Seconds are passing us by at this very moment, and there's not much you can do in one second. Well, there is something. Let's say smile, take a deep breath, close your cellphone and say thank you. Within a second you can also make decisions: click the buy button in the online store and later accept payment on your online banking website. In this post, Saku Niemi and Antti Heino are taking you on a real-time decisioning tour.
Every second is important in business, especially in customer encounters. Real-time is a growing asset, as customers increasingly appreciate the ease and convenience of things happening immediately. Fewer and fewer of us even realize how much things happen in the background when, for example, we buy something from an online store or do business in online banking. Things just work.
Digitalization and artificial intelligence have enabled new types of business models and ways to serve the customer. They can better scale services to take advantage of every second. Especially during COVID-19, digital channels and online stores have become important tools for companies to grow or even survive.
Especially during COVID-19, digital channels and online stores have become important tools for companies to grow or even survive.
It is interesting to delve into what can be done behind a seemingly simple purchase transaction. In an online store, this includes two parts: the actual transaction fixing and the payment.
AI in purchase transactions
The lifeblood of a successful online store is to utilize the data generated from the purchase as quickly and efficiently as possible. Companies with the most advanced artificial intelligence and automated decision-making processes can use real-time data. For many, however, data processing takes days or, at worst, weeks. And this delay, of course, means lost sales.
For example, let's look at following trends. For us to detect them, we need to update our algorithm, which analyzes what products are purchased together. At the same time, you should also check the ads received by the customer about the company's products. There is nothing more pointless in marketing than advertising to a customer the same product he or she has just purchased. Yet this is still happening surprisingly often.
A purchase transaction can also trigger a redefinition of the customer segment, as well as churn risk. Based on all this information, with intelligent decisioning, the company can automatically determine what would be the best follow-up for the continuation of the customer relationship. This could be a thank you message, a discount campaign or a call from customer service.
Another example of what happens in a second is payment transaction verification. Banks have a great deal of responsibility in real-time payment transactions. Regardless of whether the customer pays with debit or credit card bank. According to the EU's Second Payment Services Directive (PSD2), online payments always require strong authentication. And the bank is liable to customers in the event of abuse.
Banks must also allow third parties access to the customer's payment account if consent has been given. If something goes wrong in identifying the customer and accepting the payment, the bank has a high financial and reputational risk. Fraudsters exploit the openings they find mercilessly, quickly and at scale. The tabloids also have a field day if such cases become public.
Fraud protection in the blink of an eye
A blink of an eye is about 300 milliseconds. Accepting an online payment happens even faster. The most advanced banks' fraud prevention and control systems analyze all the data related to a payment transaction and make an automatic decision in less than a hundred milliseconds. Sophisticated technology confirms that you are the one behind the transaction and decides to accept or block the sale.
In addition to hundreds of business rules, banks use advanced algorithms and machine learning models to make decisions. They compare transaction data against customer data, geolocation data and previous payment history, including social network analysis to identify fraud.
Several experts build these machine learning models with anonymized data from hundreds of thousands of previous similar types of transactions that banks collect in order to fight financial crime.
Real-time anti-fraud systems work 24/7/365 and handle up to dozens or hundreds of simultaneous transactions in less than a blink of an eye – depending on the size of the bank. The seriousness with which financial institutions consider the potential for fraud and invest in its real-time prevention and detection tells a more convincing story of digital capability than many of the strategy slides presented at investor days.