Bank Feeds: from ANZ to Japan, one market at a time
Finance teams in Australia were spending tens of hours a month manually importing bank transactions. After rolling out ANZ Bank Feeds through Australia's Consumer Data Right framework, we used that foundation for the next step: bringing the same capability to Japan through a partnership with Moneytree.
At a glance
The problem was the same everywhere. The solution couldn't be.
Across the JAPAC region, finance teams using NetSuite had the same daily friction: download a bank transaction file, reformat it to match NetSuite's import template, and reconcile manually—for every bank account, every day. For customers with multiple accounts, this consumed 20 to 40 hours a month.
The data already existed. The question was how to get it into NetSuite reliably, without the customer doing the manual work, and in a way that complied with each market's regulatory environment. Australia and Japan both had strong business cases, but completely different constraints. That made each market its own design problem.
Australia had a framework. Our job was to make it work for customers.
Australia’s Consumer Data Right (CDR) lets accredited data recipients access banking data on a customer’s behalf, with consent. Through SISS Data Services, which connects all four major Australian banks, NetSuite customers got automated daily transaction feeds with native bank reconciliation.
But on-site research with five Australian customers revealed the core design problem: connection failures were completely silent. A broken feed looked identical to a working one, so teams only found missing data during reconciliation, sometimes weeks later. One customer described the cost of a partial failure: “If some are failed, as we get a lot of account details, to reload and exclude the ones reconciled it’ll take them a day. Huge amount of rework.”
Two roles, one dependable bank-feed experience
The Finance Administrator
ANZ // SME finance team
The NetSuite Administrator
ANZ // IT or finance lead
From daily manual imports to visible, automatic feeds
Log in to each bank portal separately to download transaction CSV files daily.
A single SuiteApp installation connects all bank accounts through SISS, covering all four major Australian banks.
Reformat files to match NetSuite's import template. Each bank uses a different format.
Transactions auto-imported daily, already mapped to the correct GL accounts.
Existing connector fails silently. Missing data is only discovered during reconciliation, with no way to trigger a retry or know which accounts failed.
Connection status remains visible, with active, error, and pending states clearly differentiated and actionable.
No way to see whether bank connections are healthy without running a reconciliation and checking for gaps.
Design target: address 20 to 40 hours of monthly manual work for a typical SME with multiple accounts.
A connection you can see, from setup to daily import
Setup follows the SISS consent and bank approval steps in the order customers actually complete them. Once connected, every account shows its status (active, error, or pending), so a broken feed surfaces before reconciliation, not during it.
Australia Bank Feeds: setup, approval, and daily transaction import
Australia Bank Feeds: connection status, activation controls, and imported accounts in NetSuite
ANZ proved the model. Japan had no open-banking framework to build on.
With ANZ Bank Feeds live, the JAPAC team turned to Japan, a market with 313 prospect customers ready at launch and an acute manual data problem. In Japan, banks provide transaction data in Zengin fixed-length files. Finance teams download these manually, convert them to CSV in Excel, then import them into a temporary table before any matching can begin. Customer research in Tokyo confirmed the demand: "Manual downloading and importing are very slow. There is strong demand for automated bank feeds."
Japan has no Consumer Data Right equivalent and no single regulated data provider. What it did have was Moneytree, a well-established Japanese fintech with API connections to over 600 financial institutions. Oracle entered into a formal Technical Licence Agreement with Moneytree after a structured evaluation involving product, legal, privacy, security, and engineering. The design challenge was translating that partnership into an experience Japanese finance professionals would trust.
Japan research: the current manual bank reconciliation process
Same feature. Completely different trust and regulatory context.
Face-to-face, bilingual, in Tokyo
The research involved four in-depth interviews (three in person and one online follow-up) conducted bilingually in Japanese and English with finance and accounting professionals at Japanese businesses. Sessions covered current bank transaction workflows, reconciliation pain points, promissory note processes, and expectations for automation.
Bank reconciliation in Japan has challenges that don't arise in Australia: bank charges cause amount differences that require automatic splitting, consolidated remittances mean one deposit covers multiple invoices, and Japanese bank data often lacks reliable identifiers for matching. The research shaped both the bank feeds setup design and the longer-term reconciliation roadmap. Privacy expectations were also clear, and they shaped a key design decision in the interface.
The Japan Bank Feeds UI, built from customer research
The interface groups bank connections by provider, shows account-level import status, and supports mapping bank accounts to NetSuite GL accounts. Its setup and account language reflects how Japanese finance teams understand their bank account structure.
Japan Bank Feeds: permission-led setup and daily Moneytree Link data transfer
Japan Bank Feeds: Moneytree ID connection and activation prompt
Japan Bank Feeds: connected Moneytree hierarchy, bank account status, and GL account mapping
Future opportunities, informed by the research
The delivered design is shown above. The concepts below are future opportunities informed by the research, not shipped functionality.
Once imports are automatic, the next friction point is matching each transaction to the right GL account or open record, and AI could suggest the most likely match. Research set the principle: suggestion, not automation. Users accept, change, or dismiss each suggestion, and nothing posts without confirmation.
Silent connection failures were the core trust problem in ANZ. AI monitoring transaction volume and timing could spot anomalies, such as a bank that normally sends 40 transactions overnight sending none, and alert users before reconciliation.
Clean daily transaction data gives AI enough signal to spot recurring patterns such as payroll, rent, and supplier payments. That could turn bank feeds into a cash-planning tool for small businesses, without overwhelming users who only came to reconcile.