Why is it so difficult to extract insights from transaction data?
Despite being extremely valuable in terms of consumer insights, raw transaction data is notoriously difficult to make use of.
The insights hidden in transaction data offer enormous opportunities. By analysing how customers spend, financial institutions can understand their needs both now and in the future, and build personal relationships with individual customers at scale through efficient digital services.
But why is the data itself so challenging to work with?
There are several factors contributing to the problem:
- Each transaction description is made up of information from multiple sources
Each piece of raw transaction data is made up of a complex web of information from multiple entities, including the merchant’s point-of-sale server, the payment scheme provider, some middleware and security software, the issuing bank, the acquiring bank and several other devices.
As a result, the data often ends up as a string of unintelligible letters, numbers and symbols that makes it difficult to decipher, both for banks and for the consumers themselves.
- Merchants are often registered as their holding company names
Merchants often don’t have a choice about how their name appears on statements. When they first apply for a POS terminal, it’s the processing centre that decides how to register the merchant. More often than not, the name of the legal holding company is used (e.g. “The Dairy Farm Company Limited”) instead of the trading name (e.g. “Wellcome”).
This makes it very difficult for financial institutions to analyse exactly where, and in what categories, consumers are spending. It also prevents banks from extracting the kind of granular insights that can be used to generate targeted offers, as the distinction between “Mannings” (health and beauty) and “Mannings Baby” (maternity) for example, is lost.
- Information is rarely updated when merchants move or change their name
This happens all the time. What was once a printers shop is now a cafe. A restaurant owner may change the name and concept of the restaurant as many times as needed. Businesses open, close, and change constantly. And most of the time, it’s cheaper just to take their existing payment terminals with them rather than registering new ones.
- A lack of industry standardisation
Visa, Mastercard, Amex and other payment scheme providers all have their own way of recording transactions, which means there is no consistency in how a transaction appears on a statement. As a result, there are countless ways one card-accepting merchant can appear in a transaction description.
- The transaction description has a limited number of characters
The transaction description itself is limited to around 23 characters or so. Therefore, the message sent from the merchant’s acquiring bank to the consumer’s issuing bank is often incomplete. When a transaction is queried, the bank often has no more information than the customer does.
The problem of unintelligible transaction data is costing the finance industry millions every year in customer support and chargeback and fraud investigations.
It’s also preventing banks from harnessing the power of big data analytics to deepen relationships and attract new customers, which could boost profits by 20% to 40%, according to a report by McKinsey & Company.
gini uses machine learning to transform unintelligible raw transaction data into clean datasets rich in consumer intelligence. Instead of a string of numbers and letters, banks get an accurate merchant trading name, category and location, along with descriptive tags. This allows for much more efficient analytics and much richer consumer insights.
To find out more about how we help financial institutions harness the full potential of their transaction data, click the link below.