Partner Identity Flow
Raising identity completion on a partner checkout
A comprehension problem wearing the costume of a conversion problem.
YEAR
2023–2024
COMPANY
Verimi
Status
Shipped · two iterations, live in 2024
Usability research
Identity verification
Onboarding
Conversion
Checkout UX
Interaction design
Plain language
Design system
B2B2C
WHAT
A high-traffic partner checkout where customers verify their identity to complete a purchase. Completion at the identity step was low.
INDUSTRY
B2B2C identity verification · unregulated
Role
Product Designer (title at the time). Ran the usability research and designed the Verimi side of the flow across two shipped iterations
Team
Product owner (owned the funnel data), engineering, one designer (me)
TIMeline
Research December 2023 – January 2024 · design completed September 2024 · optional account creation live later in 2024
Key decisions
Reframed the drop-off from a step-count problem to a comprehension problem · fixed clarity first and measured it on its own · made account creation optional with a checkbox, not a toggle
result
Completion rose from about 1 in 5 users to about 1 in 3 · clarity carried most of the lift
The three flows. Before, the email code step also created an account automatically. Iteration one kept the same steps with clearer UI and copy. Iteration two made login optional for new users, removed the email code step and moved an optional save to the end. Partner shown as a demo brand.
The problem
A large partner had integrated Verimi so its customers could verify their identity and finish a purchase. The funnel showed the problem in one line: most people who reached the first identity screen left before they finished. Only about 1 in 5 completed. The easy explanation was friction: too many steps, too much to do. The sessions pointed somewhere else.
Five sessions, one pattern
I ran five in-person usability sessions during discovery, and the same pattern showed up every time. People couldn't tell logging in from identifying themselves. Continuing also created an account with a second company, and people sensed it: the fear of being signed up by someone else was enough to make them hesitate or leave. The "Continue" button didn't say what would happen next. By the final step people were tired, and didn't realise they still had to send their verified identity back to the partner. On mobile, people skipped the dense text entirely.
The drop-off was a comprehension problem wearing the costume of a conversion problem.
Iteration one: clarity
I made each step legible before changing anything structural. I cut the copy, fixed layout and spacing, clarified the calls to action, removed decorative elements and moved the screens onto the design system. I also put the partner's logo into the Verimi screens, so people could see they were still in the purchase they started.
Completion rose to about 29%. That confirmed the diagnosis: clarity on its own carried most of the gain.
Iteration two: optional account creation
The fear of forced registration was still there, so I made login optional for people without an account: they could go straight from the partner to identification. Account creation became optional too, offered at the end of the flow, after verification. That also removed a step: the email confirmation code, which had also created the account, was gone, and seven screens became six. I used an unticked checkbox ("save identity"). A checkbox reads as one optional action with no immediate consequence, the same pattern people know from saving card details at checkout. A toggle suggests something changes in the system straight away, which wasn't true here, so it would have misdescribed the control.
The save option sat on the most expensive screen in the flow. By then we had already paid for the identification, but the partner only paid us once the user sent their data across. A drop-off here cost twice. So saving had to stay optional and quiet, and "Next" had to stay the obvious action.
I rewrote the copy in plain language (B1, the level written for a broad audience) at the same time. The change went live on this flow later in 2024, and completion reached about 32%.
Four ways to offer saving the identity after verification. The unticked checkbox at the end of the flow shipped.
What changed
Completion moved from about 1 in 5 (the low twenties percent) to about 29% after the clarity iteration, then to about 32%, roughly 1 in 3, after optional account creation went live. Two iterations, and the first carried the larger share. There was no controlled comparison, so the second step is a sequence, not proof of cause.
Optional account creation was a partner-specific setup at the time. In 2026, LEAN ID made it the default for every partner.
Completion rose from about 1 in 5 to about 1 in 3 over two shipped iterations. Clarity carried most of the lift; optional account creation added the rest. Figures are approximate and not from a controlled test.
Honest limits
Part of the journey stayed outside my reach: the external provider's identity screens and the partner's own entry screen. Provider and integration issues added noise to the tracking, so I read the numbers as direction.
The funnel analysis belonged to the product owner. I ran the research and designed the flow, and used the data as input.
A third iteration for the final hand-back step was designed but not built, so it has no data. Figures generalised for confidentiality.
What I’d do differently
Run the sessions before the first build. The comprehension problem was findable earlier, and it would have shaped iteration one instead of being confirmed by it. I'd also instrument the final hand-back step from the start, so the third iteration could be measured instead of argued for.
In one sentence
A funnel shows where people leave. Five conversations showed why, and did more for this flow than the dashboard.