Suppliers sell through big online stores by sending in details about each product, like how many units fit in a box. If one of those numbers quietly changes, orders go wrong and nobody knows why. I spent 12 weeks finding out why it happens, and designing a way for suppliers to catch it early.

Company

Amazon

Role

UX Design Intern

Timeline

Summer 2025

The problem

Suppliers only found out their product data was wrong after an order had already failed.

What I designed

A tool that warns suppliers about data problems, shows what changed, and lets them fix it themselves.

My role

I led the research and designed the experience from start to finish.

Where it landed

Four connected design concepts covering the whole journey, from spotting a problem to fixing it, refined with feedback from internal teams.

To respect a confidentiality agreement, names and details are generalized, and every visual is a simplified recreation, not a real screen.

01 · The problem

THE CRIME SCENE

Orders were failing, and nobody could say why.

Every product comes with a few key numbers: how many units are in a pack, how many packs are in a box. The store’s ordering systems use those numbers to decide how much to buy.

Sometimes one of those numbers changed in the background, without the supplier doing anything. Here’s how that plays out:

DAY 1

A box size changes from 12 to 1. Nobody is told.

DAY 3

The next order asks for the wrong amount.

DAY 10

The warehouse can’t match what arrives.

DAY 21

The supplier finds out from a failed order.

An illustrative example, not real data.

Fixing it meant emailing support and waiting while the request was passed between teams. That often took weeks.

The question I set out to answer

How might we help suppliers spot, understand and fix product data problems before they cause failed orders?

02 · Research

THE WITNESSES

I talked to the people dealing with it every day.

I started by reading existing research and past feedback. Then I interviewed suppliers who sell very different kinds of products, plus the internal teams who help them when something goes wrong.

Their stories matched, almost word for word.

TOO LATE

Every supplier I spoke with found out about problems after something broke. Never before.

OUT OF NOWHERE

All of them had seen their product numbers change without doing anything.

WEEKS

Getting a fix through support usually took weeks, sometimes much longer.

Detective’s note

Some teams spent part of every week just double-checking that nothing had changed. Not fixing anything. Just checking.

03 · Key insight

THE MOTIVE

Suppliers owned the data, but couldn’t see the version that mattered.

Suppliers create the product details, so they know them best. But the numbers actually used to place orders lived in systems suppliers couldn’t see. When the two didn’t match, the supplier was always the last to know.

What the supplier sent

12

units per box

the blind
spot

What ordering actually used

1

units per box

This changed the goal. Better error messages weren’t enough. Suppliers needed to see their own data, and have the power to fix it.

Three rules I designed by

01

Finding a problem shouldn’t be a full-time job.

02

Problems should show up before they cost money.

03

Suppliers should be able to fix their own data.

04 · What to build

THE PLAN

Three jobs the tool had to do.

I turned what I heard into three simple jobs, each written from the supplier’s point of view.

As a supplier,

Warn me when something changes

so I can fix it before it hits an order.

As a supplier,

Show me what changed, and who changed it

so I can find where the problem started.

As a supplier,

Show the costliest problems first

so I spend my time where it matters.

BEFORE

weeks of emails

An order fails

Work out which product is wrong

Email support

Wait while it’s passed between teams

Check the fix actually worked

AFTER

one place

Get a warning

Compare your number with the system’s

Fix it

Track the update

05 · Design decisions

THE FIX

Four design decisions, and the trade-offs behind them.

Each decision meant giving something up. Here’s what I chose, and why. The visuals are simplified close-ups, not real screens.

DECISION 1

Show what matters first, not everything.

The trade-off

Suppliers manage huge catalogs. Showing every detail felt thorough, but it buried the urgent problems.

What I chose

Lead with problems sorted by how much they could cost. People can still add more detail when they want it.

Needs attention

Sorted by impact

Product A

Box size doesn’t match

Product B

Missing barcode

Product C

Box size doesn’t match

DECISION 2

Put both numbers side by side.

The trade-off

Explaining exactly why a number changed would mean exposing complicated systems behind the scenes.

What I chose

Show “what you sent” next to “what’s being used,” plus who changed it, in plain words.

Units per box

You sent

12

Being used

1

Changed 3 days ago by an automatic update

DECISION 3

Let suppliers fix it, and be honest about what happens next.

The trade-off

Fixing in place is faster, but other updates can still change the number later. Hiding that would break trust.

What I chose

Let suppliers fix it right there, show the fix’s progress, and warn them if it ever changes again.

Your fix

Units per box: 1 → 12

Sent

Updating

Live

Other updates may change this later. We’ll tell you if they do.

DECISION 4

Make fixing many products at once safe.

The trade-off

Bulk edits save time, but one typo can spread across dozens of products.

What I chose

Show the per-box totals updating live, so mistakes are obvious before anything is saved.

How it adds up

Case

Pack

6 per pack

× 12 packs

72 per case

Totals update before you save.

06 · Outcome

THE VERDICT

Case closed: from chasing problems to catching them.

By the end of my internship, I delivered a complete set of design concepts covering the whole journey, from spotting a problem to fixing it. I refined them through rounds of feedback from internal teams and presented them to the team.

Here’s how the new experience compares with the way suppliers dealt with problems before.

BEFORE

AFTER

Finding problems

Found out from a failed order or a customer complaint.

→

A dashboard flags problems early and puts the costliest ones first.

Understanding them

No way to see the numbers the system was actually using.

→

What you sent and what the system is using, side by side.

Fixing them

Email support, then wait while the request moved between teams.

→

Fix the value yourself, right where you found it.

Fixing many at once

One product at a time.

→

Update a whole selection at once, with totals checked before saving.

Knowing what changed

Numbers changed with no explanation.

→

A full history of what changed, when and by whom, with undo.

Knowing it worked

Wait for an email to confirm.

→

A status tracker shows the fix’s progress, with honest notes on what to expect.

The new design turns finding out after the damage is done into spotting problems early and fixing them on your own.

What I learned

01

Complex systems need simple, consistent patterns.

When the system behind the scenes is complicated, the screens in front of it should feel familiar everywhere.

02

Showing data isn’t enough.

People act when they understand the context and the impact, not just the number.

03

Meet people where they already are.

It works better to build on how suppliers already think than to ask them to learn a new way.

Next case

PIXELNAMICS

A storefront that finally sells →