Payment Enrichment Services

We have identified three main weaknesses in Merchant Category Codes (MCCs):

• Incorrect: Wrong categorization
• Inaccurate: No categorization at all or too broad definitions of categories
• Inconsistent: Same merchant names are assigned to several different categories at the same time

In our model, we have overcome this error-prone categorization (to a high degree).

The standard category model is based on common international examples and distinguishes three main categories with varying granularity and up to 75 subcategories. In particular, our model includes a comprehensive range of Switzerland-specific categories (e.g. pension 3a or taxes). If required, the category model can be extended individually and for specific use cases. We are happy to show you what is possible.

SIX has developed a generic model that can process data from all known payment types. In addition to classic incoming and outgoing payments in e-banking or mobile banking, this also includes debit or credit card payments, TWINT payments, eBill and QR invoices.

Transactions that contain no or insufficient information cannot be categorized. Examples of this are missing intended uses, internal reference numbers without reference to a company name or similar cases.

Consumers can immediately identify transactions they have performed. This drastically reduces the chance of misinterpretation leading to erroneous fraud or chargeback requests. Also, the metadata enables the customer to provide their users with personal financial management experience in the software through analytics and filtering functions. This improves user satisfaction and increases retention.

The category module primarily makes use of our algorithm that scans description text fields for patterns to match against our inventory of over 100 high detail categories. The module can also make use of alternative data such as merchant category code data or other data that allows us to limit the scope of patterns to match. The results of the category module are categorizations with varying degree of granularity levels as well as a category logo to display. The varying category levels are customizable and can be applied directly to different use cases, such as displaying additional data on the merchant in e- and mobile banking, or performing analytics on various levels. 
For example, a transaction with description “Volt 3a - Dynamisch Risiko hoch,DA00100123456789012,XXX AAA,XXX, 8008 Zürich” will receive the categorization labels with increasing levels of granularity “Expense”, “Savings & Investment”, “Third Pillar”. 

The pretty name module extracts merchant strings from the description fields provided and matches those against our curated database of merchants. If a merchant string cannot be tied to an entry in the database, we apply a fallback pattern matching algorithm that is able to identify most Swiss merchants. Whenever merchants are identifiable the pretty name module returns a pretty name which reflects the doing business as name.

The logo module utilizes the same mechanic as the pretty name module to tie the merchant strings from the description fields to our curated database of merchant logos. If a merchant does not have a public facing logo, or the merchant counterparty is a (private) merchant on sites such as Ricardo, the logo module returns a category logo or platform logo, respectively, as replacement.

The merchant details module provides additional data regarding the merchant location where the transaction took place, such as geolocation, address, website, and phone number. This necessitates that additional geographically delimiting data fields, for instance zip codes, are provided and merchant branches/stores can be uniquely derived therefrom.