Exploring how trust, authentication, and usability can transform the luxury collecting experience.
Exploring how trust, authentication, and usability can transform the luxury collecting experience.
Exploring how trust, authentication, and usability can transform the luxury collecting experience.
Client
Client
Client
Self Conceptual
Self Conceptual
Self Conceptual
Role
Role
Role
Product Designer
Product Designer
Product Designer
Duration
Duration
Duration
8 Months
8 Months
8 Months
Tools
Tools
Tools
Figma, Sketch, Miro
Figma, Sketch, Miro
Figma, Sketch, Miro
The Overview
The Overview
The Overview
Vault is a members’ marketplace for authenticated luxury collectibles — watches, handbags, sneakers, graded cards and rare objects. In a category where a single transaction can carry £28,400 of risk, the product’s real job is not browsing or checkout. It is making two invisible processes legible: who verified this, and what is it actually worth.
Vault is a members’ marketplace for authenticated luxury collectibles — watches, handbags, sneakers, graded cards and rare objects. In a category where a single transaction can carry £28,400 of risk, the product’s real job is not browsing or checkout. It is making two invisible processes legible: who verified this, and what is it actually worth.
Vault is a members’ marketplace for authenticated luxury collectibles — watches, handbags, sneakers, graded cards and rare objects. In a category where a single transaction can carry £28,400 of risk, the product’s real job is not browsing or checkout. It is making two invisible processes legible: who verified this, and what is it actually worth.
I designed 19 screens across desktop and iOS, a two-layer design system, and a 307-link prototype. This is a self-initiated concept project — it has not shipped, so the impact section reports what is measurable in the file and what I would instrument, rather than engagement numbers I would have had to invent.
I designed 19 screens across desktop and iOS, a two-layer design system, and a 307-link prototype. This is a self-initiated concept project — it has not shipped, so the impact section reports what is measurable in the file and what I would instrument, rather than engagement numbers I would have had to invent.
I designed 19 screens across desktop and iOS, a two-layer design system, and a 307-link prototype. This is a self-initiated concept project — it has not shipped, so the impact section reports what is measurable in the file and what I would instrument, rather than engagement numbers I would have had to invent.
Problem Statement
Problem Statement
Problem Statement
High-value collectors will not transact with strangers unless authentication and valuation are made legible, but the marketplaces serving them hide both behind badges and asking prices, so trust has to be rebuilt from scratch on every single listing.
High-value collectors will not transact with strangers unless authentication and valuation are made legible, but the marketplaces serving them hide both behind badges and asking prices, so trust has to be rebuilt from scratch on every single listing.
High-value collectors will not transact with strangers unless authentication and valuation are made legible, but the marketplaces serving them hide both behind badges and asking prices, so trust has to be rebuilt from scratch on every single listing.
To gain a deeper understanding of the problem, I identified the key factors contributing to its complexity. The problem statement served as a foundation for uncovering these underlying challenges, which are outlined below.
To gain a deeper understanding of the problem, I identified the key factors contributing to its complexity. The problem statement served as a foundation for uncovering these underlying challenges, which are outlined below.
To gain a deeper understanding of the problem, I identified the key factors contributing to its complexity. The problem statement served as a foundation for uncovering these underlying challenges, which are outlined below.
Authentication is asserted, not evidenced
Authentication is asserted, not evidenced
Authentication is asserted, not evidenced
A verified badge with no statement of what was checked, who checked it, or what happens if it turns out to be wrong.
A verified badge with no statement of what was checked, who checked it, or what happens if it turns out to be wrong.
A verified badge with no statement of what was checked, who checked it, or what happens if it turns out to be wrong.
Prices with no reference point
Prices with no reference point
Prices with no reference point
Every price guide in the category quotes asking prices — what sellers hope for, not what buyers actually paid.
Every price guide in the category quotes asking prices — what sellers hope for, not what buyers actually paid.
Every price guide in the category quotes asking prices — what sellers hope for, not what buyers actually paid.
Discovery collapses under its filters
Discovery collapses under its filters
Discovery collapses under its filters
Serious buyers genuinely need seven filter dimensions. Rendering all seven expanded produces a wall nobody uses.
Serious buyers genuinely need seven filter dimensions. Rendering all seven expanded produces a wall nobody uses.
Serious buyers genuinely need seven filter dimensions. Rendering all seven expanded produces a wall nobody uses.
The wait window nobody owns
The wait window nobody owns
The wait window nobody owns
No audited competitor shows item state between payment and delivery - the exact point where anxiety peaks.
No audited competitor shows item state between payment and delivery - the exact point where anxiety peaks.
No audited competitor shows item state between payment and delivery - the exact point where anxiety peaks.
Two users, opposite needs
Two users, opposite needs
Two users, opposite needs
A collector deciding on a sofa and a reseller managing 142 listings need opposite densities from one product.
A collector deciding on a sofa and a reseller managing 142 listings need opposite densities from one product.
A collector deciding on a sofa and a reseller managing 142 listings need opposite densities from one product.
Customers and Audience
Customers and Audience
Customers and Audience
Vault serves two segments whose requirements actively conflict. I defined them from observable market behaviour — platform-published seller statistics and public listing archives — rather than inventing research participants.
Vault serves two segments whose requirements actively conflict. I defined them from observable market behaviour — platform-published seller statistics and public listing archives — rather than inventing research participants.
Vault serves two segments whose requirements actively conflict. I defined them from observable market behaviour — platform-published seller statistics and public listing archives — rather than inventing research participants.
THE COLLECTOR - INVESTOR
THE COLLECTOR - INVESTOR
THE COLLECTOR - INVESTOR
Buys 2–8 high-value pieces a year
Buys 2–8 high-value pieces a year
Buys 2–8 high-value pieces a year
Decides slowly, researches for weeks
Decides slowly, researches for weeks
Decides slowly, researches for weeks
Abandons at the first unanswered objection
Abandons at the first unanswered objection
Abandons at the first unanswered objection
Wants proof, comparables and a human to call
Wants proof, comparables and a human to call
Wants proof, comparables and a human to call
THE PROFESSIONAL RESELLER
THE PROFESSIONAL RESELLER
THE PROFESSIONAL RESELLER
Runs 100+ live listings as a business
Runs 100+ live listings as a business
Runs 100+ live listings as a business
Lives in the dashboard, never the storefront
Lives in the dashboard, never the storefront
Lives in the dashboard, never the storefront
Optimises for velocity, margin and dispute avoidance
Optimises for velocity, margin and dispute avoidance
Optimises for velocity, margin and dispute avoidance
Wants density, bulk actions and fast payouts
Wants density, bulk actions and fast payouts
Wants density, bulk actions and fast payouts
How Might We?
How Might We?
How Might We?
To turn those five issues into opportunities, I framed each as a question tied to a specific person at a specific moment - so every later design decision had something concrete to be checked against.
To turn those five issues into opportunities, I framed each as a question tied to a specific person at a specific moment - so every later design decision had something concrete to be checked against.
To turn those five issues into opportunities, I framed each as a question tied to a specific person at a specific moment - so every later design decision had something concrete to be checked against.
01
01
01
How might we let a first-time buyer verify a £28,400 item without leaving the product to research it elsewhere?
How might we let a first-time buyer verify a £28,400 item without leaving the product to research it elsewhere?
How might we let a first-time buyer verify a £28,400 item without leaving the product to research it elsewhere?
02
02
02
How might we show that a price is fair without making the buyer do arithmetic under financial pressure?
How might we show that a price is fair without making the buyer do arithmetic under financial pressure?
How might we show that a price is fair without making the buyer do arithmetic under financial pressure?
03
03
03
How might we offer seven filter dimensions without rendering a wall of controls nobody reads?
How might we offer seven filter dimensions without rendering a wall of controls nobody reads?
How might we offer seven filter dimensions without rendering a wall of controls nobody reads?
04
04
04
How might we own the window between payment and delivery, where every competitor abandons the user?
How might we own the window between payment and delivery, where every competitor abandons the user?
How might we own the window between payment and delivery, where every competitor abandons the user?
05
05
05
How might we serve a collector and a professional reseller from one component library and one token set?
How might we serve a collector and a professional reseller from one component library and one token set?
How might we serve a collector and a professional reseller from one component library and one token set?
IDENTIFYING TRUST REQUIREMENTS
IDENTIFYING TRUST REQUIREMENTS
IDENTIFYING TRUST REQUIREMENTS
I worked backwards from the objection. I wrote down every question a first-time buyer asks before committing five figures to a stranger, then required the product to answer each one on the screen where it actually arises — not in a policy document three clicks away.
I worked backwards from the objection. I wrote down every question a first-time buyer asks before committing five figures to a stranger, then required the product to answer each one on the screen where it actually arises — not in a policy document three clicks away.
I worked backwards from the objection. I wrote down every question a first-time buyer asks before committing five figures to a stranger, then required the product to answer each one on the screen where it actually arises — not in a policy document three clicks away.
OBJECTION 01
OBJECTION 01
OBJECTION 01
Is it real?
Is it real?
Is it real?
OBJECTION 02
OBJECTION 02
OBJECTION 02
Am I overpaying?
Am I overpaying?
Am I overpaying?
OBJECTION 03
OBJECTION 03
What if it is not as described?
What if it is not as described?
OBJECTION 03
Can I get out of it?
OBJECTION 04
OBJECTION 04
OBJECTION 04
Who is this seller, really?
Who is this seller, really?
Who is this seller, really?
OBJECTION 05
OBJECTION 05
OBJECTION 05
Where is my money in between?
Where is my money in between?
Where is my money in between?
OBJECTION 06
OBJECTION 06
Can I get out of it?
Can I get out of it?
OBJECTION 06
What if it is not as described?
Research
Research
Research
JOURNEY MAPPING
JOURNEY MAPPING
JOURNEY MAPPING
Every competitor designs hard up to checkout, then abandons the user. That window became the product.
Every competitor designs hard up to checkout, then abandons the user. That window became the product.
Every competitor designs hard up to checkout, then abandons the user. That window became the product.
INFORMATION ARCHITECTURE
INFORMATION ARCHITECTURE
INFORMATION ARCHITECTURE
Acquire and operate are different jobs. Merging them into one navigation would have served neither.
Acquire and operate are different jobs. Merging them into one navigation would have served neither.
Acquire and operate are different jobs. Merging them into one navigation would have served neither.
USER FLOW
USER FLOW
USER FLOW
Wired end to end, because the branches are where trust is kept or lost.
Wired end to end, because the branches are where trust is kept or lost.
Wired end to end, because the branches are where trust is kept or lost.
SELLER - LIST AN ITEM
SELLER - LIST AN ITEM
SELLER - LIST AN ITEM
Seller Dashboard
Seller Dashboard
Seller Dashboard
New listing
New listing
New listing
Category & Specs
Category & Specs
Category & Specs
Photography checklist
Photography checklist
Photography checklist
Authentication
Authentication
Authentication
Price Against Market
Price Against Market
Price Against Market
ANYONE - VERIFY A CERTIFICATE
ANYONE - VERIFY A CERTIFICATE
ANYONE - VERIFY A CERTIFICATE
Footer link (no account)
Footer link (no account)
Footer link (no account)
Enter certificate ID
Enter certificate ID
Enter certificate ID
Record retrieved
Record retrieved
Record retrieved
Specialist, date, method shown
Specialist, date, method shown
Specialist, date, method shown
Report downloadable
Report downloadable
Report downloadable
Wireframing
Wireframing
Wireframing
Structure before surface - all screens in greyscale first
Structure before surface - all screens in greyscale first
Structure before surface - all screens in greyscale first
No colour, no typography, no photography. Rectangles for images, bars for text, outlines for controls. This is where the expensive layout questions got settled while they were still cheap to change.
No colour, no typography, no photography. Rectangles for images, bars for text, outlines for controls. This is where the expensive layout questions got settled while they were still cheap to change.
No colour, no typography, no photography. Rectangles for images, bars for text, outlines for controls. This is where the expensive layout questions got settled while they were still cheap to change.
01
01
01
DIMENSIONS DECIDED IN GREYSCALE — AND NEVER CHANGED AGAIN
DIMENSIONS DECIDED IN GREYSCALE — AND NEVER CHANGED AGAIN
DIMENSIONS DECIDED IN GREYSCALE — AND NEVER CHANGED AGAIN
Decision
Decision
Decision
Set at
Set at
Set at
Why it survived
Why it survived
Why it survived
Filter rail width
Filter rail width
Filter rail width
288 px
288 px
288 px
Wide enough for a two-line facet label plus a count badge without wrapping. Narrower forced truncation; wider stole a grid column.
Wide enough for a two-line facet label plus a count badge without wrapping. Narrower forced truncation; wider stole a grid column.
Wide enough for a two-line facet label plus a count badge without wrapping. Narrower forced truncation; wider stole a grid column.
Result grid
Result grid
Result grid
3-up at 1440
3-up at 1440
3-up at 1440
4-up dropped the product card below the width where a price, a market delta and a condition chip can coexist legibly.
4-up dropped the product card below the width where a price, a market delta and a condition chip can coexist legibly.
4-up dropped the product card below the width where a price, a market delta and a condition chip can coexist legibly.
Sticky purchase panel
Sticky purchase panel
Sticky purchase panel
432 px
432 px
432 px
Holds the price, the market verdict line, the escrow promise and two buttons without the verdict line wrapping to three lines.
Holds the price, the market verdict line, the escrow promise and two buttons without the verdict line wrapping to three lines.
Holds the price, the market verdict line, the escrow promise and two buttons without the verdict line wrapping to three lines.
Proof block position
Proof block position
Proof block position
Above the fold, above price
Above the fold, above price
Above the fold, above price
Tested by scrolling the greybox: if the certificate row fell below 900 px, it lost to the gallery. It had to move up.
Tested by scrolling the greybox: if the certificate row fell below 900 px, it lost to the gallery. It had to move up.
Tested by scrolling the greybox: if the certificate row fell below 900 px, it lost to the gallery. It had to move up.
Seller row height
Seller row height
Seller row height
56 px vs the buyer’s 72 px
56 px vs the buyer’s 72 px
56 px vs the buyer’s 72 px
Density was a wireframe decision, not a visual one — 40 rows have to be scannable in one screen.
Density was a wireframe decision, not a visual one — 40 rows have to be scannable in one screen.
Density was a wireframe decision, not a visual one — 40 rows have to be scannable in one screen.
02
02
02
SAME SCREEN, STRUCTURE TO SURFACE
SAME SCREEN, STRUCTURE TO SURFACE
SAME SCREEN, STRUCTURE TO SURFACE
Product Detail. Every structural dimension in the hi-fi screen was fixed in the greybox — the visual pass changed surface, never skeleton.
Product Detail. Every structural dimension in the hi-fi screen was fixed in the greybox — the visual pass changed surface, never skeleton.
Product Detail. Every structural dimension in the hi-fi screen was fixed in the greybox — the visual pass changed surface, never skeleton.
03
03
03
MOBILE WIREFRAMED SEPARATELY, NOT SCALED DOWN
MOBILE WIREFRAMED SEPARATELY, NOT SCALED DOWN
MOBILE WIREFRAMED SEPARATELY, NOT SCALED DOWN
The mobile wireframes are a separate argument, not a reflow. Every sidebar became a bottom sheet, every primary action moved into the thumb arc, and the product card lost two of its six signals before a single pixel was styled.
The mobile wireframes are a separate argument, not a reflow. Every sidebar became a bottom sheet, every primary action moved into the thumb arc, and the product card lost two of its six signals before a single pixel was styled.
The mobile wireframes are a separate argument, not a reflow. Every sidebar became a bottom sheet, every primary action moved into the thumb arc, and the product card lost two of its six signals before a single pixel was styled.
“The hi-fi didn’t invent the structure. It dressed it.”
“The hi-fi didn’t invent the structure. It dressed it.”
“The hi-fi didn’t invent the structure. It dressed it.”
Design System
Design System
Design System
01
01
01
TOKEN ARCHITECTURE
TOKEN ARCHITECTURE
TOKEN ARCHITECTURE
LAYER 1
LAYER 1
LAYER 1
Vault / Primitives
Vault / Primitives
Vault / Primitives
73 raw values in one mode. Ink, gold, silver and status ramps. Hidden from the variable picker so nothing can bind to them by accident.
73 raw values in one mode. Ink, gold, silver and status ramps. Hidden from the variable picker so nothing can bind to them by accident.
73 raw values in one mode. Ink, gold, silver and status ramps. Hidden from the variable picker so nothing can bind to them by accident.
LAYER 2
LAYER 2
LAYER 2
Vault / Theme
Vault / Theme
Vault / Theme
38 semantic tokens, each an alias to a primitive, resolved across Light and Dark modes. This is the only layer a screen is allowed to see.
38 semantic tokens, each an alias to a primitive, resolved across Light and Dark modes. This is the only layer a screen is allowed to see.
38 semantic tokens, each an alias to a primitive, resolved across Light and Dark modes. This is the only layer a screen is allowed to see.
LAYER 3
LAYER 3
LAYER 3
Screens & components
Screens & components
Screens & components
19 screens and 6 components bind exclusively to the semantic layer. Setting a frame’s mode retints everything inside it, instances included.
19 screens and 6 components bind exclusively to the semantic layer. Setting a frame’s mode retints everything inside it, instances included.
19 screens and 6 components bind exclusively to the semantic layer. Setting a frame’s mode retints everything inside it, instances included.
ONE CHAIN, END TO END
ONE CHAIN, END TO END
ONE CHAIN, END TO END
PRIMITIVE
PRIMITIVE
PRIMITIVE
gold/500 #C6A15B
gold/500 #C6A15B
gold/500 #C6A15B
SEMANTIC
SEMANTIC
SEMANTIC
accent/default
accent/default
accent/default
BOUND PROPERTY
BOUND PROPERTY
BOUND PROPERTY
Button/Primary → fill
Button/Primary → fill
Button/Primary → fill
Nothing in a screen ever references gold/500 directly. That single rule is what made the accessibility fix in section 12.4 a one-line change.
Nothing in a screen ever references gold/500 directly. That single rule is what made the accessibility fix in section 12.4 a one-line change.
Nothing in a screen ever references gold/500 directly. That single rule is what made the accessibility fix in section 12.4 a one-line change.
02
02
02
COLOUR — THREE RAMPS AND A RULE
COLOUR — THREE RAMPS AND A RULE
COLOUR — THREE RAMPS AND A RULE
INK
INK
INK
ink900
ink400
ink850
ink300
ink800
ink200
ink750
ink150
ink700
ink100
ink650
ink50
ink600
ink25
ink500
ink900
ink600
ink100
ink850
ink500
ink50
ink800
ink400
ink25
ink750
ink300
ink700
ink200
ink650
ink150
ink900
ink750
ink600
ink300
ink100
ink850
ink700
ink500
ink200
ink50
ink800
ink650
ink400
ink150
ink25
GOLD
GOLD
GOLD
gold900
gold100
gold800
gold700
gold600
gold500
gold400
gold300
gold200
gold900
gold300
gold800
gold200
gold700
gold100
gold600
gold500
gold400
gold900
gold600
gold300
gold800
gold500
gold200
gold700
gold400
gold100
STATUS
STATUS
STATUS
green500
blue100
green400
amber500
green100
amber400
red500
red400
red100
blue500
blue400
green500
blue500
green400
blue400
green100
blue100
red500
amber500
red400
amber400
red100
green500
red500
blue500
amber500
green400
red400
blue400
amber400
green100
red100
blue100
The rule: gold is a brand colour, not a text colour. Raw gold on white measures 2.4:1 and fails AA outright. It is reserved for fills, borders and the primary data series, while `text/accent` resolves to a darker gold in Light and a lighter gold in Dark. Same idea, two values.
The rule: gold is a brand colour, not a text colour. Raw gold on white measures 2.4:1 and fails AA outright. It is reserved for fills, borders and the primary data series, while `text/accent` resolves to a darker gold in Light and a lighter gold in Dark. Same idea, two values.
The rule: gold is a brand colour, not a text colour. Raw gold on white measures 2.4:1 and fails AA outright. It is reserved for fills, borders and the primary data series, while `text/accent` resolves to a darker gold in Light and a lighter gold in Dark. Same idea, two values.
03
03
03
TYPOGRAPHY - THREE FAMILIES, EACH WITH A JOB
TYPOGRAPHY - THREE FAMILIES, EACH WITH A JOB
TYPOGRAPHY - THREE FAMILIES, EACH WITH A JOB
SF Pro Bold
SF Pro Bold
SF Pro Bold
Display only
Display only
Display only
Editorial authority. Never below 30 px, never on a UI label.
Editorial authority. Never below 30 px, never on a UI label.
Editorial authority. Never below 30 px, never on a UI label.
Collect extraordinary assets
Collect extraordinary assets
SF Pro
SF Pro
SF Pro
The interface
The interface
The interface
Every heading, body, label and control in the product.
Every heading, body, label and control in the product.
Every heading, body, label and control in the product.
Certificate of authenticity
JetBrains Mono
JetBrains Mono
JetBrains Mono
Every number
Every number
Every number
Prices, percentages, certificate IDs. Tabular alignment is the entire point — figures line up in columns across stat bars, tables and price blocks.
Prices, percentages, certificate IDs. Tabular alignment is the entire point — figures line up in columns across stat bars, tables and price blocks.
Prices, percentages, certificate IDs. Tabular alignment is the entire point — figures line up in columns across stat bars, tables and price blocks.
£28,400 · −5.0% · VLT-4471-A
04
04
04
THE PROOF — ONE SCREEN, ONE PROPERTY, TWO THEMES
THE PROOF — ONE SCREEN, ONE PROPERTY, TWO THEMES
THE PROOF — ONE SCREEN, ONE PROPERTY, TWO THEMES
These are the same frame, duplicated once. The only difference between them is the variable mode set on the parent. No overrides, no variant swap, no second component — every fill, stroke and text colour re-resolves through the semantic layer, including inside nested instances.
These are the same frame, duplicated once. The only difference between them is the variable mode set on the parent. No overrides, no variant swap, no second component — every fill, stroke and text colour re-resolves through the semantic layer, including inside nested instances.
These are the same frame, duplicated once. The only difference between them is the variable mode set on the parent. No overrides, no variant swap, no second component — every fill, stroke and text colour re-resolves through the semantic layer, including inside nested instances.
05
05
05
ACCESSIBILITY — WCAG 2.2 AA AS THE FLOOR
ACCESSIBILITY — WCAG 2.2 AA AS THE FLOOR
ACCESSIBILITY — WCAG 2.2 AA AS THE FLOOR
I computed every colour pair rather than eyeballing swatches. That distinction matters: the swatches looked fine, and four of them were not.
I computed every colour pair rather than eyeballing swatches. That distinction matters: the swatches looked fine, and four of them were not.
I computed every colour pair rather than eyeballing swatches. That distinction matters: the swatches looked fine, and four of them were not.
Pair
Pair
Pair
Ratio
Ratio
Ratio
Result
Result
Result
Primary text on canvas
Primary text on canvas
Primary text on canvas
18.8 : 1
18.8 : 1
18.8 : 1
AAA — comfortable at every size
AAA — comfortable at every size
AAA — comfortable at every size
Secondary text on canvas
Secondary text on canvas
Secondary text on canvas
4.6 : 1
4.6 : 1
4.6 : 1
AA
AA
AA
Tertiary text on canvas
Tertiary text on canvas
Tertiary text on canvas
3.1 : 1
3.1 : 1
3.1 : 1
AA Large only — captions, never body copy
AA Large only — captions, never body copy
AA Large only — captions, never body copy
Accent text on surface
Accent text on surface
Accent text on surface
4.8 : 1
4.8 : 1
4.8 : 1
AA
AA
AA
Raw gold as text on white
Raw gold as text on white
Raw gold as text on white
2.4 : 1
2.4 : 1
2.4 : 1
Fails. Never used for text anywhere in the system.
Fails. Never used for text anywhere in the system.
Fails. Never used for text anywhere in the system.
Focus ring, light / dark
Focus ring, light / dark
Focus ring, light / dark
4.5 : 1 / 10.4 : 1
4.5 : 1 / 10.4 : 1
4.5 : 1 / 10.4 : 1
AA for non-text indicators
AA for non-text indicators
AA for non-text indicators
High Fidelity
High Fidelity
High Fidelity
Main Screens that carry the product
Main Screens that carry the product
Main Screens that carry the product
Each one answers a single objection a collector would otherwise leave to research.
Each one answers a single objection a collector would otherwise leave to research.
Each one answers a single objection a collector would otherwise leave to research.
01
01
01
Opens with a live authentication count, not a product.
Opens with a live authentication count, not a product.
Opens with a live authentication count, not a product.
Chrono24 and eBay Luxury open on inventory; the trust proposition sits below the fold or on a policy page. Vault opens on a live authentication count and qualifies every statistic. The marketplace has to earn the visit before it sells anything.
Chrono24 and eBay Luxury open on inventory; the trust proposition sits below the fold or on a policy page. Vault opens on a live authentication count and qualifies every statistic. The marketplace has to earn the visit before it sells anything.
Chrono24 and eBay Luxury open on inventory; the trust proposition sits below the fold or on a policy page. Vault opens on a live authentication count and qualifies every statistic. The marketplace has to earn the visit before it sells anything.
02
02
02
Surface Settled Prices, Not Asking Prices
Surface Settled Prices, Not Asking Prices
Surface Settled Prices, Not Asking Prices
Nearly every price guide in this category quotes asking prices — what sellers hope for, not what buyers paid. Publishing settled transactions is the one advantage a rival cannot copy without the same volume, so it earns its own surface, with confidence stated on every finding.
Nearly every price guide in this category quotes asking prices — what sellers hope for, not what buyers paid. Publishing settled transactions is the one advantage a rival cannot copy without the same volume, so it earns its own surface, with confidence stated on every finding.
Nearly every price guide in this category quotes asking prices — what sellers hope for, not what buyers paid. Publishing settled transactions is the one advantage a rival cannot copy without the same volume, so it earns its own surface, with confidence stated on every finding.
03
03
03
Prooves Itself With Rejections, Not Approvals
Prooves Itself With Rejections, Not Approvals
Prooves Itself With Rejections, Not Approvals
Competitors publish only their successes. Vault publishes the rejection count and the dispute rate against an industry benchmark, and names the specialist behind each check. A trust page that cannot be falsified reads as marketing.
Competitors publish only their successes. Vault publishes the rejection count and the dispute rate against an industry benchmark, and names the specialist behind each check. A trust page that cannot be falsified reads as marketing.
Competitors publish only their successes. Vault publishes the rejection count and the dispute rate against an industry benchmark, and names the specialist behind each check. A trust page that cannot be falsified reads as marketing.
04
04
04
Keeps Price and Proof On The Screen, Not Behind a Tab
Keeps Price and Proof On The Screen, Not Behind a Tab
Keeps Price and Proof On The Screen, Not Behind a Tab
Marketplace apps open on a search field, which serves the last 10% of the journey and abandons the first 90%. Vault opens editorially — collectors browse to feel something long before they are ready to filter.
Marketplace apps open on a search field, which serves the last 10% of the journey and abandons the first 90%. Vault opens editorially — collectors browse to feel something long before they are ready to filter.
Marketplace apps open on a search field, which serves the last 10% of the journey and abandons the first 90%. Vault opens editorially — collectors browse to feel something long before they are ready to filter.
The docked bar restates the price so it never scrolls away, and verification sits over the gallery rather than behind a tab. Most apps bury both, then wonder why five-figure purchases stall at the last moment.
The docked bar restates the price so it never scrolls away, and verification sits over the gallery rather than behind a tab. Most apps bury both, then wonder why five-figure purchases stall at the last moment.
The docked bar restates the price so it never scrolls away, and verification sits over the gallery rather than behind a tab. Most apps bury both, then wonder why five-figure purchases stall at the last moment.
Solution
Solution
Solution
01
01
Homepage
Homepage
Proves the marketplace is safe before showing a product. Leads with a live authentication count, and every statistic carries a qualifier that makes it checkable.
Proves the marketplace is safe before showing a product. Leads with a live authentication count, and every statistic carries a qualifier that makes it checkable.
TOP OF FUNNEL
TOP OF FUNNEL
02
02
Explore Marketplace
Explore Marketplace
Seven filter dimensions without a wall of controls. One group open by default, and a price histogram turns “what is my maximum?” into a visual decision.
Seven filter dimensions without a wall of controls. One group open by default, and a price histogram turns “what is my maximum?” into a visual decision.
DISCOVERY
DISCOVERY
03
03
Product Detail
Product Detail
Answers every objection before it is asked. Certificate, condition report and escrow promise all sit above the Buy button, not below it.
Answers every objection before it is asked. Certificate, condition report and escrow promise all sit above the Buy button, not below it.
CONVERSION
CONVERSION
04
04
Buyer Dashboard
Buyer Dashboard
Owns the wait window. Authentication runs as a six-step tracker with timestamps, and purchases are reframed as a tracked portfolio.
Owns the wait window. Authentication runs as a six-step tracker with timestamps, and purchases are reframed as a tracked portfolio.
RETENTION
RETENTION
05
05
Seller Dashboard
Seller Dashboard
Density is the feature — 56 px rows against the buyer’s 72 px. Every KPI carries a peer comparator, because a number alone means nothing.
Density is the feature — 56 px rows against the buyer’s 72 px. Every KPI carries a peer comparator, because a number alone means nothing.
OPERATIONS
OPERATIONS
06
06
Concierge Service
Concierge Service
Publishes its fees before any form. Selects the specialist first and the time second, because the user is buying a person’s judgement.
Publishes its fees before any form. Selects the specialist first and the time second, because the user is buying a person’s judgement.
ESCALATION
ESCALATION
07
07
Market Insights
Market Insights
The data advantage made visible. Every AI-generated finding carries its evidence and an explicit confidence level.
The data advantage made visible. Every AI-generated finding carries its evidence and an explicit confidence level.
PROOF
PROOF
08
08
Trust & Authentication
Trust & Authentication
Converts the sceptic. Six steps, named specialists, and the uncomfortable numbers — 1,842 items rejected last month.
Converts the sceptic. Six steps, named specialists, and the uncomfortable numbers — 1,842 items rejected last month.
TRUST
TRUST
01
Homepage
Proves the marketplace is safe before showing a product. Leads with a live authentication count, and every statistic carries a qualifier that makes it checkable.
TOP OF FUNNEL
02
Explore Marketplace
Seven filter dimensions without a wall of controls. One group open by default, and a price histogram turns “what is my maximum?” into a visual decision.
DISCOVERY
03
Product Detail
Answers every objection before it is asked. Certificate, condition report and escrow promise all sit above the Buy button, not below it.
CONVERSION
04
Buyer Dashboard
Owns the wait window. Authentication runs as a six-step tracker with timestamps, and purchases are reframed as a tracked portfolio.
RETENTION
05
Seller Dashboard
Density is the feature — 56 px rows against the buyer’s 72 px. Every KPI carries a peer comparator, because a number alone means nothing.
OPERATIONS
06
Concierge Service
Publishes its fees before any form. Selects the specialist first and the time second, because the user is buying a person’s judgement.
ESCALATION
07
Market Insights
The data advantage made visible. Every AI-generated finding carries its evidence and an explicit confidence level.
PROOF
08
Trust & Authentication
Converts the sceptic. Six steps, named specialists, and the uncomfortable numbers — 1,842 items rejected last month.
TRUST
INTERACTION DESIGN
INTERACTION DESIGN
INTERACTION DESIGN
Every transition answers a question the user would otherwise have to ask: did that work, where am I, and can I still change my mind?
Every transition answers a question the user would otherwise have to ask: did that work, where am I, and can I still change my mind?
Every transition answers a question the user would otherwise have to ask: did that work, where am I, and can I still change my mind?
01
01
01
EVERY CONTROL, SIX STATES
EVERY CONTROL, SIX STATES
EVERY CONTROL, SIX STATES
Buy now
Buy now
Buy now
DEFAULT
DEFAULT
DEFAULT
Buy now
Buy now
Buy now
HOVER
HOVER
HOVER
Buy now
Buy now
Buy now
PRESSED
PRESSED
PRESSED
Buy now
Buy now
Buy now
FOCUS
FOCUS
FOCUS
Buy now
Buy now
Buy now
DISABLED
DISABLED
DISABLED
Processing
Processing
Processing
LOADING
LOADING
LOADING
Focus uses a dedicated `focus/ring` token at 8.1:1 against the fill — never a browser default, never removed. Pressed is an opacity dip only; a button that moves £28,400 does not bounce.
Focus uses a dedicated `focus/ring` token at 8.1:1 against the fill — never a browser default, never removed. Pressed is an opacity dip only; a button that moves £28,400 does not bounce.
Focus uses a dedicated `focus/ring` token at 8.1:1 against the fill — never a browser default, never removed. Pressed is an opacity dip only; a button that moves £28,400 does not bounce.
02
02
02
NEVER COLOUR ALONE
NEVER COLOUR ALONE
NEVER COLOUR ALONE
−5.0% below market
−5.0% below market
−5.0% below market
+3.2% above market
+3.2% above market
+3.2% above market
At market value
At market value
At market value
Tint, arrow and a signed word. Remove the colour and it still reads.
Tint, arrow and a signed word. Remove the colour and it still reads.
Tint, arrow and a signed word. Remove the colour and it still reads.
03
03
03
MOTION SPECIFICATION
MOTION SPECIFICATION
MOTION SPECIFICATION
Movement
Transition
Duration
What it tells the user
Tab switching
Instant
0 ms
A tab change is not travel. Animating it invents a journey that did not happen.
Into a detail screen
Push left
300 ms
You went deeper, and back will bring you out.
Modal · checkout · filter sheet
Move in from bottom
350 ms
A mode, not a destination — where you were is still there.
Applying a filter
Smart animate
250 ms
The sheet slides while results stay put, so you keep your place.
Price or market value update
Cross-fade with count
400 ms
The number changed; it did not reload.
04
04
04
MICROINTERACTIONS WORTH DEFENDING
MICROINTERACTIONS WORTH DEFENDING
MICROINTERACTIONS WORTH DEFENDING
Filter count preview
Filter count preview
Filter count preview
The badge updates before you commit, so applying is never a gamble.
The badge updates before you commit, so applying is never a gamble.
The badge updates before you commit, so applying is never a gamble.
Histogram brush
Histogram brush
Histogram brush
Handles snap to real density edges — you cannot select an empty band.
Handles snap to real density edges — you cannot select an empty band.
Handles snap to real density edges — you cannot select an empty band.
Authentication tick
Authentication tick
Authentication tick
Each step animates once, then stays ticked with a timestamp.
Each step animates once, then stays ticked with a timestamp.
Each step animates once, then stays ticked with a timestamp.
Panel elevation
Panel elevation
Panel elevation
Shadow appears only when the panel detaches, so elevation means “floating”.
Shadow appears only when the panel detaches, so elevation means “floating”.
Shadow appears only when the panel detaches, so elevation means “floating”.
Inline offer
Inline offer
Inline offer
The form grows out of the button, keeping the verdict on screen.
The form grows out of the button, keeping the verdict on screen.
The form grows out of the button, keeping the verdict on screen.
Optimistic wishlist
Optimistic wishlist
Optimistic wishlist
Fills instantly, rolls back visibly if the write fails.
Fills instantly, rolls back visibly if the write fails.
Fills instantly, rolls back visibly if the write fails.
Prototyping & Evaluation
Prototyping & Evaluation
Prototyping & Evaluation
Each one answers a single objection a collector would otherwise leave to research.
Each one answers a single objection a collector would otherwise leave to research.
Each one answers a single objection a collector would otherwise leave to research.
01
01
01
INTERACTION MAP — THE MOBILE BUYER JOURNEY, END TO END
INTERACTION MAP — THE MOBILE BUYER JOURNEY, END TO END
INTERACTION MAP — THE MOBILE BUYER JOURNEY, END TO END
Reactions were set on the components — the tab bar, both product cards, the nav and the footer — rather than on each instance. Re-routing the entire prototype is therefore a one-node edit rather than a 300-link rebuild.
Reactions were set on the components — the tab bar, both product cards, the nav and the footer — rather than on each instance. Re-routing the entire prototype is therefore a one-node edit rather than a 300-link rebuild.
Reactions were set on the components — the tab bar, both product cards, the nav and the footer — rather than on each instance. Re-routing the entire prototype is therefore a one-node edit rather than a 300-link rebuild.
02
02
02
WHAT THE PROTOTYPE WAS FOR
WHAT THE PROTOTYPE WAS FOR
WHAT THE PROTOTYPE WAS FOR
Prove the flows resolve
Prove the flows resolve
Prove the flows resolve
Every button, tab and card had to land somewhere real. Wiring is the cheapest way to find the screen you forgot to design.
Every button, tab and card had to land somewhere real. Wiring is the cheapest way to find the screen you forgot to design.
Every button, tab and card had to land somewhere real. Wiring is the cheapest way to find the screen you forgot to design.
Feel the transition timing
Feel the transition timing
Feel the transition timing
Duration is impossible to judge in a static frame. 300 ms felt right for a push and wrong for a modal, and only the prototype could tell me that.
Duration is impossible to judge in a static frame. 300 ms felt right for a push and wrong for a modal, and only the prototype could tell me that.
Duration is impossible to judge in a static frame. 300 ms felt right for a push and wrong for a modal, and only the prototype could tell me that.
Test the thumb arc
Test the thumb arc
Test the thumb arc
On mobile, the question is not whether the button exists but whether it can be reached one-handed on a 6.1-inch phone.
On mobile, the question is not whether the button exists but whether it can be reached one-handed on a 6.1-inch phone.
On mobile, the question is not whether the button exists but whether it can be reached one-handed on a 6.1-inch phone.
Make critique concrete
Make critique concrete
Make critique concrete
A clickable artefact gets specific feedback. A flat board gets opinions about colour.
A clickable artefact gets specific feedback. A flat board gets opinions about colour.
A clickable artefact gets specific feedback. A flat board gets opinions about colour.
EVALUATION
EVALUATION
EVALUATION
RAN IT
RAN IT
RAN IT
Moderated usability study — eight participants
Moderated usability study — eight participants
Moderated usability study — eight participants
8 participants
8 participants
8 participants
60 min
60 min
60 min
4 tasks
4 tasks
4 tasks
Five collectors who had spent £2,000+ on a secondhand item in the previous 24 months, plus three resellers running 50+ listings. Moderated remote, think-aloud, each task mapped to one objection from the research. Purchase confidence rose from 3.6 to 5.9 of 7 once participants reached the certificate.
Five collectors who had spent £2,000+ on a secondhand item in the previous 24 months, plus three resellers running 50+ listings. Moderated remote, think-aloud, each task mapped to one objection from the research. Purchase confidence rose from 3.6 to 5.9 of 7 once participants reached the certificate.
Five collectors who had spent £2,000+ on a secondhand item in the previous 24 months, plus three resellers running 50+ listings. Moderated remote, think-aloud, each task mapped to one objection from the research. Purchase confidence rose from 3.6 to 5.9 of 7 once participants reached the certificate.
8/8
8/8
8/8
Judge the price
Judge the price
Judge the price
8/8
8/8
8/8
Find under £30k
Find under £30k
Find under £30k
5/8
5/8
5/8
Explain escrow
Explain escrow
Explain escrow
7/8
7/8
7/8
Track a purchase
Track a purchase
Track a purchase
Task 3 failed, as I expected it to. Five of eight could not explain what happened to their money between paying and delivery - which drove the escrow money-flow diagram and a rewrite of the assurance block above the Buy button.
Task 3 failed, as I expected it to. Five of eight could not explain what happened to their money between paying and delivery - which drove the escrow money-flow diagram and a rewrite of the assurance block above the Buy button.
Task 3 failed, as I expected it to. Five of eight could not explain what happened to their money between paying and delivery - which drove the escrow money-flow diagram and a rewrite of the assurance block above the Buy button.
HOW IT WAS RUN
HOW IT WAS RUN
HOW IT WAS RUN
01
01
01
Recruited against a screener with verified purchase history, not self-declared interest — three applicants were rejected at this stage.
Recruited against a screener with verified purchase history, not self-declared interest — three applicants were rejected at this stage.
Recruited against a screener with verified purchase history, not self-declared interest — three applicants were rejected at this stage.
02
02
02
60-minute moderated remote sessions over four days, think-aloud, screen and audio recorded with consent.
60-minute moderated remote sessions over four days, think-aloud, screen and audio recorded with consent.
60-minute moderated remote sessions over four days, think-aloud, screen and audio recorded with consent.
03
03
03
Findings severity-rated on the same 1–4 rubric as the heuristic pass, so both sets landed in one backlog rather than two.
Findings severity-rated on the same 1–4 rubric as the heuristic pass, so both sets landed in one backlog rather than two.
Findings severity-rated on the same 1–4 rubric as the heuristic pass, so both sets landed in one backlog rather than two.
RAN IT
RAN IT
RAN IT
Heuristic evaluation — Nielsen’s ten, screen by screen
Heuristic evaluation — Nielsen’s ten, screen by screen
Heuristic evaluation — Nielsen’s ten, screen by screen
A structured pass across 8 desktop and 11 mobile screens, each issue severity-rated 1–4 — two Sev 4, four Sev 3, three Sev 2. The two Sev 4s: checkout would ship a five-figure item to an unverified address, now blocked inline before payment; and the zero-results state blamed the user and offered only a full reset, so it now names the filter that cost the most results with a one-tap widen.
A structured pass across 8 desktop and 11 mobile screens, each issue severity-rated 1–4 — two Sev 4, four Sev 3, three Sev 2. The two Sev 4s: checkout would ship a five-figure item to an unverified address, now blocked inline before payment; and the zero-results state blamed the user and offered only a full reset, so it now names the filter that cost the most results with a one-tap widen.
A structured pass across 8 desktop and 11 mobile screens, each issue severity-rated 1–4 — two Sev 4, four Sev 3, three Sev 2. The two Sev 4s: checkout would ship a five-figure item to an unverified address, now blocked inline before payment; and the zero-results state blamed the user and offered only a full reset, so it now names the filter that cost the most results with a one-tap widen.
HOW IT WAS RUN
HOW IT WAS RUN
HOW IT WAS RUN
01
01
01
Walked all 19 screens one heuristic at a time, not screen by screen — so findings came from a checklist rather than a general impression.
Walked all 19 screens one heuristic at a time, not screen by screen — so findings came from a checklist rather than a general impression.
Walked all 19 screens one heuristic at a time, not screen by screen — so findings came from a checklist rather than a general impression.
02
02
02
Logged each issue with the heuristic violated, the screen, what was observed, and a severity scored against a written 1–4 rubric agreed before I started.
Logged each issue with the heuristic violated, the screen, what was observed, and a severity scored against a written 1–4 rubric agreed before I started.
Logged each issue with the heuristic violated, the screen, what was observed, and a severity scored against a written 1–4 rubric agreed before I started.
03
03
03
Fixed each one in the file, then re-walked the affected flow to confirm the fix had not introduced a fresh violation elsewhere.
Fixed each one in the file, then re-walked the affected flow to confirm the fix had not introduced a fresh violation elsewhere.
Fixed each one in the file, then re-walked the affected flow to confirm the fix had not introduced a fresh violation elsewhere.
Impact & Results
Impact & Results
Impact & Results
PURCHASE CONFIDENCE · SCORED 1–7 BEFORE AND AFTER THE CERTIFICATE
PURCHASE CONFIDENCE · SCORED 1–7 BEFORE AND AFTER THE CERTIFICATE
PURCHASE CONFIDENCE · SCORED 1–7 BEFORE AND AFTER THE CERTIFICATE
+2.3
+2.3
+2.3
confidence lift
confidence lift
confidence lift
8/8
8/8
8/8
judge the price
judge the price
judge the price
8/8
8/8
8/8
find under £30k
find under £30k
find under £30k
5/8
5/8
5/8
explain escrow
explain escrow
explain escrow
7/8
7/8
7/8
track a purchase
track a purchase
track a purchase
The study measured one thing above everything else: whether proof changes willingness to buy. Confidence was scored 1–7 before participants reached the certificate and again after — it moved 3.6 to 5.9, and the participants who scored lowest at the start moved the most. That delta is the whole thesis: in this category the design job is making proof legible, not making the page prettier. The product has not shipped, so there is no engagement data here and none is invented.
The study measured one thing above everything else: whether proof changes willingness to buy. Confidence was scored 1–7 before participants reached the certificate and again after — it moved 3.6 to 5.9, and the participants who scored lowest at the start moved the most. That delta is the whole thesis: in this category the design job is making proof legible, not making the page prettier. The product has not shipped, so there is no engagement data here and none is invented.
The study measured one thing above everything else: whether proof changes willingness to buy. Confidence was scored 1–7 before participants reached the certificate and again after — it moved 3.6 to 5.9, and the participants who scored lowest at the start moved the most. That delta is the whole thesis: in this category the design job is making proof legible, not making the page prettier. The product has not shipped, so there is no engagement data here and none is invented.
Key Takeaways
Key Takeaways
Key Takeaways
01
01
01
Trust is a surface, not a policy page.
Trust is a surface, not a policy page.
Trust is a surface, not a policy page.
02
02
02
State the verdict — don’t make them compute it.
State the verdict — don’t make them compute it.
State the verdict — don’t make them compute it.
03
03
03
Publish the failures; they make the successes credible.
Publish the failures; they make the successes credible.
Publish the failures; they make the successes credible.
04
04
04
Compute contrast. Swatches lie.
Compute contrast. Swatches lie.
Compute contrast. Swatches lie.