Work / EMV® 3DS

Designing the moment payments feel safe

UX guidelines for payment authentication, now used by banks, merchants and card networks worldwide.

Role
Product designer
Client
EMVCo, with Ipsos
Shape
Two engagements, about six months each
The EMV 3DS challenge screen in the fictional BriteSpark brand: two numbered steps and one button to open the bank app.

What I did

  • Designed for the person mid-purchase, keeping every flow, screen and word focused on reducing anxiety at the moment of authentication
  • Helped build a multilingual design system so one template family could serve 20 flow variants and 80 test prototypes without drifting off-standard
  • Sat in on user testing across both engagements and turned findings into design changes, sometimes between sessions
  • Helped shape the research into the published UX guidelines, and updated the live site for the second edition

Where it landed

  • A single, evidence-backed answer to how authentication should look and behave, adopted as guidance worldwide
  • Guidelines that live as an interactive web product rather than a report nobody opens
  • emvco.com/3ds-guidelines

The three seconds that decide a sale

You're buying a coffee machine online. You hit "Pay" and a screen appears asking you to prove you're really you. In those three seconds you either feel reassured or you feel suspicious.

Three parties sit behind that screen. Merchants want conversion. Banks want security. Card networks want both. The person buying just wants their coffee machine, without feeling scammed on the way.

My focus throughout was that last person. Every decision came back to one question: does this make someone mid-purchase feel clearer and safer, or more anxious?

A person on a sofa approving a card payment in a banking app on their phone, with a clear two-step checkout screen.
The moment in question: a bank asking someone to confirm it's really them, mid-purchase.

Mapping the territory

In workshops with EMVCo I helped map all 20 payment flow variants. Visual journey maps got six competing card networks aligned quickly, because everyone could point at the same thing.

A Figma canvas mapping authentication flows column by column, with several collaborators' cursors on the same board.
Alignment, live on the canvas: card network cursors working the same board.

Most flows differ only by authentication method: a one-time passcode here, knowledge-based questions there. So the real design work wasn't in any single screen. It was a multilingual Figma design system flexible enough to spin up 80 prototype variants across six markets, and consistent enough that none of them drifted off-standard.

A Figma component properties panel showing header, notification, challenge type, variant and step options.
One component family, configured by property: type, variant, market, step. I built it alongside the other designer on the team, pairing on the complex flows and splitting the rest.
Real logos made people relax. So every bank, card network and merchant in the study was fake.

Designing honest tests

Recognising your own bank calms you down, which corrupts the data. To recreate genuine payment anxiety, every party in the study got a fictional brand. Even the cards were fake.

Hands holding a bank card for YourBank, a fictional bank created for the research.
YourBank, a bank that doesn't exist, on a card that does.

The first engagement validated existing flows with coded prototypes: realistic, but slow to change, so we piloted, updated, then ran full testing. The second explored unproven flows in Figma, which let us adjust copy and interactions between sessions as findings came in.

A user testing suite viewed through glass, with FigJam boards of session notes on screen.
I sat in on testing across both rounds. Findings were captured live in FigJam and ported to Airtable for the wider team.

Copy as UX

The smallest words carried the most weight, especially on unhappy paths. A payment failing mid-purchase is exactly when someone needs calm, clear instruction. With Ipsos' researchers I helped test and refine every notification and error state, front-loading the key details for scanning and clarity.

Before and after versions of the same YourBank error screen. Before: a "Payment security" heading over a vague technical message. After: the heading says "Please open your bank app to continue", with the explanation underneath.
The old heading named the problem. The new one tells you what to do, and the explanation moves underneath.

What it all came down to

Across both engagements, every finding kept resolving to the same four things. A challenge earns a customer's trust when:

  1. Control stays with the customer.
  2. Familiarity makes it feel expected.
  3. Efficiency keeps the purchase moving.
  4. Design quality signals it can be trusted.
The EMVCo guidelines site showing four guiding principles for trust in authentication.
Those four principles became the organising frame for the entire guidelines product.

From research report to living guidelines

Research at this scale produces mountains of findings that usually die in a PDF. Instead they became an interactive guidelines site, navigable and rich with prototypes.

This was often the longest part of each engagement: taking the Ipsos reports and shaping them into clear, usable guidance. For the second edition I made the template updates to the live site and published the revised recommendations.

The EMVCo guidelines site showing challenge anatomy and a WCAG accessibility checklist.
Accessibility sits inside the guidance itself: contrast checks right alongside the anatomy of each screen.

Guidelines, applied

The test of a template is what happens when a real brand moves in. The anatomy holds from research brand to real bank: same zones, same two-step structure, same hierarchy.

The secure checkout challenge screen in the fictional BriteSpark brand. The same challenge template applied to Starling Bank's brand, as a concept.
Same anatomy, dressed for the real world. My Starling concept surfaces the amount and card, the way challenger banks do, so the customer can see exactly what they're approving.

The takeaway

The right amount of friction isn't zero. In testing, a pause of three to five seconds worked best. When security is on someone's mind, a well-designed pause builds trust. Complex systems don't need to feel complex. They need to feel considered.

Say hello.

Got a product that's harder to use than it should be? I'd love to hear about it.

[email protected]LinkedIn