Nate DouglasSenior Product Designer, home

Rebuilding how millions of drivers pay a toll.

  • Pay types
  • Dashboards
  • Login
  • Personal plans
  • Support
  • Correspondence
Role
Senior product designer
Team
Led a team of three designers, then the only designer on E-ZPass in 2026. Worked with 5 PMs and 40+ engineers
Timeline
June 2023 to June 2026
Tools
Figma, Claude Code and Cursor for coded prototypes and the entry alternatives, Maze with Respondent for testing, Power BI and Storybook.
Platforms
Web and native mobile, NY and NJApp StoreGoogle Play
62% → 94%correctly understood payment statusMaze tests, 50 people per round
68% → 92%found their balance and chose a payment method without helpThe same two test rounds
36 → 14payment-status support contacts per 1,000 paymentsPower BI, 90 days before and after launch

Mine: the design and the Apollo design system. Not mine: the build, the follow-up test round and the support data.

How these were measured

Error recovery: in prototype testing of the validation, confirmation and error-recovery flows, recovery without help rose from 58% to 92% and median recovery time fell from 92 to 54 seconds. Percentages above: baseline and follow-up usability testing; another group ran the follow-up round. The 7 of 15 in What I found comes from a separate study of 15 customers, so it isn’t the same measure as the 62% baseline. Support contacts: post-launch data shared by the E-ZPass program team. The 45 seconds to under 12 comes from my own timed walkthroughs (same task, same account, three runs per flow), not customer testing. Tolls NY, the app this shipped in, has 3M+ downloads; that is the product’s reach, not a result of my work.

The redesigned Tolls NY home page: Pay Toll Bill, Log in, Open an account and E-ZPass On-the-Go over a bridge photo

The short version

I design what a product does when it isn’t sure.

On E-ZPass, that moment came right after Pay: did it go through, and is it safe to pay again? Every chapter designs for it: payment status, guided billing, which shipped, and two AI assistant concepts that didn’t ship.

The rule behind every screen: data that isn’t confirmed must never look like a fact. See it in three states

Problem

A service people can’t avoid, with an interface that hadn’t kept up.

E-ZPass isn’t optional for most Northeast drivers, so when it’s confusing, they’re stuck.

  1. Hard to find. Balance and payment methods were buried.
  2. Hard to read. Plans were codes like MCC and NRC.
  3. Hard to trust. After paying, people had to guess whether it went through.
The legacy E-ZPass payment flow in five screens: account overview, plan selection, payment setup, review and payment history
Where it started: the legacy payment flow, five screens from account overview to payment history.
What I found
Paying wasn’t the hard part.

People could find Pay. They couldn’t tell what happened after they pressed it, so the redesign centers on what people see after they pay.

7 of 15 customers we studied on the legacy site couldn’t tell if their payment went through.

It ran as a loop.

Every step fed the next one. Each step links to the chapter that shows it.

1Understand

Hands offOne problem, with success metrics

1 · Understand

Scope and cross-check

Audit the inputsTickets, analytics and stakeholder goals in one place, to find the single highest-impact problem.

Set guardrailsSuccess metrics up front, so the scope can’t creep.

Tools

See it: Finding the problem, Prioritizing

2Explore

Hands offOne tested direction engineering can build

2 · Explore

Explore wide, with engineering

Start with engineeringSketch rough flows with the tech lead and check feasibility on the spot.

Prototype and testClickable prototypes with slow and unsure data built in, then tests with real users, run in Maze with people recruited through Respondent.

Tools

See it: Exploring options, Tradeoffs, Working with engineering, Guided billing

3Build

Hands offA release built from Apollo

3 · Build

Lean production

Use the systemBuild with Apollo components and its shared values, so no time goes to redrawing layouts.

Define version 1Release when the core flow unblocks people without adding risk.

Tools

See it: Ready for build, Design system, Handoff, Design QA

4Learn
4 · Learn

Keep iterating

Track and adaptWatch drop-off and support contacts against the goals set up front.

Pick the next problemWhatever is still left over goes back to step 1.

Tools

See it: After launch

Hands off

Live data picks the next problem · back to step 1, every release

What I cut
One long form on a phone.

The legacy site split paying across several screens. The agency asked for one page. On a phone it was too long, so each screen got one decision: pick the bill, pay, done.

Legacy E-ZPass payment screens: payment setup, review payment, and payment history
Legacy: paying took several separate screens.
The one-page toll bill payment, full length
Scroll the page ↓
The one page the agency asked for: one long scroll on a phone. Scroll it.
Redesigned mobile payment: Pay Toll Bill to pick a method, then Add Checking Account, one decision per screen
Cut down: one decision per screen.
Wireframes: the consolidated payment page beside the same payment split into choose payment, bank account, and review and pay
Wireframes: the one consolidated page beside the same payment as one decision per screen.
Where people start

Starting from what people came to do.

People arrive for different jobs: account holders want their balance, bill payers want to pay a notice.

Compared
Account first, sign-in with bill lookup below, or task first.
Chose
Task first. Bill payers start without an account; account holders keep a direct sign-in.
Cost
Carrying the chosen task through navigation and sign-in.
Agreed
I walked a small group, including two engineers, through the options. Task first was kept.
Tolls NY start screen with sign-in first and a create account button
Option AAccount first: sign in before anything else.
Tolls NY start screen with sign-in at the top and pay a toll bill below it
Option BSign-in on top, bill lookup below it.
Tolls NY start screen asking how can we help, with pay or find a toll bill, manage my E-ZPass account and get E-ZPass
ChosenTask first: pay a bill, manage an account, or get E-ZPass.
Decisions

Three decisions that shaped it.

  1. Account analytics by area, with payments and statements flagged
    Situation
    The account area covered sign-up, payments, statements, plans and profile. We couldn’t do it all at once.
    Call
    Start with payments and statements, where the analytics showed the clearest drop-off and failures cost the most.
    Cost
    Sign-up, plans and profile waited, even though plan names confused people too.
    Result
    The data decided the order of the work.
  2. The legacy payment flow: find the action, enter payment, unclear outcome
    Situation
    It looked like people couldn’t find things. Research showed a trust problem: they couldn’t tell what happened to their money.
    Call
    Status, balance change, next steps and recovery go at the center. One decision per screen on the phone.
    Cost
    More screens.
    Result
    A one-time bank payment went from about 45 seconds to under 12 in my timed walkthroughs.
  3. Auto-replenishment authorization, legacy beside the approved redesign
    Situation
    The bank authorization wording couldn’t change.
    Call
    Same words, new layout: a plain summary first, full terms below, consent next to what it authorizes.
    Cost
    The full legal text stayed on the page.
    Result
    Approved with the wording unchanged, and without the wall of text.

Guided billing over open chat.

Guided billing beat open chat in prototype tests, so guided shipped. It shows whether the payment went through and what to do next, without people needing to know what to ask.

What the AI does

  1. People askin their own words, or by picking a choice
  2. The model structures itinto choices: which bill, which payment, what they want to know
  3. People confirmor adjust the choices
  4. The records answerpayment match and bank status come from the account’s records, not the model
Guided billing help: the $161.50 bill and three choices, did my payment go through, what amount is on my bill, I can’t find my payment
Guided billing starts by asking what to check.
Side by side: the guided result saying the payment is processing and not posted, and Dara’s chat panel giving the same answer
Left: guided billing, which shipped. Right: Dara, the open chat version I designed and tested against it. Guided won, so Dara didn’t ship.
Before and after

Same tasks, rebuilt.

These are just a sample of the full redesign.

The whole payment journey

Five screens from account to payment status, before and after.

Five-screen legacy E-ZPass journey showing account overview, plan selection, payment setup, payment review, and payment history
Before: the screens followed how the system was built.

Screen by screen

A handful of the roughly 3,000 screens in the redesign, across web, native and responsive. Pick a task to see the legacy screen and the redesign together.

  1. Before: the legacy home page After: the redesigned home page with four entry points

    The four main tasks are up front, and alerts got their own spot.

  2. Before: a coded plan table After: plans as plain-language cards

    The coded rows became cards with a real plan name, what it covers and the price.

  3. Before: raw bank fields under legal text After: saved payment methods listed up front

    Saved methods show first with the primary one marked, and there’s one way to add a bank or card.

  4. A dense legacy E-ZPass dashboard on desktop and squeezed into a mobile viewport The redesigned E-ZPass dashboard on desktop and intentionally responsive mobile

    Status comes first, there’s one main action, and the controls are big enough to tap.

    Before: the legacy website, and the same page squeezed onto a phone, tables and tiny actions included.

    After: the redesigned website, with a phone layout designed from the start.

  5. Pay Toll Bill, amount due Add Checking Account, filled Statements on mobile

    Every step follows the same pattern.

    Amount due up top, payment methods below.

    One field per line, with a check diagram where people need it.

    Statements are searchable, and any period is one tap away.

Design system

The Apollo design system, rebuilt.

I rebuilt Conduent’s Apollo design system myself. For the first two to three months it took about half my time.

14
button styles found in an audit of the account flows
9
input treatments in the same flows
Weekly
updates, sometimes daily
Button and input audit of the E-ZPass account flows: button roles and states, and input treatments grouped by behavior
The audit: every button and input in the account flows, grouped by what it does.
  1. The Add Checking Account screen for E-ZPass New York and E-ZPass New Jersey side by side. The name, routing and account number fields are one shared component; only the authorization text below them differs between New York and New Jersey.

    One component. Only the legal text changes between New York and New Jersey.

    Fields, validation and errors work the same way. Only the authorization language differs, so we never kept two versions.

  2. Apollo text input states from the form spec: placeholder, enabled, focus, active, error with the message replacing helper text, warning with an icon, disabled and loading

    The frontend lead and I reviewed the working, coded components together, not just design specs.

    We went through every state of an input together. Design and code pulled color, type and spacing from one shared list of values.

    Input states from the Apollo form spec.

  3. A visual test review for the text input error state: baseline, new and difference views show the error message moved 9 px below the field, with Accept and Deny buttons. Other input states pass.

    Automatic screenshot comparisons ran on every component change.

    40+component screens changed by one spacing update in early 2024, which exposed three places where values had been typed in by hand.

    A difference wasn’t always a bug, so we reviewed it before accepting. Keyboard use, validation and the full payment flow we still checked by hand.

  4. Apollo UI Guide accessibility page: a table of WCAG 2.2 AA requirements in every component spec (contrast, labels, keyboard access, visible focus, target sizes, error messages) and a do and don't for error messages, where a red border alone is not enough.

    Accessibility requirements, from the WCAG 2.2 AA standard, were written into each component spec.

    Contrast, labels, keyboard access, visible focus, target sizes and error messages. A red field wasn’t enough: the message had to say what was wrong and which field to fix. Designed and tested to that standard, not certified.

  5. Four E-ZPass mobile screens with one failure pattern, each split into what happened, what it means and what to do: an expired card with Update card; a returned bank payment with the bank's reason, Fix bank details and Use another method; a failed auto-replenishment showing the balance, Fix payment method and Make a payment; and a payment still being confirmed, with only Check payment status so nobody pays twice.

    Every failure says what happened, what it means for the payment, and what to do next.

    Declined cards, returned bank payments, expired methods and failed auto-replenishment share one structure. If the status is unknown, people check it before paying again.

  6. New York and New Jersey bank account screens: the New Jersey authorization became a content variant of the shared form.

    Every request answered three questions: what’s the problem, where does it show up, and why doesn’t the existing component work?

    1. Reuse what exists
    2. Adapt an existing one
    3. Build something new

    The frontend lead and I made the call together.

Live data picks the next problem.

Launch isn’t the end. I check the goals I set up front, then look at what’s still left over.

Payment-status support contacts fell from 36 to 14 per 1,000 payments, from post-launch data shared by the E-ZPass program team.

Checked against the goals set up front.

The same three goals, written before the first wireframe. They are the three numbers at the top of this page.

GoalTargetResult
People knowCan say whether a payment went through, without help62% → 90%+94%
People find it themselvesFind their balance and choose a payment method without help68% → 85%+92%
Fewer people have to askPayment-status support contacts per 1,000 payments36 → under 2014
Result3 of 3 met
Looking back

Payment status needs to be more than one screen.

What didn’t work

Moving the Pay button and making it more visible. Nothing really changed, because people could already find Pay. That pointed us to what happened after they pressed it.

What I’d change

Test rough end-to-end flows in week one. The problems were in the handoffs between steps, not the screens.

What I’d do next

Payment status as a full system. Seven states, each saying what it means for the balance and whether it’s safe to try again.

What I’d measure next

  • How often Processing ends in rejected or returned, and whether people found out in time.
  • How often guided billing hands off to a person, and whether the agent has to ask again.
  • Whether people still come back within a day to check.
Unshipped concept: designing trust controls for an E-ZPass AI assistant
Leadership

Three designers, 5 PMs, 40+ engineers.

How each group got to yes.

  • The agency

    Asked for one page. On a phone it was too long, so it became one decision per screen.

    What I cut
  • Engineering

    I walked a small group, including two engineers, through the entry options. Task-based entry was kept.

    Where people start
  • Legal

    The bank wording couldn’t change. Approved with the legal wording unchanged.

    Decisions
  • Frontend

    We reviewed the coded components and made new-component calls together.

    Design system

Who made which call

My calls
  • Starting with payments
  • Status at the center, one decision per screen
  • Layout of the fixed legal text
  • Testing guided billing against open chat
Made together
  • Task-based entry, with engineering
  • Version 1 status, with the backend lead
  • New components, with the frontend lead
Not mine to make
  • Bank authorization wording
  • The bank’s nightly confirmation file
  • Final sign-off, by MTA leadership
What people said

From people I worked with.

Saurabh Prabhat
“He approached problems thoughtfully, collaborated effectively with colleagues, and communicated design decisions with clarity. I valued his ownership of projects…”
Saurabh Prabhat, Senior Global Director of AI and UX at Conduent · my manager Read the recommendation ↗