Related Blogs


Avoid “Garbage In, Garbage Out” with Your MI Packages

Risk Management
Reporting on the results of credit decision strategies is a vital function – but it’s often given less thought than other components of risk management. With the multitude of data variables available, selecting the key performance indicators relevant to the decision being made will be essential in making sure your management information packages deliver what is needed to the right people.

Leading organisations create concise MI packages targeted to a specific audience. Here are a few hints and tips:

  1. Along with defining clear objectives when designing a new strategy also determine the MI required to report on the performance against objectives.
  2. Make use of the RAG approach (red, amber, green) to clearly indicate results against target. Amber can be calculated based on a % tolerance to the target. This approach is clear and easy to understand.
  3. Explore all data sources to optimize availability.
  4. Be aware of differing data definitions across locations. One of our clients greatly improved their data availability and hence their decisioning capabilities by creating a warehouse with data from multiple products. This provided a consistent data definition, enriched the data available for each product and enabled an improved total customer approach to be applied.
  5. Consider the audience before deciding on the MI metrics and number to be included.
  6. Create easy to understand graphics, including thinking how the colors will appear in presentations or when printed.

Thanks to Stacey West of Fair Isaac Advisors, who contributed information for this post.


Infographic: Keep the Car and the Paintings, I’ll Take the Gold!

Cars, diamonds, gold, stock, art, baseball cards – which collateral holds most sway with bankers when it comes to approving a consumer loan?

We included this somewhat unusual question in the latest installment of our quarterly survey of North American risk managers. We asked bank risk officers which type of collateral, other than real estate, they would prefer for consumer loans, if they had their druthers. Respondents had six choices, ranging from very traditional to not so traditional. Answers are shown in our infographic:



Perhaps not surprisingly, gold was the big winner. Au (as my high school chemistry teacher would say) was favored by 41 percent of respondents.

The next most compelling form of collateral was stock certificates (29 percent), followed by cars at 14 percent. Does that mean bankers have more faith in the stock market than the used-car market?

Further down the list were diamond jewelry (10 percent) and works of art (4 percent). Bringing up the rear was a baseball card collection at 2 percent. So don't plan to leverage your Reggie Jackson rookie card to buy a boat!

While we were obviously being a bit lighthearted with this question, the results are interesting. Although we set up our question by telling respondents to assume that the value of all the items could be verified, there was a clear lean towards traditional types of collateral. I suppose that is to be expected. When money is changing hands, it makes sense that lenders would want to minimize risk – be it real or perceived.


Deep Learning Analytics: The Next Breakthrough in Artificial Intelligence?

With countless analytic advances based on our understanding of how the brain works, many of us in the data science field were intrigued to hear about Google's recent acquisition of artificial intelligence (AI) company DeepMind. It reinforces the potential of a new field within AI, that of "deep learning." Indeed, we here at FICO are well aware of the business applications of artificial intelligence, having pioneered the use of these methods to solve business problems, most notably our fraud detection software that utilizes neural network models.

Much like neural network fraud models, deep learning research focuses on mimicking how the brain learns. A neural network model utilizes “shallow” learning where the inputs are complex domain-specific variables. By contrast, deep learning allows simplistic inputs but leverages deep networks to learn complex relationships between these simpler inputs.  In this way, deep learning is trying to get closer to the intricacy of the human brain.

Looking at model architecture helps to illustrate the difference. The figure below represents the architecture of a typical neural network. The "hidden layer" is the mathematical core of a neural net. It selects the combinations of inputs (e.g., dollar amount, transaction type) that are most predictive of the output (e.g., likelihood of fraud).

Shallow Learning (Fraud Example)


A single hidden layer is typically what classifies this network as shallow. By contrast, deep learning networks include multiple hidden layers between the input layer and the output layer. This increased complexity means we can solve more multifaceted problems using larger, more complex data sources – a huge plus in these days of Big Data. For those who would like to dig deeper into these differences, I encourage you to read my more in-depth analysis in a recent FICO Labs Blog post.

As an example of complex problem-solving, deep learning shows strong promise in the areas of image and video analytics. Analytic recognition of images is a rather multidimensional process, involving the consideration of multiple features like edges, pixel clusters and shading that would need to be automatically learned.

Of course, deep learning is in its relative infancy when compared to its shallow learning counterparts. At FICO, we've honed our fraud models for more than two decades, and in fact, the use of less complex shallow networks provides multiple benefits, including fast computation, and even more importantly, narrower scope to prevent over-training of the models. This results in robust models that can make quick decisions – paramount for fraud detection.

As the potential for deep learning is being recognized, it's no wonder that Google and others are so interested in these methods to leverage their Big Data stores. We at FICO continue to research these types of advances to ensure that we are constantly building the most predictive analytics for our clients.


Originations: Are You Ready to Step onto the Cloud?

Cloud vs on-premises
Should our company deploy origination software on-premises or in the cloud? We’ve been fielding quite a few questions along these lines after recently launching our own FICO® Origination Manager 4.5, the first release that can be deployed either way. In this post, I’ll answer some of the top questions we’re hearing, including:

  • Why move to the cloud? What are the operational benefits?
  • Will it hinder regulatory compliance?
  • Will our data be secure?
  • What are the financial benefits?

The answers below can help you decide whether on-premises or cloud delivery is right for your organization.

The basics: Why chose cloud-based software over traditional on-premises?

Cloud-based delivery shifts operational responsibility for everything associated with you using the software – software license, implementation, servers, associated data storage, data center space and utilities, security, and more – to the provider. As a result, the cloud offers strong benefits in several areas:

  • Fast time-to-value, sincesoftware can be available in just a few days, with full configuration in weeks. Client configuration is also available more quickly in the cloud environment, compared to “getting in line” in the long project queues that many IT departments have.
  • Optimized performance, with your vendor implementing all application software, operating systems and hardware, plus updates and upgrades. Again, you won’t need to negotiate these projects with your internal IT department.
  • Seamless scalability, an important capability if your originations business is growing fast, or has demand spikes driven by seasonal factors or promotions. FICO delivers capacity on-demand to handle any increases in traffic, helping to ensure high throughputs and an excellent customer experience.

For our clients, we deploy FICO® Origination Manager in the cloud (at our data centers) and establish interfaces with other key systems, which can be inside “the four walls” of your organization. Users access the software through a highly secure online connection; data generated by the application can be stored in the cloud or on-premises, at your organization.

Regulatory compliance: Am I covered in the cloud?

In short, the answer is yes. Choosing cloud-based solution that meets PCI and/or other regional data security requirements, as well as a provider that has deep experience with regulatory compliance, can help to ensure adherence to local government regulations. In the US, these include the Truth in Lending Act, Equal Credit Opportunity Act and others. FICO has worked with many customers in this area, and can help incorporate appropriate regulatory requirements into your solution, wherever you do business worldwide.

Data security: Will my data be protected in the cloud?

Again, the answer is yes—provided your vendor follows industry-standard data security protocols. Some companies, however, feel more confident and secure with storing financial data in-house at their own data centers. This translates into a significant ongoing security burden, including maintaining all available updates to their origination software, as well as associated physical, network and database security measures.

With many organizations struggling to rein in their data center sprawl, cloud-based software and storage are attractive and highly secure. FICO maintains the utmost in security measures at every layer of the application delivery stack, from operating system and network, to secure customer access, to multi-layer security at our data centers.

The bottom line: What’s the difference financially?

Across all industries, many companies are choosing cloud-based software delivery because it can offer greater financial flexibility. Importantly, the cloud delivery model changes the very cost structure of software. Whereas on-premises deployments present a significant, upfront capital expense (CapEx), cloud-based delivery typically entails a (much lower) monthly operating expense (OpEx).

Getting started with cloud-based delivery is usually less costly (and much faster) than with on-premises software. For most enterprise applications, the cost advantage extends as far as a decade after initial deployment.

But that doesn’t take into consideration a final and very important cost benefit of cloud-based delivery: Your software will always be up-to-date. You don’t need to plan for hardware or software upgrades every few years.

Overall, we find that most customers will see operational, financial and customer-centric benefits from moving to a cloud-based origination solution. But for some, on-premises is still the right answer.

What are your thoughts on on-premises vs. cloud origination solutions?


Attack of the Cyber Men (and Women)


This week I am attending and presenting at couple of industry events: the Fraud Conference and the FICO-sponsored Fraud Women's Network. Both of these events are focused on the pervasive fraud threats posed online and the measures being adopted to improve cyber safety and security.

The events and the challenges of authentication in the online environment have brought to mind one of my favorite conundrums: "The more you take, the more you leave behind." The traditional answer here (spoiler alert!) is footsteps, but it could equally apply to one's online footprint. The larger your presence and engagement online, the greater the potential risk of data or credential compromise.

Past wisdom has suggested that adequate authentication defenses are vested in something you know (like a password or personal data) and something you have (such as a device, card or token), with a drive toward the introduction of dynamic, variable challenges rather than a reliance on simply static and time-agnostic data.

But this wisdom is being challenged as personal data content online becomes ever richer — especially through social media profiles — and availability of data increases as a consequence of insecure customer processes, compromised devices, or unintended or manipulated disclosure.

Current wisdom calls for the use of out-of-band (i.e., a different concurrent channel) authentication, geo-location and proximity correlation, and even behavioral biometrics (such as screen navigation). And the nirvana is, of course, the use of biometrics or something you are.

Whatever level of authentication is adopted, however, this should of course only be one element of a layered defense. To criminals, compromising personal credentials is interesting but is just a means to an end. Their intent is to use such data to plunder funds. So the way to best defend assets (data, goods, services, finances) is by running checks across every interaction on a holistic basis.

Over the last couple of days, industry leaders have focused on keeping an acceptable balance between risk exposure and customer experience. And this means that holistic checking needs to be seamless, and only interventional where the level of risk is high or the liaison with the customer demands it. This speaks to a sound enterprise fraud management approach.

Cyber risks may indeed be pervasive, but the more the payments industry takes a concerted and holistic layered defense approach, the more likely that compromises will be contained and potential loss mitigated.


Bad Debt Ruining Your Balance Sheet? Look at Bust-Out Fraud

Lock and chain
Sometimes I feel like fighting financial fraud is what PIMCO’s Bill Gross calls
“[s]ort of a reverse ‘Sisyphus’ moment – two steps upward, one step back.” He was talking about investment yields, but the same principle applies to financial organized crime. Despite increased awareness and prosecution of financial fraud (two steps up), its perpetrators are increasingly sophisticated and operating 24/7, worldwide (one step back).

Bust-out fraud damages every organization

I’m talking about bust-out fraud rings, of which there are many examples across industries. Credit card fraud is the classic example; fraudsters use stolen identities to apply for and use credit. They make regular payments for a short while, then quickly max out their credit and abandon the cards, with no plans of future repayment.

Bust-out fraud is a major force wherever there is a trusted exchange of money – from consumer and commercial loans, to healthcare and home insurance reimbursements, to tax refunds. From recent headlines, here are a few examples:

Fraud Ring Headlines

Skewed financial results

Because bust-out fraudsters appear, on the surface, to be legitimate customers, the financial damage they incur is typically classified as defaults or other bad debt. But bust-outs are a fraud problem, not a default problem, and a big one.

Analysts estimate that between 10% and 15% of all banks’ unsecured bad debt is actually bust-out fraud, resulting in tens of billions in losses every year. Financial results are skewed, and untold resources are spent chasing down debt that will never be collected.

Big Data: A giant step upward

Today, advances in Big Data analytics are making it possible to achieve big breakthroughs in detecting and busting bust-out fraud rings. This week, I’m speaking on this topic at the NACHA Payments 2014 conference, at the session: “Busting Fraud Rings with Social Link Analysis.”

You might want to keep an eye out for my next post; in it, I’ll go into detail about social link analysis, a key enabling technology in stopping today’s sophisticated fraud rings. By proactively exposing fraud rings’ biggest vulnerability, shared identity information, social link analysis uses more information to identify more relationships – and bust more bust-out fraud rings.

Until then, thanks for your comments, opinions and social shares below. And may all your pursuits take you upwards, not back.


Banker Survey: Delinquencies Expected to Rise

For the first time in a long time, I sense a bit of pessimism creeping into our quarterly North American risk survey results. In the latest survey of U.S. and Canadian bank risk professionals, expectations for delinquencies on credit cards and auto loans, as well as total delinquencies on all consumer loans, reached their highest levels since Q4 2011.

Risk Survey Delinquency Predictions

In the survey, 44% of respondents expected delinquencies on credit cards to increase during the next six months, while 35% said delinquencies on car loans would increase. Some 43% expected the total number of delinquencies on all consumer loans to increase.

These numbers certainly don’t signal any sort of imminent catastrophe. However, this is the fourth consecutive quarter in which pessimism has increased with regard to delinquencies on auto loans and credit cards. That’s starting to look like a clear trend.

The glass-is-half-full crowd can interpret this trend as a healthy sign after lenders spent much of the time from 2008-2013 constricting credit availability and avoiding risk. These numbers mean more people are gaining access to credit. But the glass-is-half-empty crowd will be quick to say that we need to keep a close eye on risk levels. If actual delinquencies reach an uncomfortable level, then lenders may need to pull back again.

Of course, both views are true. You always need to keep an eye on credit quality, but with card delinquency levels at record lows and US credit quality approaching pre-reccession levels, it feels like we are not in imminent danger.

Interestingly, our survey found bankers expect consumer re-leveraging to continue and perhaps accelerate. In the survey, 65% of bankers expected average balances on credit cards to increase over the next six months—the highest percentage in our survey’s four-year history. In addition, 61% expected the amount of new credit requested by consumers to increase, which is the second-highest figure ever recorded for that question.

When it comes to small business lending, our survey indicated that things would remain status quo, and perhaps improve just a bit. Among those polled, 94% expected the amount of credit requested by small businesses to remain steady or increase. Some 84% of respondents believed the amount of credit extended to small businesses would remain steady or increase. And 74% expected the supply of credit for small business loans to satisfy demand.

Risk Survey Small Business Outlook

To view all survey responses, I invite you to download the full report. It will be interesting to see what the next few quarters have in store, particularly for consumer loan delinquencies. Hopefully lenders are reaching a point of healthy equilibrium—expanding access to credit without taking big risks.


As Debt Collection Complaints Rise, Greater Regulatory Scrutiny Follows

It is no secret that debt collection practices are under increasing regulatory scrutiny. And with three recent reports revealing a rise in consumer complaints related to debt management, it's a safe bet that state and federal regulators will continue to keep a close eye on the collections industry.

In 2013, according to the Federal Trade Commission (FTC), debt collection had the second-highest number of all consumer complaints reported. Each year, the FTC releases a summary report detailing the complaints reported to its Consumer Sentinel Network (CSN), an online database available only to law enforcement. Last year, the CSN received more than 2 million consumer complaints from numerous federal and state agencies, US and Canadian Better Business Bureaus, and consumer organizations. 

The FTC report also revealed that the number of debt collection complaints (10%) was the second highest, behind only identity theft (14%). However, identity theft complaints had declined a significant 20% from 2012, while debt complaints rose by 2.5%. In addition, the report’s state-by-state breakdown shows that in nearly every state, debt collection complaints were either the highest or second-most reported category.

A recent study by US PIRG Education of consumer complaints to the Consumer Financial Protection Bureau (CFPB) also reinforced the growing frequency of debt collection complaints from consumers. Looking at July 2013—when the Bureau began receiving debt collection complaints—through January 16, 2014, PIRG noted that debt collection represented the second-highest volume of complaints behind those related to mortgage. During that timeframe, the CFPB received an average of 2,000 debt collection complaints per month.

In a third debt collection-related report, released several weeks ago by the CFPB, the Bureau confirmed that debt collection concerns now comprise the largest monthly source of complaints it receives.

While the CSN and CFPB complaints represent a relatively small percentage of the approximately 30 million Americans with debt in collections, the increase in complaints cannot be ignored. Keep in mind that the complaints inform regulators of problem areas that need to be addressed. In a recent post, I discussed the CFPB taking the first step towards drafting new debt collection rules by inviting public input on a wide range of issues. I stressed the importance of those involved in debt management to share their insights with the Bureau. In February, FICO submitted its comments along with more than 20,000 other responses.

Besides engaging in the rulemaking process, it’s equally important to manage and resolve customer concerns internally before consumers take further action. FICO is working with our clients to implement analytics and automation that help address various regulatory challenges, both here in the US, as well as abroad where conduct risk has become a central regulatory focus. (For more on conduct risk, read my colleague David Molyneaux’s recent blog post.) The stakes are increasing, and proactively addressing customer frustration is not only a best practice, it can also keep your organization in good standing with the regulators.


Infographic: How Fast Is Card Fraud Detected? 5X Faster Than the Eye Blinks

In financial services, many things are described as “real time,” including numerous online banking and payment transactions. However, there is no universal definition of real time. Is it real time if a transaction or event takes place within 10 seconds? One second? Certainly most people would say the time it takes to blink is fast enough to be considered real time.

We’ve just developed a new infographic that may make you re-evaluate what real time means. It illustrates how fast FICO® Falcon® Fraud Manager detects bogus credit and debit card transactions—in just 40-60 milliseconds. That’s 5X faster than the average person blinks. It’s faster than a car airbag inflates. It’s even faster than a helicopter blade rotates.

Click on the infographic above to view it in full or go to

Perhaps even more impressively, during those 40-60 milliseconds, the Falcon software completes 15,000 fraud detection calculations based on many pieces of data associated with each payment card swipe – including transaction amount, merchant profile, transaction location, point-of-sale device, time of day and account history.

The result is more security for card issuers and merchants, and a seamless customer experience. Everybody wins … except the bad guys.


Paying Cash May Cost Americans More Than They Realize

A recent New York Times article "Newly Wary, Shoppers Trust Cash" highlighted a rather troubling trend for those of us in the payments industry: Americans making a conscious effort to use cash instead of credit cards. The catalyst for this paradigm shift isn’t a need to conserve funds or avoid finance charges, as you might expect. Instead, these consumers are choosing cash over plastic to avoid becoming future fraud victims.

With steady news coverage of data breaches and other fraud, it makes perfect sense that consumers are feeling a bit tender. But the New York Times article gave me pause. Perhaps we, as an industry, have not done a good enough job of instilling trust with consumers. Let’s face it, carrying large amounts of cash is not necessarily safe, nor is it convenient. And using credit responsibly is a smart way to build a credit history, which would benefit consumers in multiple ways over the long run.

My colleague TJ Horan recently blogged about consumers wanting payment systems that are frictionless and secure. There’s a powerful message here, and it’s coming from the customer!

So how do we build trust back into the equation? How do we encourage cardholders to reach for their payment cards with renewed confidence? Here are a few ideas:

  • Crow about your security practices. The days of discreetly operating a top-shelf fraud department are somewhat over. Make consumers aware that you are fighting the good fight and using the very best tools in the marketplace. You don’t have to give up trade secrets to reassure your current and future customers on how safe they truly are. And as part of this communications, you should…
  • Show that you plan for the worst. For instance, your website should have a “safety and security” section that features information about what customers can expect should they fall prey to some form of financial crime. Include how you plan to share important information, including your choice of communication channel like email, phone or US mail.
  • Focus on consumer education. Many consumers think that a decent credit score is maintained simply by having available credit. They don’t know or may forget about the importance of using credit regularly and responsibility to establish an adequate credit history. Educate your cardholders on smart credit behaviors and how these can affect their credit scores.
  • Use loyalty programs and other incentives. Contests and giveaways based on card usage are truly great ways to inject new life into any card program. Incentives work, and it’s also another great way to build ongoing customer relationships.

If you have additional ideas, please share them in our comments section!

Customers are constantly discriminating between safety, convenience and quality in the marketplace. It's our job to find a way to deliver on all of these components whenever possible.