The Neobank Squeeze: Why Classic Banks Need to Learn to Move at Challenger Speed
The threat to Australia's classic banks has stopped being theoretical.
For years, the neobank threat in Australia had a comfortable shape. A handful of digital challengers with clever branding and low fees, popular with students and early adopters, but never quite big enough to worry institutions built on decades of trust and a national branch network.
That's no longer the case. Three separate developments in 2026 show the ground is shifting beneath classic banks.
In July, Revolut became the first global fintech to get a full, unrestricted banking license in Australia. The company has invested roughly 400 million dollars over five years to get there, and it's moving more than one million existing Australian customers straight into the newly licensed bank.
That same month, OFX agreed to a 247 million dollar all-cash acquisition by UK fintech Equals at a 100 percent premium. This is a reminder that this crop of challengers gets priced on earnings, like any other bank, not on a speculative growth story.
The third example: Up, the digital bank now owned by Bendigo and Adelaide Bank, reached one million customers, with 29 percent year-on-year growth. More than 85 percent of its customers are under 40, and most of that growth is from word of mouth. Bank of Queensland’s own answer has been to run ME Bank as a separate digital-first brand since 2021, a tacit admission that it couldn’t achieve that speed within its core systems.
What's actually changing
The pattern isn't that Australians are suddenly abandoning their banks. What's changed is narrower and more serious: an entire generation is forming its first financial habits somewhere else. The institutions built to capture that generation are now backed by global capital, full banking licenses, and takeover valuations that treat them as permanent fixtures rather than experiments.
Revolut is arriving with a tested product and a five-year reinvestment plan. OFX shows the sector now has enough capital to buy its way to scale. And Up’s growth curve shows the clearest risk to classic banks: the generation that banks need to fuel the next thirty years of deposits and mortgages is forming its expectations somewhere the classic banks don’t control
This rarely shows up as a dramatic deposit outflow. It shows up as a slow bleed in the highest-value segment. These customers are forming their first real financial relationship elsewhere, with no reason to switch to another bank.
Classic banks measure their position against other classic banks. NPS surveys. Market share tables, refreshed annually. These measures aren't built to catch a demographic bleed happening one customer at a time through word of mouth. These metrics can look stable while losing the relationship that matters most for the next three decades.
How classic banks actually compete at this pace, with control
A large bank's instinctive response is to:
- freeze, waiting for more data before investing in a segment a challenger is already winning, or
- compartmentalize, standing up a digital innovation lab that produces good prototypes but never touches the core product that millions of customers actually use.
Neither response works because the problem is structural. Digital-first banks don't run an annual roadmap. They run continuous experimentation, testing dozens of variants a year on onboarding and messaging at a fraction of the cost and time of a traditional release cycle. The earlier piece in this series described exactly this cadence: one digital bank running roughly thirty experiments on its onboarding funnel in a single year, doubling conversion at some steps along the way. A classic bank shipping a handful of major digital releases a year is not competing in the same contest, regardless of how good any single release is.
The way through isn’t to become a neobank overnight. Nor is it to keep the two worlds permanently separate. The answer lies in separating what must move at bank speed from what doesn’t.
Banks need to move carefully on core pricing, credit risk, and regulatory disclosure. They can move faster on rethinking onboarding flows, in-app messaging, offer sequencing, etc. The latter category is far larger than banks realize.
Once you make that separation, the job gets simple: watch how customers actually behave across your sites and apps in real time, then test and ship changes against that behavior all the time, not twice a year. This is where a combined analytics and experimentation layer earns its place as core infrastructure, not an add-on.
Amplitude provides that behavioral visibility, empowering financial institutions with the same view that a five-person growth team at a neobank builds its whole rhythm around, applied at the scale of a bank with millions of customers. Statsig, now part of Amplitude, provides the experimentation engine to test and ship changes against that behavior with statistical rigor, at a cadence that can match a challenger’s, but with the audit trail, guardrails, and stopping rules a listed bank’s risk committee actually requires before anything reaches a live customer.
Speed and control have rarely been built into the same stack. This is why so many transformation programs have delivered one at the expense of the other. A bank that can observe behavior in real time and test it against proper governance is not choosing between the two. It’s doing what challenger banks have done from day one, inside an operating model built for a regulated institution rather than a five-year-old startup.
Behavioral visibility and governed experimentation are the foundation, but the layer that compounds fastest on top of them is personalization, increasingly driven by AI rather than static rules. A neobank’s advantage isn't just that it tests more; it’s that what it learns from one customer's journey shapes the next customer's experience almost immediately. Classic banks generally have the data to do the same thing, but it’s rarely connected to a single, live decisioning layer. AI makes that connection tractable at scale, not by replacing the experimentation discipline described above, but by turning its output into a continuously adapting customer experience rather than a quarterly report a product team reads and acts on manually.
The practical starting point is to pick one high-traffic, high-friction journey and expand from there. Onboarding is an obvious candidate. Instrument it well, and you can prove a small team can spot a behavioral problem and ship a governed fix in weeks, not a full release cycle. Once that loop is proven, and risk and compliance are comfortable with how it runs, expanding to the next journey gets easier. So does layering on real personalization.
Start small and proving value is much easier than asking for enterprise-wide approval before anyone’s seen it work once.
What this means for the next decade
2026 will likely be remembered as the year the neobank threat in Australia became real. A fully licensed global competitor, a fintech sector mature enough to be bought and sold at real valuations, and a customer base in its twenties and thirties that has never had a branch conversation about which bank to choose. None of that is reversible, and none of it should be met with panic or denial.
The classic banks that win the next decade will not be the ones with the prettiest app or the biggest transformation budget. They'll be the ones that build the operating discipline neobanks have had from birth, but with the same seriousness already applied to credit risk: continuous behavioral intelligence, governed experimentation, and AI-driven personalization working as one system. Speed must become a capability the classic banks control, built into how they operate, rather than a threat they spend the next decade reacting to.

Mark Drasutis
Head of Value, Asia Pacific and Japan, Amplitude
Mark Drasutis is the Head of Value, APJ at the leading digital analytics platform Amplitude. With more than 25 years of experience leading digital products, transformation and driving innovation, Mark is passionate about turning complex challenges into opportunities for growth.
More from MarkRecommended Reading

Your agents are only as good as your data context
Sep 4, 2026
9 min read

The New Trust Economy in Financial Services
Sep 2, 2026
11 min read

How to Secure AI Agent Traces Without Losing the Signal
Aug 28, 2026
7 min read

Connecting agent performance to product outcomes
Aug 20, 2026
10 min read

