If you had to name the companies that were founded in the 1950s and are creating cutting edge software today, I bet you could count them on one hand. And I bet FICO would be one that you would miss.
FICO is a company with a household name that dates back to its founding in 1956. In the U.S., consumers are familiar with FICO scores which are used to determine the credit risk associated with individuals. The brand is so ubiquitous in the financial industry that FICO even got a mention in the movie The Big Short, one of this year’s Oscar favourites.
Today, FICO is about a lot more than just credit scores. While the financial industry is still a key vertical for this vendor, it has diversified its software offerings to marketing and more recently, cybersecurity functions. I had the chance to chat with Scott Zoldi, the chief analytics officer of FICO, about how the company has evolved and how its algorithms are being used by its clients today. Watch the whole conversation above, and I’ll provide a brief summary below.
FICO’s evolution to marketing solutions
FICO evolved from working on credit scores for consumers to other algorithms that can predict customer behaviour. From there, they started using them as tools in their decision management solutions.
In marketing, FICO specializes in collecting a consumer’s transaction history and draws on variables that help segment that person. Factors such as items purchased and the periods between purchases are just a couple of examples of ways FICO can predict a customer’s future actions.
These models can be used by marketers to more effectively target customers based on offers they are more likely to want, and avoid flooding them with unwanted offers. One of FICO’s Canadian customers is Loblaw Companies Ltd., which has developed a loyalty program and a mobile app that pushes offers based on FICO’s algorithms.
FICO is able to look at the millions of shoppers visiting Loblaw’s grocery stores and determine patterns around how often an individual is buying milk, for example. A prediction is made about when you’re likely to be interested in buying milk again and an offer is pushed to your app to motivate that purchase.
While the algorithms offer very specific personalization, FICO keeps privacy in mind while designing its solutions. There’s no personally identifying data to match you as a person to your transaction history – because it’s not necessary to solve the problem of pushing you offers.
FICO’s different marketing solutions are tailored to different parts of the customer journey, so marketers can use the tool to help with the part of the process that they are most interested in. Some marketers that use FICO solutions are looking to develop customer archetypes to better understand how to cater to their audience. This is one of FICO’s main differentiator when it comes to its approach to marketing software. The company has many patented algorithms around analysis and the core offering it always comes back to is deep analytics.
New ventures into cybersecurity
For the past three years, FICO has been focusing on predicting how computers should be behaving in addition to people. It thinks that the current approaches to cybersecurity aren’t working – signature-based analytics are falling short, and heuristic-based approaches have also failed to prevent zero-day attacks. For FICO, it again takes an analytics approach and considers the typical behaviour of computer systems to detect potential threats.
FICO develops profiles around machines much in the same way it does around the customers. When intrusions take place, it can stick out like a sore thumb when you’re looking at the machine in this way, because it’s not looking for a specific signature, but rather comparing it against the typical usage of the machine.
Adaptive analytics – the next chapter
FICO just recently was awarded patents around adaptive analytics, which have the capability to be self-learning. This is the new chapter of fraud prevention for the company that has helped prevent credit and loan fraud for decades. As crooks learn to change their techniques to fool established systems, self-learning models are required to consider data in real-time. It also needs to be able to apply solutions and detect new risks with little feedback from its users.
The idea’s not to replace a human analyst that would make decisions about fraud prevention, but to try and assist them in doing so even more quickly. It’s the next progression beyond the rules-based fraud prevention of the 1990s, and is needed to reinforce the pattern-based models used in fraud prevention today. All models erode over time as knowledge of how to break them is developed, but a self-learning model could fight back.
Analytics in the cloud
Much like other software vendors, FICO also offers a software as a service option for its solutions. For customers that are interested in perhaps experimenting with FICO’s tools and algorithms, having the option to turn on a cloud service for a specific function is easier than having to organize an internal roll-out in some situations. FICO’s cloud provides it with a broader potential install base as a result.
