Categorisation and Enrichment Engine
Turn raw, ambiguous banking data into clear, actionable insights.
98%
accuracy in categorisation
Trained on a broad range of client data collected over 15 years, consistently maintained and retuned to ensure continued accuracy.
4
levels of granularity
Multi-level taxonomy provides up to three levels of detail for all transactions, with a fourth level specifically for categorising loan types.
99%
accuracy in merchant ID
Identify when new merchants enter the market, ensuring accuracy for both major household names and small, independent merchants.
extra
layers of enrichment
Transactions are enriched with merchant data like logos, geolocation and telephone numbers, alongside industry standards: amount, category, date and name.
Use Cases
How it works: Categorisation and Enrichment Engine
The engine receives raw banking transaction data in real-time, either via an API or a direct link to your Kafka broker. It can accept data from Moneyhub’s Open Banking connectivity or from a client’s own proprietary data.
The engine uses proprietary AI to balance and process four key sources of data: Moneyhub’s exclusive labelled dataset, your data, Open Banking data, and open data. Model weights are assigned to each data type, ensuring that input values are accurately mapped to the correct output classes.
The engine detects spending analysis, regular transactions, and merchant names from the data. This information is then fed into a machine learning model that categorises transactions into a four-level taxonomy, from a broad category group (Level 1) down to a specific loan type (Level 4).
Once categorised, the engine enriches transactions with contextual data. This includes merchant metadata like logos, websites, and telephone numbers. Optional geolocation data can also be added for card-present transactions.
The final output is clean, enriched data that generates customer and financial insights. This empowers you to build a fuller financial picture, create nudges, and develop targeted cross-selling offers.
Secure
cloud infrastructure
multiple availability zones with automated failover capabilities.
1 billion
transactions per day
at an average of 12,000 transactions per second (TPS).
48,000
transactions per second
scalability, beyond Amazon’s average of 29,000 product searches per second.
GDPR
compliant
all data is encrypted in-transit and at rest for security and regulation.
Core
Multi-level customer centric taxonomy
12 x Level 1 Category Groups and 65 x Level 2 Categories, providing granular categorisation.
Core
Merchant detection
Thousands of merchants and associated metadata, such as logos, telephone numbers and websites.
Optional
Geolocation
Provides highly accurate location data for where the transaction took place such as latitude, longitude and postcode.
Optional
Retrieving static map
Integration of dynamic map displays. This service connects seamlessly to retrieve map images, offering users a visually engaging way to view transaction locations.
Optional
Enhanced transaction information
Integration of enhanced merchant data to retrieve their information, including; phone number, website, and opening hours.
Optional
Regular transaction detection
Identifies a regular series of transactions from all transaction types (both income and expenditure), beyond simply Direct Debit.
“Partnering with Moneyhub will allow us to rapidly deliver far richer and more valuable insights for our customers.”
Ranil Boteju Group Chief Data and Analytics Officer at Lloyds Banking Group