Tier-1 Bank
Evaluating Vendor Solutions
In the risky world of finance, tier-1 banks are continuously exploring advanced technologies to enhance their security and compliance measures.
One such institution sought to evaluate a leading homomorphic encryption solution to bolster its financial crime detection capabilities. Leveraging Syntheticr Institution and Syntheticr Custom, they tested the solution and ultimately saved millions of dollars by avoiding a costly implementation that would have underperformed in real-world conditions.
The tier-1 bank recognized the potential of homomorphic encryption to allow secure data processing and peer risk exposure verification, without exposing sensitive information. However, they faced three critical challenges:
How to accurately assess the solution’s effectiveness given real-world data was not viable for this phase of the engagement.
How to validate the behaviour of the when applied across multiple sources of data.
How to replicate persistent data quality issues they experienced with their own data to create a real-world testing scenario.
Using Syntheticr 1.0 - a comprehensive synthetic financial transaction dataset with embedded financial crime scenarios - they conducted performance testing of a homomorphic encryption solution. The initial results were promising, as various types of financial crime activity were accurately detected and positive feedback was provided from peer institutions.
However, to ensure a thorough evaluation, the bank used Syntheticr Custom to replicate their persistent data quality issues. Using this customised dataset, they were able to perform a more realistic test of the solution. During this second test the performance of the solution deteriorated significantly for both analytics and anonymised peer lookups.
The Testing Process:
Initial Testing (Syntheticr Institution)
Using this robust synthetic dataset, the institution initially tested the homomorphic encryption solution by adding the synthetic dataset and comparing the results to the labelled dataset. This yielded positive results, accurately notifying known instances of anomalies and suspicious transactions.
Adapting Data Quality (Syntheticr Custom)
To better simulate their own data, the bank used Syntheticr Custom to incorporate the persistent data quality issues they experience into Syntheticr 1.0. This new dataset included characteristics like records, inaccurate transaction timestamps, and inconsistent data formats.
Refined Testing (Syntheticr Custom)
With the custom dataset in place, the financial institution conducted a second round of testing on the homomorphic encryption solution. This time, the results were concerning; the solution's performance in identifying other known instances of financial crime activity diminished significantly, revealing that it struggled under conditions that closely resembled the institution's operational realities.
Based on the findings from the second round of testing, the tier-1 financial institution decided not to purchase the homomorphic encryption solution. By rapidly proving the technology would not perform as anticipated, they avoided a multi-million dollar investment in purchasing and deploying a solution that would have failed to meet their requirements.
Syntheticr Ecosystem and Syntheticr Custom enabled the timely and cost-effective testing of a cutting-edge financial crime solution that would not have been possible using real-world data.