Amazon

•

5 min

From screws to $3B: designing Amazon's B2B marketplace from scratch.

Founding UX member · Employee #6 · Bauhaus Design · Radical Research

Role:

Sr. UX Designer / Manager

Team:

12

$

40

B

Revenue between 2015-2016

$

40

B

Revenue between 2015-2016

400
770

%

Repeat purchase rate at 90 days

400
770

%

Repeat purchase rate at 90 days

3000

K

Shadow buyers converted in 8 weeks

3000

K

Shadow buyers converted in 8 weeks

2300

K

Healthcare practices unblocked through research

2300

K

Healthcare practices unblocked through research

Challenge

Amazon's browse, search, and checkout were built for consumers and we lacked the organizational purchasing infrastructure to support B2B customers.

strategy

Working directly with buyers, I made the research, design, and architectural decisions that designed Amazon's B2B purchasing experience and connected it to their procurement systems. To scale that work, I led the integration.

Results

Large B2B customers could now fulfill operational orders on Amazon. Repeat purchases climbed from 40% to 77%. Annual B2B revenue reached $3 billion. The design patterns scaled across 90% of Amazon's catalog and lifted standards for the consumer experience.

Challenge

When "one-size-fits-all" fails.

Amazon's default acquisition strategy of retooling existing stores with retail patterns had become a liability. A mixture of qualitative and quantitative research revealed that B2B customers saw Amazon as a stopgap, not a procurement partner. While our patterns made products discoverable, our onsie-twosie shopping experience and lack of procurement support meant higher-than-normal returns.

Challenge

When "one-size-fits-all" fails.

Amazon's default acquisition strategy of retooling existing stores with retail patterns had become a liability. A mixture of qualitative and quantitative research revealed that B2B customers saw Amazon as a stopgap, not a procurement partner. While our patterns made products discoverable, our onsie-twosie shopping experience and lack of procurement support meant higher-than-normal returns.

"It feels like I'm shopping for airplane components at Toys R Us."

— Business buyer, Amazon Supply research session

After 6 weeks of visiting warehouses, shops, and restaurants, my journey maps revealed that for small, medium, and large businesses, shopping accounted for less than 5% of how they stocked their storerooms and factory floors. To become a trusted B2B supplier, Amazon needed to focus beyond the shopping cart.

Download to See Full Size

Recreation of the procurement swimlanes used to illustrate B2B scope

Strategy

A purchasing platform built around how businesses work.

My work focused on 3 parallel tracks:

  1. Developing new B2B shopping patterns optimized for error-free purchasing.

  2. Creating accounts that unlocked permission-managed purchasing for companies with 1 to 1000+ buyers.

  3. Protecting the privacy of over 500K private Amazon accounts that would need to migrate to corporate accounts.

Strategy

A purchasing platform built around how businesses work.

My work focused on 3 parallel tracks:

  1. Developing new B2B shopping patterns optimized for error-free purchasing.

  2. Creating accounts that unlocked permission-managed purchasing for companies with 1 to 1000+ buyers.

  3. Protecting the privacy of over 500K private Amazon accounts that would need to migrate to corporate accounts.

A purchasing experience built for professional evaluation

B2B buyers did not shop the way consumers browsed Amazon. Fasteners alone could have 50+ variations, yet the product page treated them almost identically.

I redesigned the experience around how professionals actually evaluated parts: 80% larger images, specification-first layouts, and comparison patterns built for operational purchasing.

Decision time dropped from 6–10 seconds to 3–5. Returns fell from 10% to 6%. Many of these patterns later spread across Amazon.

A purchasing experience built for professional evaluation

B2B buyers did not shop the way consumers browsed Amazon. Fasteners alone could have 50+ variations, yet the product page treated them almost identically.

I redesigned the experience around how professionals actually evaluated parts: 80% larger images, specification-first layouts, and comparison patterns built for operational purchasing.

Decision time dropped from 6–10 seconds to 3–5. Returns fell from 10% to 6%. Many of these patterns later spread across Amazon.

Side-by-side comparison of key features/patterns Amazon Retail and Amazon Supply browse patterns (2011)

Results

My resulting new B2B design patterns reduced returns by 40% (10→6%), and task completion times by 45% (9s→4s). The radical elimination of color from all but links and product images shocked many, but it performed so well that it later inspired Amazon's "beacon" retail redesign a year later.

Results

My resulting new B2B design patterns reduced returns by 40% (10→6%), and task completion times by 45% (9s→4s). The radical elimination of color from all but links and product images shocked many, but it performed so well that it later inspired Amazon's "beacon" retail redesign a year later.

Amazon - 2011
Amazon Supply

Side-by-side comparison of Amazon Retail and Amazon Supply (2011)

A scalable organizational account architecture

Amazon was built around one email and one account. Companies could not purchase that way. Before organizations could route orders through procurement systems, the underlying account model had to be rebuilt.

Working closely with legal, I helped architect the migration strategy to ensure employees’ personal purchase histories were never exposed to employers. In 8 weeks, 300K shadow business buyers either converted to or were merged into business accounts.

A scalable organizational account architecture

Amazon was built around one email and one account. Companies could not purchase that way. Before organizations could route orders through procurement systems, the underlying account model had to be rebuilt.

Working closely with legal, I helped architect the migration strategy to ensure employees’ personal purchase histories were never exposed to employers. In 8 weeks, 300K shadow business buyers either converted to or were merged into business accounts.

Example workflow illustrating the in-depth logic behind the registration/account merger process

Hard Decisions

Building an account model that could scale

Business Admin accounts: Launch fast or build what lasts?

Under pressure to launch, I made a difficult call with engineering: ship phase one using a hierarchical account model so we could uncover onboarding and support problems at scale while building the object-oriented version in parallel. Phase two is still in use today.

Hard Decisions

Building an account model that could scale

Business Admin accounts: Launch fast or build what lasts?

Under pressure to launch, I made a difficult call with engineering: ship phase one using a hierarchical account model so we could uncover onboarding and support problems at scale while building the object-oriented version in parallel. Phase two is still in use today.

Download to See Full Size

Model comparison "white paper" circulated to inform phased account manager launch.

Restricted certifications

Don’t ask users for data we can verify ourselves

Healthcare providers originally had to upload photos of their licenses to create accounts. The result was catastrophic: 87% of registration attempts were abandoned.

Working with legal, I showed that almost all of the certification data being requested already existed in public records. We reduced physical proof requirements to the single state that legally required it. Average verification time dropped from 3 days to 15 minutes, unlocking 230K healthcare provider accounts.

Restricted certifications

Don’t ask users for data we can verify ourselves

Healthcare providers originally had to upload photos of their licenses to create accounts. The result was catastrophic: 87% of registration attempts were abandoned.

Working with legal, I showed that almost all of the certification data being requested already existed in public records. We reduced physical proof requirements to the single state that legally required it. Average verification time dropped from 3 days to 15 minutes, unlocking 230K healthcare provider accounts.

Example flow diagram illustrating the steps, processes, times, and emotional impact of our restricted & regulated (R&R) product certification.

Results

A new standard for Amazon

Large organizations could now place operational orders through Amazon. The patterns we built for professional buyers were later formalized into the Business Interface Guidelines and spread across Amazon’s broader shopping experience, influencing browse, search, and ordering patterns across the catalog.

More than 100 of the patterns I developed still live on at Amazon, including the inline image filter I created, which reduced load overhead by 80% and later became a standard search feature. Phase two of the account architecture is still in use today.

Results

A new standard for Amazon

Large organizations could now place operational orders through Amazon. The patterns we built for professional buyers were later formalized into the Business Interface Guidelines and spread across Amazon’s broader shopping experience, influencing browse, search, and ordering patterns across the catalog.

More than 100 of the patterns I developed still live on at Amazon, including the inline image filter I created, which reduced load overhead by 80% and later became a standard search feature. Phase two of the account architecture is still in use today.

Accounts
2000

K

10

M

Accounts were activated during year 2 through onboarding efforts.

Accounts
2000

K

10

M

Accounts were activated during year 2 through onboarding efforts.

B2B Scope
900

%

% of Amazon's catalog covered by B2B design patterns

B2B Scope
900

%

% of Amazon's catalog covered by B2B design patterns

Shopping Behavior
400
770

%

Repeat purchase rate at 90 days using history/peer activity

Shopping Behavior
400
770

%

Repeat purchase rate at 90 days using history/peer activity

Revenue

$

30

B

$1B in annual revenue in year 1, $3B in year 2.

Revenue

$

30

B

$1B in annual revenue in year 1, $3B in year 2.

Accelerating global COVID-19 research with AI

IQVIA

•

7 min

Accelerating global COVID-19 research with AI

IQVIA

•

7 min

Replacing feature delivery with faster threat response

FINRA

•

5 min

Safeguarding 162M investors by revealing risks.

FINRA

•

6 min

Safeguarding 162M investors by revealing risks.

FINRA

•

6 min

© 2026 James Ricks
© 2026 James Ricks