ABN AMRO Bank optimizes financial crime detection with Azure Data Explorer

ABN AMRO, a Netherlands-based financial institution, is committed to delivering personal banking services that meet the dynamic needs of its clients. "We take our role in society seriously, and work hard to build and sustain trust with our clients,” shares Jurriaan Amesz, Lead Product Owner at ABN AMRO.
To this end, ABN AMRO stays vigilant against financial crime. “We’ve established a group-wide due diligence environment driven by data and smart tooling,” explains Amesz. “We see our Detecting Financial Crime (DFC) department as the gatekeeper of the financial system.”

New analytics system

ABN AMRO previously conducted reviews through intelligent dashboards that required analysts to go through customer information in batches, weekly. Then more recently, the bank identified vulnerabilities that could potentially enable bad actors to exploit the system for money laundering and other illicit activities.

Initially, the bank's response was to onboard more analysts for due diligence and transaction monitoring. "We soon realized it wasn't sustainable. The case inflow was more than we could handle, and we weren't meeting our targets,” Amesz explains. The bank had to sift through billions of transactions from seven million customers, covering everything from customer, contract, and transaction information, to reference and external data.

On top of that, for money laundering cases, the bank needed to analyze networks of customers instead of individuals. All these considerations necessitated a new approach and system. “We needed to generate better insights and make the whole process more efficient with data science solutions, rules, and dashboards,” Amesz adds.

Due diligence evolution

To enhance real-time data handling and concurrent users, the bank introduced Microsoft Power BI and Azure Synapse Analytics. “This provided us with the performance for hundreds of concurrent users handling tens of billions of records. It’s enabled us to execute searches more quickly and more thoroughly,” Amesz proudly says.

Around the time ABN AMRO adopted Synapse Analytics, Azure Data Explorer (ADX) in Synapse became available. The bank ran a proof of concept with a few months' worth of transaction data, linking ADX to Power BI. “We even received training from Microsoft, it was a real partnership where they helped us think about the architecture, and it was tremendously useful,” shares Jorg Klein, Domain Architect a.i. for the DFC Department at ABN AMRO. “It was unbelievably fast. I have worked with data analytics products for 15 years and I’ve never seen anything process data this quickly.”

Knowing that they could push ADX even further, DFC worked with the Microsoft team to optimize Power BI queries and reports. “We deal with many alerts and triggers from our machine learning models and rule sets indicating potential financial crime activity. Given the volume of analysts and alerts, even minor performance enhancements to our Power BI reports, such as reducing load times from almost a minute to a few seconds, can save a substantial amount of time and money over a year,” Klein explains. “Power BI also provides excellent data visualization capabilities, which enable our analysts to better understand and manage alerts.”

The bank also created a bridge between data and workflows, speeding the process along further. “We have integrated our data & analytics platform with our operational applications, which is quite unique,” says Klein. “We source data from the operational applications, generate actionable insights in the platform, and go back to the business processes and end-users, where manual interference is by exception only.”

As part of the transition, some teams switched to Kusto Query Language (KQL), the primary language for ADX, from the traditional SQL they’ve been using. Learning the new language was straightforward for these ABN AMRO DFC teams. "Some of our team members used the Kusto Detective Agency game, which made learning the new language fun,” Klein shares. “If you understand SQL, learning Kusto is easy."

Optimized financial crime detection

Following the implementation of ADX, there was a clear surge in the bank's productivity and efficiency. Klein points out, "With large data sets, you usually need to worry about all sorts of things like partitioning, data optimization, and ordering. With Data Explorer, the process is quite streamlined—you simply load the data, and it functions efficiently, even while enforcing fine-grained access control policies. All columns are indexed and optimized right out of the box, which is a unique feature that enables the DFC department to handle tens of billions of transactions.”

The difference is tangible. Previously, the average wait time for some of the very heavy reports was about a minute. With ADX, it's reduced to 15 seconds. “With over 2,500 analysts, this saves us a lot of time overall,” Klein adds. “What’s more, we can now also investigate queries that were unthinkable before, like free text analysis or text mining scenarios. ADX is a hidden gem within Azure Data Analytics, and many individuals are unaware of the potential it holds.”