Every one of Macquarie's 400,000 personal banking customers, about 60% of them online, lands on the dashboard after signing in. In 2021 I redesigned it, from a year of feedback through nine rounds of usability testing to an A/B test, and built its components into OneUI, Macquarie's design system.
ease of use, A/B tested with 1,831 customers
final dashboard task success
participants over nine test rounds
The problem
A banking dashboard should show customers where they stand, then get out of the way. Macquarie's had grown by addition: every team that wanted a place got a tile, and by 2021 there were nine in a rigid grid. Checking a balance sat at the same weight as a Qantas rewards card.
The dashboard drew 13.9 million views a month, and most visits left straight away.
Where customers went next, in one month
Three quarters of sessions reached the second row of tiles, and only half the third.
With the analytics team I pulled twelve months of complaints from Medallia, CX&O call notes and app store reviews together, and clustered them by theme.
A year of dashboard feedback, by theme
The affinity map, in customers' words
I have to scroll to find my accounts
31% of commentsI find it difficult to get to where I want to be… which is get to my accounts
Not sure why when I log in it goes to the dashboard. It isn't all that helpful
Cannot find what I am looking for… navigation
Too many steps to pay or transfer
24% of commentsWish I can make a quick payment
It's confusing how to make a payment. I have to go through multiple steps
Tiles I never use
19% of commentsI don't really care about the Rewards
Spending categories don't work too well
Want a quick way to look at the budgets
I don't like seeing the negative balances, not sure why I need 'net wealth'
Balance or available, which is it?
14% of commentsWas that login me?
12% of commentsWant to ensure that my account is secure
My role
I was the product designer for the Personal Banking dashboard, working with a product owner, a tech lead and a squad. I led research, design, testing and accessibility, and product-owned OneUI, so every component the dashboard needed could be built once for every team. The design team grew from two to five.
Research
Macquarie's personas described who customers were, not what they did. With the analytics team I built goal-directed personas from data instead, plotting 390,559 clients in Tableau by digital engagement and products held.
| Low product holding | High product holding | |
|---|---|---|
| Highly active online | 27,193 | 124,821 |
| Rarely online | 115,338 | 123,207 |
Of seven personas, two shaped the dashboard most, because each needed nearly the opposite thing first.
Everyday banking · the largest group
Administrative officer, married
He is saving for a home loan and worries most about missing a bill.
Wealth · fewer customers, higher value
Private banking client, married
She is confident online and wants to see and protect her whole wealth at a glance.
One-on-one interviews
Survey responses, aged 18 to 75+
Dashboard views in Google Analytics
All 11 interviewees had a transaction account, 8 had savings and 2 a home loan. They signed in with a purpose: check a balance, see whether they've been paid, then pay or transfer. The surprise was security. Several wanted to see the last login, to check nobody else had been in.
QWhen you first log in, what would you like to see first?
Account balance for each of my accounts.
My account balance, and maybe my most recent transaction. Have I been paid?
A button to immediately transfer $100 from savings to my transaction account, for when my card is declined at the register.
The survey's top three matched what the analytics said customers left to do.
Ranked in respondents' top three
I reviewed ten dashboards, and the big four Australian banks in depth.
| CommBank | NAB | ANZ | Westpac | |
|---|---|---|---|---|
| Accounts first | Yes | Yes | Yes | Yes |
| Primary and sidebar columns | Yes | Yes | No | No |
| Transfer or pay from the dashboard | Sidebar | No | No | Yes |
| Last signed in | No | No | No | Yes |
| Net position | Yes | Yes | No | No |
Take
Accounts first, with both balances. Transfer and pay without leaving. The last sign-in and upcoming bills.
Leave
Accounts as cards, repeating the Accounts page, proof of balance, and filling space because it is there.
CommBank: a two-column layout with the transaction account first, and upcoming bills and a transfer form in the sidebar. The red line traces the reading order
Design
In rows, every tile matched the tallest beside it, so the most important one fought its neighbours for attention. Nielsen Norman Group found 74% of viewing time falls within the first two screenfuls, so I moved to two columns on OneUI's 12-column grid: a wide one for the jobs customers came to do and a narrow one for context, with accounts top-left, where people look first.
Before
Fixed heights and nine tiles of equal weight.
After
Variable heights, a clear first read, and collapsible secondary widgets.
A card sort with three participants checked the groupings, and the survey and analytics set the order.
| Widget | Customers ranked it | Before, on the page | After, everyday banking |
|---|---|---|---|
| Account summary | 1st | 3rd | Primary, top |
| Transfer | 2nd | Not on the dashboard | Secondary, top |
| Recent transactions | 3rd | 6th | Primary, second |
| Bills and upcoming payments | 4th | 7th | Primary, fourth |
| Budgets | 5th | 5th | Primary, third |
| Net wealth | 6th | 1st | Secondary, collapsible |
| Spending categories | 7th | 4th | Secondary, collapsed |
| Cash flow | 8th | 9th | Secondary, collapsed |
| Rewards | 9th | 2nd | Moved to the Rewards page |
That ranking served Peter, but Jane signs in for her net wealth. Rather than average the two, each gets its own default built from the same widgets.
Default dashboards
Same widgets, a different order
Everyday banking
Wealth
Customers choose once, with a live preview, and can change it later.
Personalise your banking
Blue-sky ideas like savings goals mostly didn't survive the budget or timeline, but they gave the team a direction and showed what the first release should skip.
Testing
Each round tested one question, so a failure pointed at one decision.
Usability testing, rounds 1 to 9
98%task success on the final dashboard, with 87 participants
93% liked their card beside the account details.40 of 43, round 9, account summary.
95% called a quick dashboard transfer essential.38 of 40, round 5.
88% preferred a list of upcoming bills to a calendar.35 of 40, round 2.
87% found "Yesterday" and "Sunday" clearer than a date.33 of 38, round 5.
100% found "Last log in" helpful for spotting misuse.40 of 40, round 1.
Account summary, before and after: a plain list of accounts becomes rows with a card image, BSB and account number, current and available balances, and a menu of quick actions
QIn their words
I know Netflix is every month. I don't need to know how it fits in my calendar. All I want to know is how many days until I need to pay it.
I love being able to go in, do a quick payment to my mate, and leave.
I would not know when the 2nd of September was, but if you said yesterday I would.
What did not work, and what changed
Net wealth as the headline.A big home loan showed as a big negative at every sign-in, so it now leads only for wealth customers.
Cash flow periods behind a settings icon.Almost nobody found them, so they became Week, Fortnight, Month and Year tabs.
A feedback widget as a score.People would only rate a bad experience, so we kept it as a signal, not a measure.
Delivery
The tech lead, product owner, OneUI and accessibility signed off each widget. I then sized them with the squad and sequenced them by impact against effort.
| Widget | Impact | Effort | Release |
|---|---|---|---|
| New layout and Last log in | High | Low | 1 |
| Account summary | High | Low | 1 |
| Recent transactions | High | Medium | 1 |
| Transfer | High | High | 2 |
| Budgets | Medium | Medium | 2 |
| Bills and upcoming payments | Medium | Medium | 2 |
| Net wealth | Medium | Low | 2 |
| Cash flow and spending | Low | Medium | 3 |
We set targets before the build, so the A/B test had something to be judged against.
Targets, set before the build
Up 20%, from 6.3 to 7.5 out of 10.
Targets, set before the build
Up 5%, from 44.0 to 46.2.
Targets, set before the build
Down 15%.
Targets, set before the build
Up from 85%.
On mobile the order carried over, with the secondary column stacked below and collapsed.
Light
Dark
Accessibility
The dashboard was built to WCAG 2.1 AA. I annotated the designs, reviewed the code for semantics and focus order, and tested with screen readers, keyboard and zoom on desktop, iOS and Android before each release.
I was also the principal accessibility consultant across Macquarie's business units, including vendor products: Macquarie.com and Macquarie.com.au, Personal Banking, Advisory, Business Bank, Property Management and the Broker Portal.
The system underneath
When I joined, two sprint teams had already built plenty of duplicate UI, and an audit made the case for a design system. I then built components ahead of the sprints that needed them, so engineering never waited on design.
Products built on OneUI
Of UI design output covered by the system
The dashboard gave OneUI some of its most reused patterns: the account row, the collapsible card, the day-grouped transaction list and the inline transfer form.
Then, search
Next came search: every section had its own, and none could find a payment by name. We bet that one search for everything would cut frustration and calls to CX&O.
Redesigning global search
We tested real tasks, like paying a friend $10 for dinner. Results are grouped by type, with filters, without leaving the dashboard.
Global search
The impact
In a Google Optimize A/B test with 1,831 customers, the new dashboard beat every target.
| Measure | Target | Result |
|---|---|---|
| Ease of use | 6.3 → 7.5 | 6.3 → 8.9, up 41% |
| eBanking NPS | 44.0 → 46.2 | 44.0 → 68.2 |
| Time to make a payment | Down 15% | Down 23% |
| Task success | Above 85% | 98% |
Ease of use, from 6.3 to 8.9 out of 10
Time to make a payment
Designers on the team by the end
It became the reference point for later Personal Banking design.
Your amazing positive energy that you bring to every meeting and every interaction. It's a delight to work with you. There was nothing that I have personally asked you to do and you haven't been able to do it. You always go above and beyond.
Reflection
I would say plainly which evidence backs which decision. The ranking stood on 55 responses and 13.9 million events; the card sort on three people.
I would also ship the strongest default first, since we never measured how many customers changed theirs, and add a choice only where data showed a group poorly served.
