Experian CIO Barry Libenson explains role of data in credit monitoring

Experian Global CIO Barry Libenson sets a technology strategy that spreads across 39 countries, overseeing a 1,500 person software development team that builds products used by financial institutions around the world, helping them to understand the spending habits, attitudes and behaviours of customers.

Libenson joined the credit reporting agency in 2015 after three years as CIO of Safeway, becoming the first person in the company’s history with global responsibility for technology. In every year since his arrival, Experian has been named one of the “World’s Most Innovative Companies” by Forbes magazine, thanks in no small part to its use of data analytics.

“That’s one of the things that Experian does really well,” Libenson tells CIO UK in Experian’s central London offices. “We curate massive amounts of data on a regular basis and we have petabytes of information that we use in a decisioning process.”

He estimates that Experian offers around 800 different products, covering everything from fraud detection and credit-worthiness to targeting software. Data is the core component of the Experian product portfolio, which needs to keep pace with a rapidly evolving landscape.

On-demand analytics

IDC predicts that by 2020 the data the world creates and copies annually will reach 44 trillion gigabytes. This flood of information can be overwhelming. Experian estimates that just 30% of organisations use analytics to improve their data insights. The company’s research suggests that four in ten companies struggling to cope with the volume and complexity of data, while 71% said enhancing their analytics capability was a top priority for their business.

To help these organisations, Experian recently rolled out an analytics on-demand platform called Experian Ascend. The hybrid-cloud platform is designed to quickly produce reliable insights on changing consumer behaviours by providing access to a range of data sources in real-time via a set of self-service tools.

Customers can use Ascend to build their own predictive models and make decisions by applying machine learning and AI techniques to Experian’s hordes of credit, client and alternative consumer data on more than 220 million people. The content is mined in an analytical sandbox to unearth real-time insights into consumers and businesses.

Traditionally, the likes of Barclays Bank and Lloyds would approach Experian to understand the effect of changing the parameters that determine someone’s credit-worthiness, which they factor into their calculations on what interest rate to charge on a loan.

“The way that historically would happen is you would come to someone in Experian, a salesperson or somebody, and say we want you to run this campaign,” explains Libenson.

“They would then take that back to the mothership and we would run it for them, and it would take some period of time to get them the results. Then we would provide them with a report with those results.

“What the sandbox allows them to do is sit down in front of a terminal and use SAS or R or H20, or any other of the analytical languages that are out there and do it themselves in real-time. In the case of the US, it has the data on all 300 million individuals, which basically represents the entire credit-worthy population of the United States, because that’s the size of our data source. We’re rolling that out now, and the demand has been off the chart.”

Hybrid cloud benefits

Ascend uses a hybrid cloud approach to optimise capacity for seasonality, especially around the busy Christmas period. “We only need it for maybe a month or so and then the load drops back down to sort of a sustainable regular load that we anticipate,” says Libenson.

“By building the sandbox this way, if processor power becomes an issue, it’ll dynamically scale. We can set it to add cores to the environment, and we pay for those cores as long as we use them but then if the load drops that processing power dissipates or is removed.”

It also gives them further data storage flexibility. Experian uses Kubernetes to orchestrate the movement of containers between data centres and the cloud within minutes.

“The application will continue to run and it won’t miss a beat,” Libenson explains. “It just automatically adjusts its pointers through orchestration back to where it needs to basically point in order to get the data,” he explains.

Emerging technologies

The methods of analysis are rapidly changing. Experian is applying machine learning to understand how applications behave and to identify potential risks by using Dynatrace to monitor their performance and then funnelling the data into Splunk for analysis.

“We can find things that are outside of normal tolerance much more quickly using machine learning than we were able to do previously because human beings just can’t process the information that way and aren’t nearly as likely to say that something is out of the ordinary as a machine,” says Libenson.

“Machines have no emotions. They don’t care. If the machine sees something that it thinks is abnormal, it’s just going to say: ‘I can’t make sense of this, somebody should look at it’. Whereas a human being is much more likely to put a subjective inference into that. That’s why the machine technology’s much better at monitoring this stuff and watching for bad behaviour and potential areas of problems.”

Experian also recently introduced a free Child ID Scan to help protect children in the US from identity theft. The risk is growing as children are leaving online footprints at ever younger ages. This data is valuable as they have no existing record, so a thief can create a fake credit history based on their name and security number.

Javelin Strategy & Research estimates that in 2017 more than a million children were affected by identity fraud, resulting in losses totalling $2.6 billion and families paying over $540 million out of pocket.

The Child ID Scan checks if their social security number is associated with an Experian credit file. If it finds one, the company will help the family to resolve the issue.

“If they have no credit history at all, it’ll allow you to establish a legitimate credit profile for them, making it much more difficult for somebody to steal their data later down the road,” says Libenson.

The power of APIs

Libenson is a big believer in the benefits of application programming interfaces (APIs). Experian initially used these building blocks to help credit companies integrate data into other parts of the organisation.

“The idea was: let’s create an internal API Hub and instead of building point to point applications, we’ll build a hub and spoke architecture so that when anybody wants a data element they can call the API Hub, and the API Hub will go to the data source to retrieve the information,” he says.

“That way, if the backend data source ever changes, the actual application code doesn’t need to change, the API interface does. It’s much easier to change that API interface than it is all of the code. In theory, if you change the data source you only have to change the API code once, and all the applications that are calling for that data will now hook up to the new API entry points instead of the old ones.

“Essentially, you only have to rewrite a very small piece of code, whereas in the old way of doing things if you changed the data source and there were six things calling, all six had to be rewritten. Now we rewrite one.”

Experian then added APIs for financial institutions and customers who wanted real-time access to only a limited number of data elements in their credit reports.

Finally, the company rolled out a set of public APIs of redacted or starred data that anyone can access for free to design their own application. They can then enter into a commercial agreement with Experian to build them with the full dataset.

Experian built the API hub based on Google’s Apigee framework, after giving everyone in the team a chance to participate in the decision-making process.

“That’s one thing we always do,” says Libenson. “Even if we don’t always agree 100%, as long as people are given their opportunity to provide input and participate in the process, they generally will get behind the solution. We’ve had really tremendous success with Apigee and so we’re really pleased with that as the choice at this point.”

Building with microservices

Libensen has also embraced microservices to make software development more efficient.

“There’s been kind of this paradigm shift to the way applications are being built and it’s really around the use of microservices and building reusable components,” he says.

“What we’ve started to do is encourage the development teams to build their applications using microservices so that the services can be shared and reused by other parts of the organisation.”

This approach helped Experian extend the “pinning” technique the company uses to connect data to individuals.

“We don’t want multiple pinning algorithms, we want one, so we created a microservice that the data fabric team uses for doing all data ingestion and pinning,” says Libenson.

“If you’re somebody in the company building an app and you need to pin data, you go use that microservice and you don’t write anything. The code already exists. There’s a published specification on how you use it and you hook that into your app and use that microservice as you build your product.

Complex communication

Libenson manages a vast team spread across so many locations that he racked up more than a quarter of a million air miles last year.

To integrate the diverse needs of his teams and their customers, the CIO relies on a variety of communication methods, a comprehensive organisational structure, and a set of universal policies for developing products.

“We do a lot of face-to-face meetings, we use telepresence and a lot of video conferencing, and we try to run the technology organisation consistently around the world,” he says.

“There’s a regional CIO in place in the United Kingdom, there’s one in place in the United States, there’s one in Asia, and there’s one in Brazil. They have responsibility within the region. They’re part of my direct team and they’re part of the decision-making body around technology standards. They then in turn work very closely in their regions with the business leaders.”

Libenson has a thorough cyber security strategy to keep all of them secure. It starts by measuring the external threats through tools including anti-stuff, IPS technology, and web application firewalls that looks out for attacks.

Meanwhile, monitoring systems identify if someone breaches these defences, and if they do Experian can lock down the estate, data can be encrypted to minimise the damage and access cut to anyone who doesn’t have a valid reason to be there.

Libenson compares the approach to an old saying from his native USA: “If we’re walking in the woods and we come upon a bear, I don’t have to run faster than the bear, I just have to run faster than you.

“The goal of our security posture and profile is to create the most difficult environment that we possibly can for somebody to monetise externally. There’s plenty of other targets out in the industry that don’t have as robust a security profile and are much easier for the bad guys to compromise. Our goal is to be as tough as possible and to do everything we can, so that they don’t even want to try. They’d rather go somewhere else.”