How to extract value from transactional data. Pt. 1

They share firsthand how to design a successful banking product and the journey a development takes from the Proof of Concept to its implementation in banking apps, where they are used by millions of people.

The Product and IT departments (led respectively by JC and Óscar) face numerous challenges daily that help us keep improving and enhancing the security of our modules.

Innovation never rests, and that is why we wanted to share with you the secrets of categorising banking data, enriching financial information and how to extract value from insights that help people improve their economic situation.

Let’s get started!

The birth of Coinscrap and its drive to improve financial health

To kick off the talk, there’s nothing better than recalling the beginnings of our company back in 2016. Juanjo handed over to Óscar, who explained how this Galician startup began with a B2C app focused on helping people manage their finances and savings.

The following year, we had already caught the attention of several national entities, with Caser being the first company we signed an agreement with. We realized there was a market to pivot to B2B, and since then, we have continued collaborating as a tech partner with the world’s leading banks.

Coinscrap Finance arrives: A strategic ally bringing innovation to the financial sector

The extensive knowledge gained from transactional data kept surprising us in the company’s early stages. The introduction of the PSD2 regulation led to a significant increase in digital payments and allowed the entities we work with to accumulate valuable information for their business strategies.

The case of EVO Banco: An intelligent piggy bank for automatic savings

Our CTO explained how the automated rules used by the Coinscrap app caught the interest of banks like EVO.

Tools that allow saving a percentage of the salary, allocating a fixed amount each month to the piggy bank, or setting aside an euro every time our team scores a goal, increase the entity’s NPS and boost user retention.

Micro-savings Santander: Personalizing savings goals encourages achievements

In 2019, one of the largest global banks knocked on our door. Coinscrap Finance’s innovative digital solutions are consumed through a white-label approach in their online banking.

Thanks to this, the entity provides its clients with a tool that allows them to control spending and reach savings goals in a fun and easy way. Gamification plays a crucial role in this module, leading to savings of 161 million euros for the 80,000 clients adhering to the program.

💡 Learn more about…

How personalising savings targets encourages achievement with the Santander Piggy Bank

Don’t miss the podcast with all the details!

Fintech innovation through banking transaction categorization

Juanjo then handed over to Juan Carlos to discuss the raw material that makes the magic happen: Transactional data. “They are fundamental for understanding users’ finances and spending habits,” said our CPO.

“The categorization tool classifies banking transactions into income and expenses, currently handling 17 categories and over 114 subcategories,” he explained. Additionally, it provides an enrichment layer that generates valuable insights. “Categorization helps entities offer hyper-personalized financial products by taking users’ consumption habits into account.”

Artificial intelligence and machine learning to process and extract value from banking data

On a technological level, having an AI engine with near 100% efficiency in interpreting banking transactions has been crucial for explaining the success of our tools.

COCO, the name of this engine, was named in honor of Óscar, as he explains in this post. Its operation is based on open banking technology, enabling us to decipher banking movements to show users and entities where, how, when, and in what way a deposit or expense was made.

Additionally, we detect the carbon footprint of transactions thanks to our COCO {CO2} tool, which this year received TÜV Rheinland certification. We are one of the five companies in Europe to achieve this milestone, meaning compliance with ISO 14064 and the GHG Protocol for organizations.

Our engine uses machine learning algorithms to understand the long, incoherent lines of data resulting from a purchase or deposit, returning organized and curated information that is highly valuable to banks and their customers.

We can achieve homogeneity and establish a system that unifies the codes used internally by each entity. This represents a tremendous challenge, and today, we offer the best categorization and enrichment tool in Spanish on the market.

The importance of training the AI engine with high-quality data volumes

Having a vast amount of reliable and high-quality information, like transactional data, makes COCO‘s training optimal. The origin of the data is 100% traceable and comes from infallible sources, something implicit in the heavily regulated banking sector.

From there, the implementation of improvements comes from the human supervision of the algorithm, extracting movements randomly for checks and can extract value, conducting numerous Proofs of Concept to facilitate the training of each category and the balancing of data.

“Training a garnishments category is much more complex than training a salary category, for example,” Óscar pointed out.

How detailed transaction analysis allows COCO to offer financial recommendations

Juanjo took the opportunity to ask Juan Carlos what exactly is meant by banking data enrichment. “COCO not only categorizes transactional data but also enriches it to provide more detailed and useful analysis.” Instead of simply identifying an expense as “restaurant,” COCO goes further to specify if it was at an Italian restaurant, fast food, or a café.

This level of detail helps users better understand their spending patterns and make adjustments to their personal finances. The ability to enrich data with additional details like the type of establishment and its specialty also allows banks to offer financial planning tools to their client base, positioning themselves as a strategic ally.

Use cases of enrichment and financial insights

Avoiding chargebacks thanks to the extract value of banking data

Our AI engine can associate transactions with logos and commercial locations, display the establishment’s website, or show the carbon footprint generated by an expense, among many other things. You can extract the data value, which is particularly useful for clarifying confusing bank movements and reducing chargebacks.

Juanjo then made a point by mentioning data from a study by Mastercard, explaining that 25% of chargebacks could be avoided if users better understood their transactions. Adding logos, names, addresses, and more to banking movements helps users feel more secure about their spending.

Recommendations based on financial insights

From the entities’ perspective, enrichment allows a deep understanding of the end user, knowing their specific needs at their current life stage. For example, if someone hires a company to organize their wedding, they might be interested in using a micro-savings module for their honeymoon.

Similarly, using financial insights, if a bank detects nursery expenses, it can check if the user has a transaction categorized as “life insurance.” It would be the ideal time to offer one if they don’t have it or make an advantageous offer for them to switch companies. The up-selling and cross-selling opportunities are endless.

With the participation of our CTO, we pause this first part of our Meets. The second part of this summary will be released soon, where you will learn more about integrations, security, and other technical aspects of our developments.

As Óscar says: “Collaboration between the IT and Product departments is essential to achieve this level of excellence and work in the top tier of finance.” In this post, you can see in-depth the stages of the development process of our innovative tools. 

Thanks for reading!