender referencing?
+ No new behavior to learn
+ Users already have email
- Context gets lost across threads
- Can't reference specific data points
THE TRAP
Option B: Contextual In-App Messaging
Built-in chat linked to each loan estimate. Tap any data point to
reference it in a message. The lender sees exactly which number Alex is
asking about.
+ Messages tied to specific data points
+ Reduces intimidation (text vs call)
+ Lender sees exactly what you're asking about
+ Full conversation history per estimate
CHOSEN
Verdict — Backed by User Research
Option B won. Our research showed 4/6 participants avoided contacting lenders entirely because they "didn't know what to ask." Linking
messages to specific data points removed this barrier. In testing, participants sent 2-3x more messages to lenders when they could tap a
number and ask about it directly, compared to a generic "compose" button.
Tradeoff I Accepted
Building in-app messaging meant we couldn't launch without a backend messaging infrastructure. The engineering team estimated 3 extra weeks. I
pushed for it because the research signal was too strong — users who can't communicate with lenders don't negotiate, and borrowers who don't negotiate
overpay by thousands. The PM and I aligned that messaging was core to the value proposition, not a nice-to-have. We cut the rate alerts feature to stay on
timeline.
MICRO DECISION
Why Comparison Lives Before Individual Details
The Context
The app served two modes: comparison (estimates uploaded) and exploration (still
shopping). The IA question: what does Alex see first after uploading 3 estimates?
Option A: Single Feed
One scrolling feed mixing lender discovery, saved estimates, and
comparison tools. Similar to a social feed pattern. Everything in one
view.
+ Simple mental model (one feed)
+ Discoverable features
- Mixing modes confuses intent
- Comparison buried in feed
Option B: Tab-Based Navigation
Separate tabs: Explore (find lenders), Compare (side-by-side),
Messages (lender communication). Each tab matches a mental model
of the mortgage process.
+ Maps to user's mental model
+ Comparison is always one tap away
+ Clear intent per screen
+ Matches mortgage process phases
CHOSEN
Verdict — Backed by Usability Testing
Option B won. Usability testing showed participants expected distinct spaces for browsing vs. comparing. When presented with a single feed,
5/6 participants asked "where's the comparison?" The tab model matched their mental model: mortgage shopping has distinct phases
(research, compare, decide) and the navigation should reflect that.
Tradeoff I Accepted
Tabs create a harder onboarding challenge — new users need to understand 3 sections instead of one scrollable feed. I mitigated this with an onboarding
walkthrough that highlights each tab's purpose on first launch. The alternative (removing the walkthrough and relying on discovery) felt risky for a financial
tool where users arrive anxious and unfamiliar. I chose guided clarity over organic exploration.
UI Showcase
40+ Screens Built on a Modular System
Clean, high-contrast UI designed for financial clarity. Every component built in Figma with auto-layout for
responsive scaling.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Upload Estimates — Snap a photo or upload PDF. The app extracts key numbers automatically.
Confer Prototype
Confer Prototype
Confer Prototype
Design System
The System Behind the Screens
40+ screens needed to feel like one product. These are the principles I built
the system around, and the micro decisions that made the experience feel
polished.
Color = Meaning, Not Decoration
Every color in the system carries semantic meaning. Green means "best
deal" or "savings." Red/amber means "watch out" or "higher than
average." Blue is interactive/actionable. This isn't a style choice — it's an
information design decision.
Example: Outlier fees turn amber automatically. Users spotted fee
differences 3x faster after this change.
4px Grid, 8px Spacing Scale
All components sit on a 4px grid with spacing tokens at 8/16/24/32/48px.
This wasn't just for visual consistency — it meant any new screen could
be assembled from existing components in under 30 minutes. During 4
rounds of iteration, this saved an estimated 3+ hours of redesign work.
Tokens: spacing-xs (8px), spacing-sm (16px), spacing-md (24px),
spacing-lg (32px), spacing-xl (48px)
Progressive Disclosure Everywhere
The core interaction pattern: show the 3-4 numbers that matter, hide the
rest behind a tap. This applied to loan cards, fee breakdowns, jargon
explainers, and lender profiles. One pattern, applied consistently, so users
never had to learn a new interaction model.
Result: 89% of users tapped to expand at least one detail. Zero users
reported feeling "lost" in testing.
Accessibility as a Constraint, Not an Afterthought
All text meets WCAG AA contrast ratios. Touch targets are minimum
44x44px. Color is never the only indicator — outlier fees get both color
AND a text badge. Screen reader labels on every interactive element.
Financial tools can't afford to exclude anyone.
Tested with VoiceOver on iOS. Comparison cards read as: "Lender A, rate
3.5%, monthly payment $1,420, tap to expand details."
Micro Decisions
The Small Choices That Made It Feel Right
These aren't in any PRD. They're the interaction-level decisions I made because I was close enough to the problem
to notice they mattered.
Number Formatting
$1,420/mo not $1420.00
Mortgage numbers are big and scary. I
formatted all dollar amounts with commas,
abbreviated to "/mo" for monthly payments,
and used relative language ("$200 less than
Lender B") alongside absolute numbers.
Why: In testing, participants processed relative
comparisons 2x faster than absolute numbers
alone.
Sticky Comparison Bar
Key metrics follow you down the page
When users scrolled into detail views, a sticky
bar at the top showed the 3 key numbers
(rate, payment, total cost) for the selected
lender. This prevented the "wait, which one
was cheaper?" moment.
Why: 4/6 participants scrolled back up to re-
check numbers before this fix. Sticky bar
eliminated that entirely.
Upload Confidence
"We found 12 data points" confirmation
After PDF upload, instead of silently moving to
comparison, I added a confirmation screen:
"We extracted 12 data points from your loan
estimate. Review them before comparing."
This turned an anxious moment into a
confidence moment.
Why: Upload abandonment dropped from 50% to
near-zero after adding this single screen.
Jargon Tap Targets
Dotted underlines on financial terms
Every mortgage term has a dotted underline
(not solid — solid implies a link). Tapping
shows a plain-language tooltip in-context,
without navigating away. "APR = the true
yearly cost of your loan, including fees."
Why: 89% usage rate. The separate glossary
page it replaced had 0% usage.
Empty State Design
"Upload your first estimate to get
started"
The comparison screen starts empty with a
clear CTA and a 3-step visual showing what
will happen: Upload → Review → Compare.
No blank screen anxiety. The empty state
teaches the core flow.
Why: Onboarding completion increased 35% after
replacing the generic "No data yet" empty state.
Message Anchoring
Tap a number to ask about it
In the messaging feature, users can tap any
data point on a loan card and it pre-fills a
contextual message: "I have a question about
the $3,200 origination fee on Estimate #2."
The lender sees exactly what the borrower is
asking about.
Why: Participants sent 3x more messages when
they didn't have to formulate the question from
scratch.
Design System
The System Behind the Screens
40+ screens needed to feel like one product. These are the principles I built
the system around, and the micro decisions that made the experience feel
polished.
Color = Meaning, Not Decoration
Every color in the system carries semantic meaning. Green means "best
deal" or "savings." Red/amber means "watch out" or "higher than
average." Blue is interactive/actionable. This isn't a style choice — it's an
information design decision.
Example: Outlier fees turn amber automatically. Users spotted fee
differences 3x faster after this change.
4px Grid, 8px Spacing Scale
All components sit on a 4px grid with spacing tokens at 8/16/24/32/48px.
This wasn't just for visual consistency — it meant any new screen could
be assembled from existing components in under 30 minutes. During 4
rounds of iteration, this saved an estimated 3+ hours of redesign work.
Tokens: spacing-xs (8px), spacing-sm (16px), spacing-md (24px),
spacing-lg (32px), spacing-xl (48px)
Progressive Disclosure Everywhere
The core interaction pattern: show the 3-4 numbers that matter, hide the
rest behind a tap. This applied to loan cards, fee breakdowns, jargon
explainers, and lender profiles. One pattern, applied consistently, so users
never had to learn a new interaction model.
Result: 89% of users tapped to expand at least one detail. Zero users
reported feeling "lost" in testing.
Accessibility as a Constraint, Not an Afterthought
All text meets WCAG AA contrast ratios. Touch targets are minimum
44x44px. Color is never the only indicator — outlier fees get both color
AND a text badge. Screen reader labels on every interactive element.
Financial tools can't afford to exclude anyone.
Tested with VoiceOver on iOS. Comparison cards read as: "Lender A, rate
3.5%, monthly payment $1,420, tap to expand details."
Micro Decisions
The Small Choices That Made It Feel Right
These aren't in any PRD. They're the interaction-level decisions I made because I was close enough to the problem
to notice they mattered.
Number Formatting
$1,420/mo not $1420.00
Mortgage numbers are big and scary. I
formatted all dollar amounts with commas,
abbreviated to "/mo" for monthly payments,
and used relative language ("$200 less than
Lender B") alongside absolute numbers.
Why: In testing, participants processed relative
comparisons 2x faster than absolute numbers
alone.
Sticky Comparison Bar
Key metrics follow you down the page
When users scrolled into detail views, a sticky
bar at the top showed the 3 key numbers
(rate, payment, total cost) for the selected
lender. This prevented the "wait, which one
was cheaper?" moment.
Why: 4/6 participants scrolled back up to re-
check numbers before this fix. Sticky bar
eliminated that entirely.
Upload Confidence
"We found 12 data points" confirmation
After PDF upload, instead of silently moving to
comparison, I added a confirmation screen:
"We extracted 12 data points from your loan
estimate. Review them before comparing."
This turned an anxious moment into a
confidence moment.
Why: Upload abandonment dropped from 50% to
near-zero after adding this single screen.
Jargon Tap Targets
Dotted underlines on financial terms
Every mortgage term has a dotted underline
(not solid — solid implies a link). Tapping
shows a plain-language tooltip in-context,
without navigating away. "APR = the true
yearly cost of your loan, including fees."
Why: 89% usage rate. The separate glossary
page it replaced had 0% usage.
Empty State Design
"Upload your first estimate to get
started"
The comparison screen starts empty with a
clear CTA and a 3-step visual showing what
will happen: Upload → Review → Compare.
No blank screen anxiety. The empty state
teaches the core flow.
Why: Onboarding completion increased 35% after
replacing the generic "No data yet" empty state.
Message Anchoring
Tap a number to ask about it
In the messaging feature, users can tap any
data point on a loan card and it pre-fills a
contextual message: "I have a question about
the $3,200 origination fee on Estimate #2."
The lender sees exactly what the borrower is
asking about.
Why: Participants sent 3x more messages when
they didn't have to formulate the question from
scratch.
Design System
The System Behind the Screens
40+ screens needed to feel like one product. These are the principles I built
the system around, and the micro decisions that made the experience feel
polished.
Color = Meaning, Not Decoration
Every color in the system carries semantic meaning. Green means "best
deal" or "savings." Red/amber means "watch out" or "higher than
average." Blue is interactive/actionable. This isn't a style choice — it's an
information design decision.
Example: Outlier fees turn amber automatically. Users spotted fee
differences 3x faster after this change.
4px Grid, 8px Spacing Scale
All components sit on a 4px grid with spacing tokens at 8/16/24/32/48px.
This wasn't just for visual consistency — it meant any new screen could
be assembled from existing components in under 30 minutes. During 4
rounds of iteration, this saved an estimated 3+ hours of redesign work.
Tokens: spacing-xs (8px), spacing-sm (16px), spacing-md (24px),
spacing-lg (32px), spacing-xl (48px)
Progressive Disclosure Everywhere
The core interaction pattern: show the 3-4 numbers that matter, hide the
rest behind a tap. This applied to loan cards, fee breakdowns, jargon
explainers, and lender profiles. One pattern, applied consistently, so users
never had to learn a new interaction model.
Result: 89% of users tapped to expand at least one detail. Zero users
reported feeling "lost" in testing.
Accessibility as a Constraint, Not an Afterthought
All text meets WCAG AA contrast ratios. Touch targets are minimum
44x44px. Color is never the only indicator — outlier fees get both color
AND a text badge. Screen reader labels on every interactive element.
Financial tools can't afford to exclude anyone.
Tested with VoiceOver on iOS. Comparison cards read as: "Lender A, rate
3.5%, monthly payment $1,420, tap to expand details."
Micro Decisions
The Small Choices That Made It Feel Right
These aren't in any PRD. They're the interaction-level decisions I made because I was close enough to the problem
to notice they mattered.
Number Formatting
$1,420/mo not $1420.00
Mortgage numbers are big and scary. I
formatted all dollar amounts with commas,
abbreviated to "/mo" for monthly payments,
and used relative language ("$200 less than
Lender B") alongside absolute numbers.
Why: In testing, participants processed relative
comparisons 2x faster than absolute numbers
alone.
Sticky Comparison Bar
Key metrics follow you down the page
When users scrolled into detail views, a sticky
bar at the top showed the 3 key numbers
(rate, payment, total cost) for the selected
lender. This prevented the "wait, which one
was cheaper?" moment.
Why: 4/6 participants scrolled back up to re-
check numbers before this fix. Sticky bar
eliminated that entirely.
Upload Confidence
"We found 12 data points" confirmation
After PDF upload, instead of silently moving to
comparison, I added a confirmation screen:
"We extracted 12 data points from your loan
estimate. Review them before comparing."
This turned an anxious moment into a
confidence moment.
Why: Upload abandonment dropped from 50% to
near-zero after adding this single screen.
Jargon Tap Targets
Dotted underlines on financial terms
Every mortgage term has a dotted underline
(not solid — solid implies a link). Tapping
shows a plain-language tooltip in-context,
without navigating away. "APR = the true
yearly cost of your loan, including fees."
Why: 89% usage rate. The separate glossary
page it replaced had 0% usage.
Empty State Design
"Upload your first estimate to get
started"
The comparison screen starts empty with a
clear CTA and a 3-step visual showing what
will happen: Upload → Review → Compare.
No blank screen anxiety. The empty state
teaches the core flow.
Why: Onboarding completion increased 35% after
replacing the generic "No data yet" empty state.
Message Anchoring
Tap a number to ask about it
In the messaging feature, users can tap any
data point on a loan card and it pre-fills a
contextual message: "I have a question about
the $3,200 origination fee on Estimate #2."
The lender sees exactly what the borrower is
asking about.
Why: Participants sent 3x more messages when
they didn't have to formulate the question from
scratch.
Testing & Iterations
How 4 Rounds of Testing Shaped the
Final Design
Each round of usability testing revealed something I didn't expect. The final
design looks nothing like where I started, and that's the point.
6
Participants tested
3
Testing rounds
92%
Task success rate (final)
4.2/5
Confidence rating (final)
Testing method: Moderated usability testing via Zoom. Each session lasted 30-45 minutes. Participants were given 3 sample loan estimates and asked to
identify the best deal. I measured task completion time, error rate, confidence self-rating (1-5), and collected verbal think-aloud feedback. Recordings were
affinity-mapped after each round to identify patterns.
1
Reduce 5 Loan Estimates to 3 for MVP
TRIGGER: 4/6 participants froze when presented with 5 options
Users felt overwhelmed comparing 5 loan estimates simultaneously. Cognitive load testing showed decision quality dropped sharply past 3
options. This aligns with Hick's Law and the "paradox of choice" research.
BEFORE
→
Problem: 5 cards caused scroll fatigue. Participants compared only the
first 2-3 and ignored the rest. Average decision time: 12+ minutes.
AFTER
→
Fix: Capped at 3 for MVP. All cards visible without scrolling. Participants
engaged with all 3 equally. Decision time dropped to ~5 minutes.
2
Reduce Upload and Processing Time
TRIGGER: 3/6 participants abandoned the upload flow mid-process
Users abandoned the upload flow when processing took more than 8 seconds with no feedback. The silent loading spinner created anxiety
about whether their documents were actually being read.
BEFORE
→
Problem: Generic spinner with no progress indication. Users didn't know
if it was working. 3/6 hit "back" within 8 seconds.
AFTER
→
Fix: Step-by-step progress: "Reading document... Extracting rates...
Comparing lenders..." Reduced perceived wait time. Zero abandonment
in next round.
3
Surface Hidden Fees with Visual Highlighting
TRIGGER: 5/6 participants missed a $1,200 fee difference between lenders
Participants missed critical fee differences between lenders because data was presented in identical formats. When everything looks the
same, nothing stands out. Outlier values needed visual emphasis.
BEFORE
→
Problem: All numbers displayed in the same style. A $1,200 fee
difference looked identical to a $12 difference. Users scanned past
critical outliers.
AFTER
→
Fix: Red/amber highlighting for outlier lander. "recommended" and "High
Savings" and "Moderate Savings" columns. In the next test, 6/6 players immediately
spotted the fee difference.
Testing & Iterations
How 4 Rounds of Testing Shaped the
Final Design
Each round of usability testing revealed something I didn't expect. The final
design looks nothing like where I started, and that's the point.
6
Participants tested
3
Testing rounds
92%
Task success rate (final)
4.2/5
Confidence rating (final)
Testing method: Moderated usability testing via Zoom. Each session lasted 30-45 minutes. Participants were given 3 sample loan estimates and asked to
identify the best deal. I measured task completion time, error rate, confidence self-rating (1-5), and collected verbal think-aloud feedback. Recordings were
affinity-mapped after each round to identify patterns.
1
Reduce 5 Loan Estimates to 3 for MVP
TRIGGER: 4/6 participants froze when presented with 5 options
Users felt overwhelmed comparing 5 loan estimates simultaneously. Cognitive load testing showed decision quality dropped sharply past 3
options. This aligns with Hick's Law and the "paradox of choice" research.
BEFORE
→
Problem: 5 cards caused scroll fatigue. Participants compared only the
first 2-3 and ignored the rest. Average decision time: 12+ minutes.
AFTER
→
Fix: Capped at 3 for MVP. All cards visible without scrolling. Participants
engaged with all 3 equally. Decision time dropped to ~5 minutes.
2
Reduce Upload and Processing Time
TRIGGER: 3/6 participants abandoned the upload flow mid-process
Users abandoned the upload flow when processing took more than 8 seconds with no feedback. The silent loading spinner created anxiety
about whether their documents were actually being read.
BEFORE
→
Problem: Generic spinner with no progress indication. Users didn't know
if it was working. 3/6 hit "back" within 8 seconds.
AFTER
→
Fix: Step-by-step progress: "Reading document... Extracting rates...
Comparing lenders..." Reduced perceived wait time. Zero abandonment
in next round.
3
Surface Hidden Fees with Visual Highlighting
TRIGGER: 5/6 participants missed a $1,200 fee difference between lenders
Participants missed critical fee differences between lenders because data was presented in identical formats. When everything looks the
same, nothing stands out. Outlier values needed visual emphasis.
BEFORE
→
Problem: All numbers displayed in the same style. A $1,200 fee
difference looked identical to a $12 difference. Users scanned past
critical outliers.
AFTER
→
Fix: Red/amber highlighting for outlier lander. "recommended" and "High
Savings" and "Moderate Savings" columns. In the next test, 6/6 players immediately
spotted the fee difference.
Testing & Iterations
How 4 Rounds of Testing Shaped the
Final Design
Each round of usability testing revealed something I didn't expect. The final
design looks nothing like where I started, and that's the point.
6
Participants tested
3
Testing rounds
92%
Task success rate (final)
4.2/5
Confidence rating (final)
Testing method: Moderated usability testing via Zoom. Each session lasted 30-45 minutes. Participants were given 3 sample loan estimates and asked to
identify the best deal. I measured task completion time, error rate, confidence self-rating (1-5), and collected verbal think-aloud feedback. Recordings were
affinity-mapped after each round to identify patterns.
1
Reduce 5 Loan Estimates to 3 for MVP
TRIGGER: 4/6 participants froze when presented with 5 options
Users felt overwhelmed comparing 5 loan estimates simultaneously. Cognitive load testing showed decision quality dropped sharply past 3
options. This aligns with Hick's Law and the "paradox of choice" research.
BEFORE
→
Problem: 5 cards caused scroll fatigue. Participants compared only the
first 2-3 and ignored the rest. Average decision time: 12+ minutes.
AFTER
→
Fix: Capped at 3 for MVP. All cards visible without scrolling. Participants
engaged with all 3 equally. Decision time dropped to ~5 minutes.
2
Reduce Upload and Processing Time
TRIGGER: 3/6 participants abandoned the upload flow mid-process
Users abandoned the upload flow when processing took more than 8 seconds with no feedback. The silent loading spinner created anxiety
about whether their documents were actually being read.
BEFORE
→
Problem: Generic spinner with no progress indication. Users didn't know
if it was working. 3/6 hit "back" within 8 seconds.
AFTER
→
Fix: Step-by-step progress: "Reading document... Extracting rates...
Comparing lenders..." Reduced perceived wait time. Zero abandonment
in next round.
3
Surface Hidden Fees with Visual Highlighting
TRIGGER: 5/6 participants missed a $1,200 fee difference between lenders
Participants missed critical fee differences between lenders because data was presented in identical formats. When everything looks the
same, nothing stands out. Outlier values needed visual emphasis.
BEFORE
→
Problem: All numbers displayed in the same style. A $1,200 fee
difference looked identical to a $12 difference. Users scanned past
critical outliers.
AFTER
→
Fix: Red/amber highlighting for outlier lander. "recommended" and "High
Savings" and "Moderate Savings" columns. In the next test, 6/6 players immediately
spotted the fee difference.
Impact
From 45-Minute Confusion to 5-
Minute Clarity
Three numbers that tell the story of what changed.
5 min
Average comparison time
Down from 30+ minutes manually comparing
PDFs
Measured in usability testing (n=6)
$3K+
Potential savings per borrower
When borrowers compare 5 lenders instead of
1
Based on Freddie Mac research data
80%
Increase in form completion
After 4 rounds of iteration on the upload flow
Measured in usability testing (n=6)
A
Alex
RESOLVED
"I went from panicking about the biggest purchase of my life to actually understanding my options. I ended up switching from my bank's offer to a lender that saved me $4,100 over the loan. I would have never done that without being able to see the numbers side by side."
These metrics came from usability testing with 6 participants and projected estimates based on CFPB/Freddie Mac research. As a concept
project, real-world deployment data would be the next validation step. Being transparent about this is intentional — the methodology is sound,
and the directional signal is strong.
Business Context
Confer Inc. was an early-stage fintech startup targeting the $11.1 trillion U.S. mortgage market. The company's thesis: borrowers who compare
5+ lenders save an average of $3,000 (per Freddie Mac research), but 47% accept the first offer they receive because comparison is too hard.
My design work was used in the investor demo that helped secure the next phase of funding discussions.
$11.1T
U.S. mortgage market
47%
Borrowers who don't compare
Seed
Stage when I joined
Who I Worked With
CEO/Founder — Aligned on MVP scope, feature prioritization, and investor
demo requirements. Weekly design reviews.
PM — Sprint planning, user story writing, and backlog grooming. Co-
facilitated 2 of 4 usability testing rounds.
Backend Engineer — Defined PDF parsing constraints, API response shapes,
and data accuracy thresholds that directly shaped the UI.
Legal Advisor — Reviewed all mortgage terminology, disclosure
requirements, and comparison methodology for TILA-RESPA compliance.
Impact
From 45-Minute Confusion to 5-
Minute Clarity
Three numbers that tell the story of what changed.
5 min
Average comparison time
Down from 30+ minutes manually comparing
PDFs
Measured in usability testing (n=6)
$3K+
Potential savings per borrower
When borrowers compare 5 lenders instead of
1
Based on Freddie Mac research data
80%
Increase in form completion
After 4 rounds of iteration on the upload flow
Measured in usability testing (n=6)
A
Alex
RESOLVED
"I went from panicking about the biggest purchase of my life to actually understanding my options. I ended up switching from my bank's offer to a lender that saved me $4,100 over the loan. I would have never done that without being able to see the numbers side by side."
These metrics came from usability testing with 6 participants and projected estimates based on CFPB/Freddie Mac research. As a concept
project, real-world deployment data would be the next validation step. Being transparent about this is intentional — the methodology is sound,
and the directional signal is strong.
Business Context
Confer Inc. was an early-stage fintech startup targeting the $11.1 trillion U.S. mortgage market. The company's thesis: borrowers who compare
5+ lenders save an average of $3,000 (per Freddie Mac research), but 47% accept the first offer they receive because comparison is too hard.
My design work was used in the investor demo that helped secure the next phase of funding discussions.
$11.1T
U.S. mortgage market
47%
Borrowers who don't compare
Seed
Stage when I joined
Who I Worked With
CEO/Founder — Aligned on MVP scope, feature prioritization, and investor
demo requirements. Weekly design reviews.
PM — Sprint planning, user story writing, and backlog grooming. Co-
facilitated 2 of 4 usability testing rounds.
Backend Engineer — Defined PDF parsing constraints, API response shapes,
and data accuracy thresholds that directly shaped the UI.
Legal Advisor — Reviewed all mortgage terminology, disclosure
requirements, and comparison methodology for TILA-RESPA compliance.
Impact
From 45-Minute Confusion to 5-
Minute Clarity
Three numbers that tell the story of what changed.
5 min
Average comparison time
Down from 30+ minutes manually comparing
PDFs
Measured in usability testing (n=6)
$3K+
Potential savings per borrower
When borrowers compare 5 lenders instead of
1
Based on Freddie Mac research data
80%
Increase in form completion
After 4 rounds of iteration on the upload flow
Measured in usability testing (n=6)
A
Alex
RESOLVED
"I went from panicking about the biggest purchase of my life to actually understanding my options. I ended up switching from my bank's offer to a lender that saved me $4,100 over the loan. I would have never done that without being able to see the numbers side by side."
These metrics came from usability testing with 6 participants and projected estimates based on CFPB/Freddie Mac research. As a concept
project, real-world deployment data would be the next validation step. Being transparent about this is intentional — the methodology is sound,
and the directional signal is strong.
Business Context
Confer Inc. was an early-stage fintech startup targeting the $11.1 trillion U.S. mortgage market. The company's thesis: borrowers who compare
5+ lenders save an average of $3,000 (per Freddie Mac research), but 47% accept the first offer they receive because comparison is too hard.
My design work was used in the investor demo that helped secure the next phase of funding discussions.
$11.1T
U.S. mortgage market
47%
Borrowers who don't compare
Seed
Stage when I joined
Who I Worked With
CEO/Founder — Aligned on MVP scope, feature prioritization, and investor
demo requirements. Weekly design reviews.
PM — Sprint planning, user story writing, and backlog grooming. Co-
facilitated 2 of 4 usability testing rounds.
Backend Engineer — Defined PDF parsing constraints, API response shapes,
and data accuracy thresholds that directly shaped the UI.
Legal Advisor — Reviewed all mortgage terminology, disclosure
requirements, and comparison methodology for TILA-RESPA compliance.
Reflections
What I Learned About Designing for
Financial Anxiety
Three lessons from this project that changed how I think about designing
for complex, high-stakes decisions.
FROM ITERATION 1: Reducing 5 LEs to 3
In high-stakes decisions, less information creates more confidence
My instinct was to show everything: every fee, every rate, every line item. Testing proved the opposite. When I reduced from 5 loan estimates to
3 and surfaced only 4 key metrics per card, decision quality went up and decision time went down. This changed how I think about
information architecture for any complex domain. The job isn't to present all the data — it's to decide which data matters most, show that first,
and make the rest accessible without cluttering the primary view. Progressive disclosure isn't just a UI pattern. It's a trust-building strategy.
FROM CONSTRAINT: PDF parsing at ~70% accuracy
Designing around unreliable technology taught me to plan for failure states first
The PDF parser worked about 70% of the time. My first instinct was to wait for engineering to fix it. Instead, I designed a review-and-confirm
flow that turned the limitation into a feature — users felt more in control because they verified every number before comparing. This taught
me that constraints aren't blockers, they're design inputs. If I'd waited for perfect parsing, we'd have shipped nothing. By designing for the
imperfect state, we shipped on time and users reported feeling more confident because they'd verified the data themselves.
FROM COLLABORATION: Pushing back on the CEO's feature request
Saying "not yet" to a good feature is harder than saying "no" to a bad one
The CEO wanted a lender marketplace in v1. It's a great feature — 4/6 participants asked for it. But adding it meant delaying the core
comparison tool by 6+ weeks. I made the case (with testing data) that comparison alone solved the primary pain point, and that marketplace
without comparison was useless anyway. We shipped comparison first, kept marketplace for v2. The lesson: as a designer on a small team,
you're not just designing screens. You're helping the business decide what to build and when. That requires being comfortable using research
to challenge stakeholder assumptions, even when the stakeholder is the person who hired you.
FROM RESEARCH: 6 interviews, strong signal, real limits
I'd double my research sample size next time — 6 participants gave direction, not confidence
Six interviews surfaced clear patterns (4/6 didn't know they could negotiate, 5/6 couldn't distinguish APR from interest rate). But for a fintech
product where a wrong decision costs thousands, I'd want 8-10 participants to feel confident the patterns hold at scale. I'd also test with real
loan estimates instead of sample data. The emotional weight of real money changes how people interact with the tool. Sample data made
testing feel like a game. Real data would have revealed the anxiety-driven behaviors I'm actually designing for.
What's Next
1
Mortgage Marketplace — Connect borrowers with vetted lenders directly through the app, turning comparison into action.
Evidence: 4/6 participants said they wanted to apply to a new lender right from the comparison screen but had to leave the app to do it.
2
Shop for Services — Extend the transparent comparison model to home insurance, title companies, and inspectors.
Evidence: 3/6 participants asked "can I compare home insurance the same way?" during testing. The comparison pattern is transferable.
3
Real-Time Rate Alerts — Notify borrowers when rates drop below their current best offer, creating a reason to return to the
app.
Evidence: Mortgage rates changed 3 times during our testing period. Participants who knew about the change renegotiated. Those who didn't, missed
savings.
A
Alex
EPILOGUE
"I closed on my house last month. The lender I chose wasn't the one my realtor recommended — it was the one the numbers told me to
choose. I saved $4,100 over the life of my loan. That's a vacation I wouldn't have had."
Reflections
What I Learned About Designing for
Financial Anxiety
Three lessons from this project that changed how I think about designing
for complex, high-stakes decisions.
FROM ITERATION 1: Reducing 5 LEs to 3
In high-stakes decisions, less information creates more confidence
My instinct was to show everything: every fee, every rate, every line item. Testing proved the opposite. When I reduced from 5 loan estimates to
3 and surfaced only 4 key metrics per card, decision quality went up and decision time went down. This changed how I think about
information architecture for any complex domain. The job isn't to present all the data — it's to decide which data matters most, show that first,
and make the rest accessible without cluttering the primary view. Progressive disclosure isn't just a UI pattern. It's a trust-building strategy.
FROM CONSTRAINT: PDF parsing at ~70% accuracy
Designing around unreliable technology taught me to plan for failure states first
The PDF parser worked about 70% of the time. My first instinct was to wait for engineering to fix it. Instead, I designed a review-and-confirm
flow that turned the limitation into a feature — users felt more in control because they verified every number before comparing. This taught
me that constraints aren't blockers, they're design inputs. If I'd waited for perfect parsing, we'd have shipped nothing. By designing for the
imperfect state, we shipped on time and users reported feeling more confident because they'd verified the data themselves.
FROM COLLABORATION: Pushing back on the CEO's feature request
Saying "not yet" to a good feature is harder than saying "no" to a bad one
The CEO wanted a lender marketplace in v1. It's a great feature — 4/6 participants asked for it. But adding it meant delaying the core
comparison tool by 6+ weeks. I made the case (with testing data) that comparison alone solved the primary pain point, and that marketplace
without comparison was useless anyway. We shipped comparison first, kept marketplace for v2. The lesson: as a designer on a small team,
you're not just designing screens. You're helping the business decide what to build and when. That requires being comfortable using research
to challenge stakeholder assumptions, even when the stakeholder is the person who hired you.
FROM RESEARCH: 6 interviews, strong signal, real limits
I'd double my research sample size next time — 6 participants gave direction, not confidence
Six interviews surfaced clear patterns (4/6 didn't know they could negotiate, 5/6 couldn't distinguish APR from interest rate). But for a fintech
product where a wrong decision costs thousands, I'd want 8-10 participants to feel confident the patterns hold at scale. I'd also test with real
loan estimates instead of sample data. The emotional weight of real money changes how people interact with the tool. Sample data made
testing feel like a game. Real data would have revealed the anxiety-driven behaviors I'm actually designing for.
What's Next
1
Mortgage Marketplace — Connect borrowers with vetted lenders directly through the app, turning comparison into action.
Evidence: 4/6 participants said they wanted to apply to a new lender right from the comparison screen but had to leave the app to do it.
2
Shop for Services — Extend the transparent comparison model to home insurance, title companies, and inspectors.
Evidence: 3/6 participants asked "can I compare home insurance the same way?" during testing. The comparison pattern is transferable.
3
Real-Time Rate Alerts — Notify borrowers when rates drop below their current best offer, creating a reason to return to the
app.
Evidence: Mortgage rates changed 3 times during our testing period. Participants who knew about the change renegotiated. Those who didn't, missed
savings.
A
Alex
EPILOGUE
"I closed on my house last month. The lender I chose wasn't the one my realtor recommended — it was the one the numbers told me to
choose. I saved $4,100 over the life of my loan. That's a vacation I wouldn't have had."
Reflections
What I Learned About Designing for
Financial Anxiety
Three lessons from this project that changed how I think about designing
for complex, high-stakes decisions.
FROM ITERATION 1: Reducing 5 LEs to 3
In high-stakes decisions, less information creates more confidence
My instinct was to show everything: every fee, every rate, every line item. Testing proved the opposite. When I reduced from 5 loan estimates to
3 and surfaced only 4 key metrics per card, decision quality went up and decision time went down. This changed how I think about
information architecture for any complex domain. The job isn't to present all the data — it's to decide which data matters most, show that first,
and make the rest accessible without cluttering the primary view. Progressive disclosure isn't just a UI pattern. It's a trust-building strategy.
FROM CONSTRAINT: PDF parsing at ~70% accuracy
Designing around unreliable technology taught me to plan for failure states first
The PDF parser worked about 70% of the time. My first instinct was to wait for engineering to fix it. Instead, I designed a review-and-confirm
flow that turned the limitation into a feature — users felt more in control because they verified every number before comparing. This taught
me that constraints aren't blockers, they're design inputs. If I'd waited for perfect parsing, we'd have shipped nothing. By designing for the
imperfect state, we shipped on time and users reported feeling more confident because they'd verified the data themselves.
FROM COLLABORATION: Pushing back on the CEO's feature request
Saying "not yet" to a good feature is harder than saying "no" to a bad one
The CEO wanted a lender marketplace in v1. It's a great feature — 4/6 participants asked for it. But adding it meant delaying the core
comparison tool by 6+ weeks. I made the case (with testing data) that comparison alone solved the primary pain point, and that marketplace
without comparison was useless anyway. We shipped comparison first, kept marketplace for v2. The lesson: as a designer on a small team,
you're not just designing screens. You're helping the business decide what to build and when. That requires being comfortable using research
to challenge stakeholder assumptions, even when the stakeholder is the person who hired you.
FROM RESEARCH: 6 interviews, strong signal, real limits
I'd double my research sample size next time — 6 participants gave direction, not confidence
Six interviews surfaced clear patterns (4/6 didn't know they could negotiate, 5/6 couldn't distinguish APR from interest rate). But for a fintech
product where a wrong decision costs thousands, I'd want 8-10 participants to feel confident the patterns hold at scale. I'd also test with real
loan estimates instead of sample data. The emotional weight of real money changes how people interact with the tool. Sample data made
testing feel like a game. Real data would have revealed the anxiety-driven behaviors I'm actually designing for.
What's Next
1
Mortgage Marketplace — Connect borrowers with vetted lenders directly through the app, turning comparison into action.
Evidence: 4/6 participants said they wanted to apply to a new lender right from the comparison screen but had to leave the app to do it.
2
Shop for Services — Extend the transparent comparison model to home insurance, title companies, and inspectors.
Evidence: 3/6 participants asked "can I compare home insurance the same way?" during testing. The comparison pattern is transferable.
3
Real-Time Rate Alerts — Notify borrowers when rates drop below their current best offer, creating a reason to return to the
app.
Evidence: Mortgage rates changed 3 times during our testing period. Participants who knew about the change renegotiated. Those who didn't, missed
savings.
A
Alex
EPILOGUE
"I closed on my house last month. The lender I chose wasn't the one my realtor recommended — it was the one the numbers told me to
choose. I saved $4,100 over the life of my loan. That's a vacation I wouldn't have had."
Back To Home
Back To Home
Back To Home
Back To Home
Back To Home
Back To Home
Back To Home
Back To Home
Back To Home
What’s Next?
👋🏻 Say hello
I am open to 🧑🏻💻 full-time or Contract roles. I’d love to hear from you if you have a question, a project idea, or would like to schedule a call.
Amit Goyani
UI/UX Designer
Let's Work
Let's Work
Let's Work
©2026 Designed with ❤️ and hand-coded by Amit Goyani. All rights reserved.
What’s Next?
👋🏻 Say hello
I am open to 🧑🏻💻 full-time or Contract roles. I’d love to hear from you if you have a question, a project idea, or would like to schedule a call.
Amit Goyani
UI/UX Designer
Let's Work
Let's Work
Let's Work
©2026 Designed with ❤️ and hand-coded by Amit Goyani. All rights reserved.
What’s Next?
👋🏻 Say hello
I am open to 🧑🏻💻 full-time or Contract roles. I’d love to hear from you if you have a question, a project idea, or would like to schedule a call.
Amit Goyani
UI/UX Designer
Let's Work
Let's Work
Let's Work
©2026 Designed with ❤️ and hand-coded by Amit Goyani. All rights reserved.