Growth design at one of
Canada's biggest banks.

Forms are where conversion goes to die.

The Auto Loan lead form was a high-stakes page with high abandonment. Customers who reached it were already interested — but the form was losing them before they could complete it. Drop-off data pointed to specific fields where users stalled or abandoned entirely.

The hypothesis: the form's cognitive load was too high. Too many open text fields, no visual hierarchy to guide users through the flow, and a layout that felt equally demanding on desktop and mobile.

"The fields with the highest drop-offs shared a pattern: they required freeform text input where structured input could have done the job better."

The design approach.

Rather than a full-form overhaul, the redesign targeted friction at its source — the fields with the steepest abandonment curves.

Two research streams. One aligned problem statement.

The redesign was grounded in both quantitative behavioural data and qualitative user research — run in parallel so each could validate the other before any design decisions were made.

Competitive analysis reviewed traditional financial institutions, fintechs, and indirect competitors against a structured grading system covering branding, innovation, navigation, technical quality, and accessibility. Designers, stakeholders, and agile team members were brought into workshops to score and discuss findings.

User research was conducted with BMO's dedicated research team and included synthesis of prior research, heuristic review, internal stakeholder interviews, and moderated usability testing — both on competitor sites and on bmo.com.

Two structural changes drove most of the improvement.

The new UI was visually simpler — better contrast, micro-interactions, animation, and white space. But the changes that mattered most were structural: a rebuilt IA and a new modular layout system.

Five teams. One consistent direction.

A homepage redesign at a major bank touches more teams than a typical product feature. Each brought constraints that shaped the final design — and aligning them without losing design integrity was part of the work.

Holding the design direction steady while incorporating feedback from five different stakeholder groups — and still shipping on time — is the kind of cross-functional coordination that doesn't show up in a prototype but is felt across every phase of delivery.

What I'd do differently.

The competitive analysis was thorough but primarily a benchmark exercise — evaluating what others had done rather than stress-testing our own assumptions. In retrospect, I'd push for lightweight generative research with BMO users earlier in the process, before the competitive audit, so we were grounding competitor observations in actual user behaviour rather than the other way around.

I'd also advocate for a longer A/B testing window on the module test. The results were significant, but a longer run would have given more confidence on lower-traffic segments — particularly mobile users in smaller markets.

What this reinforced.

Growth design teaches you to hold two things at once: conviction about the problem, and humility about the solution. You can be confident in your hypothesis and still get surprised by the data. The best test results weren't the ones that confirmed what we designed — they were the ones that told us something we didn't expect.

The position-matters finding from the module test is a good example. We didn't design to test that; we discovered it because we were reading the data carefully. That kind of accidental learning only happens when the team treats A/B tests as genuine experiments, not just validation exercises.