Loading...

We will contact you

GET STARTED

Fill in the form and we will contact you shortly.

How AI is transforming payments

Hero Fcg Bambora 1 Hero Fcg Bambora 1 Placehoder

In just a few years, artificial intelligence has gone from a promising technology that excelled in board games to creating real improvements in everyday life. During the fall and winter of 2018 Bambora and FCG launched a pilot to explore the possibilities of this technology in the world of payments. The results showed both a great potential for improvements in the short term and endless possibilities for the future.

Payments are in many ways the perfect application for AI. Every day, payment companies handle enormous amounts of structured data, surrounded by a strict regulatory system, just the type of data that AI is already great at analyzing and finding patterns in. Patterns that can then be used to reduce costs and prevent fraud.

At Bambora, this analytical approach isn’t new. Customer’s data and transaction patterns are already being analyzed to find new ways to optimize both transaction flows and costs through what is called Payment Performance. At the same time, AI is also seen as the natural next step when developing and streamlining the Payment Performance approach.

Pilot project put AI to the test

In the fall of 2018, Bambora launched a first pilot project into AI together with FCG, a consultancy and technology firm specialized in services and solutions for companies in the financial sector. The thought behind the project was to combine Bambora’s data and payments expertise with FCG’s unique experience from working with AI in the finance industry. The basis for the project was data from a half a billion transactions that was then analyzed using FCG’s own AI platform, DeepSea.

”The essence of DeepSea is that it combines data scientists and data engineers with a technical platform. The people have in-depth knowledge and skills in data science, data engineering, software development and support. The platform makes use of Amazon’s AWS infrastructure which allows for agile implementations, timely deliverables, scalability, quality and resilience. All that translates into a complete offering, from data capture to fully data-driven automation and analytics-as-a-service”, says Joel Nisses, who headed up the project from FCG.

"We found a new variable that we call ‘unclean transaction’ and that helps explain the cost of a transaction"

Joel Nisses, FCG.

Joel 5 Joel 5 Placehoder

A key to lowered transaction fees

DeepSea’s mission was clear: by analyzing the transaction data it was to identify areas that could be optimized to benefit both Bambora and our merchants. One such area was the fees that card schemes and card issuing banks charge for transactions. These fees can vary significantly between different types of transactions and the prize for any given transaction depends on a great number of variables and how the acquirer chooses to handle the transaction.

Bambora already uses an advanced model to calculate transaction fees and to analyze how transactions should be handled to minimize fees. But with the help of DeepSea, FCG built machine learning algorithms that gave an even clearer view of which factors actually drive card fees, as well as the different cost structures of the various card schemes.

”Among other things, we found a new variable that we call ‘unclean transaction’ and that helps explain the cost of a transaction. This is just one of many variables, but it has a rather large degree of explanation. The concept ‘unclean transaction’ doesn’t necessarily mean there is something wrong with the transaction, but it can give the acquirer a clue about how to minimize transaction costs and consequently lower the price for the merchants”, says Joel Nisses.

More effective risk assessment

Another area where Bambora and FCG identified a large potential for AI was risk assessment. Risk assessments are done when an acquirer is boarding new merchants and continuously throughout the customer lifecycle. The assessment is based on a long line of factors, from fraud risk to credit rating and compliance.

With the help of DeepSeas algorithms, this assessment could be done in a much more advanced way, that could find new factors to improve risk prediction and prevention. These insights could also be used to improve the risk based pricing, based on each companies individual risk level.

“We also see that this technology could play a really important role when it comes to increasing the effectiveness in predicting, preventing and identifying fraud. It’s clear that AI could be used to find new patterns in our transaction data that would make us even more accurate when identifying suspected fraud and making risk assessment. That means we can focus our resources more on what’s most important and less on irrelevant alerts”, says Stina Granberg, Head of Operations for Global Acquiring at Bambora.


"This means we can focus our resources more on what’s most important and less on irrelevant alerts"

Stina Granberg, Bambora.

Text Image Stina Text Image Stina Placehoder

Potential for new services

According to Joel Nisses, the pilot project with Bambora provided clear evidence of the amazing potential that AI has in the world of payments. Beyond lower fees and better risk assessments, he thinks that AI will be instrumental in providing the industry with better forecasting tools that can handle transaction data in real-time to predict the future. One of the benefits of this would be that Bambora could offer its merchants more cost effective transactions.

In addition, Joel Nisses also thinks that AI will allow payment companies to develop completely new types of services. One example is the possibility for merchants to see their own improvement potential by making real-time comparisons with a benchmark of similar businesses. This type of service is something that Joel Nisses believes will be an important advantage for payment companies on an increasingly competitive market. But to get there, he thinks it’s important for payment companies to start testing the technology. No one becomes an AI expert over-night and therefore you have to start small. That sentiment is also shared by Stina Granberg at Bambora.

“This pilot project was a great opportunity for us to explore how this technology can be used and what challenges we have to overcome when we start using it on a daily basis. It also showed us that our current way of working is correct, now we just have to look at how we can use AI to take it to the next level”.