The need for companies to maintain competitive edge through the application of data analytics is gaining broad recognition: less well known is FICO, and the organization’s efforts to build analytics solutions that unlock the business value in very large, unstructured data sets. Over the past several months, the San Jose, California-based analytics and decision management software company has introduced innovations to support analysts working with Big Data, but also to extend access to its capabilities to a broader community.
Founded in 1956, FICO is perhaps best known for the FICOScore – the standard measure of consumer credit risk in the U.S. – and for solutions that help customers manage credit accounts, identify and minimize the impact of fraud, and customize consumer offers. Among its clients, FICO counts nine of the top 10 Fortune 500 companies, two-thirds of the top 100 global banks, 90 of the 100 largest U.S. financial institutions and all the 100 largest U.S. credit card issuers. FICO also works with many of the leading insurers, retailers, pharmaceutical businesses and government agencies (has a personal finance management tool). Its offerings fall into three broad categories: decision management tools, including software for business rules management, model development and optimization; decision management applications, or integrated systems that apply analytics, decision logic and industry expertise to strategy and execution across the customer lifecycle; and standard scores and models to manage risk, identify opportunities, and help clients forecast customer behaviour.
FICO’s strength in credit, risk and decision management has translated into a number of ‘firsts’, such as the first commercially available credit and insurance underwriting scoring systems, the first adaptive control systems for managing card accounts, the first neural network-based fraud solutions, and the first analytic systems for retailers to optimize offers and their timing. And this long term experience with analytics has allowed FICO to more recently rise above a lot of the Big Data noise, to differentiate through long term experience in the credit, risk and decisioning domain, and through expertise in the development of appropriate tools. As Andy Flint, senior director for analytic product management at FICO put it: “You hear Big Data analytics come up a lot now in conversation. But we have really been doing this throughout our 60 year history. FICO was founded on the premise that we can make better decisions, and better decisions on behalf of our customers, by leveraging historical data, by leveraging empirical data. What sets FICO apart is our position that analytics should really be about making better, smarter, more profitable and more customer delightful decisions.” Flint contrasted this FICO approach with that of other BI organizations that may be less focused on generating insights that are actionable.
Recently, FICO has added to its innovation roster with new analytics capabilities designed to turn the Big Data challenge to opportunity. As Stuart Wells, chief technology officer at FICO explained: “Modeling massive data sets can be challenging, because most traditional data analysis tools have been built for the ‘small data’ world, with highly structured, mostly numerical data. Today, data volumes are orders of magnitude larger, they defy a simple ‘rows and columns’ structure, and they’re mostly made up of messy, unstructured information like text, voice and even video.”
To help companies create actionable insight from this messy, raw data, the company has introduced the FICOModel Builder for Big Data, which uses text mining algorithms, the company’s Semantic Scorecard formulation, and embedded Lucene and Tika indexing and extraction libraries to find associations in keyword, concept or contextual scouring of large text-based data sources. A part of the FICODecision Management Platform, the tool is designed to enable data scientists to design and deploy predictive models, based on signals identified in the data. As use case examples, Flint described the search and analysis of massive amounts of customer behavior and other data that are conducted each time a credit card transaction is approved, or the scour of call centre text files at the end of the day to model and predict where the best outcomes may be on follow up calls – and where work should be prioritized for the next day.
To manage issues around Big Data volume, the Model Builder integrates with Hadoop, the open-source storage framework, and works with Cloudera’s proven, enterprise-ready Hadoop distribution. The solution also supports R language, a free programming software that is widely used by statisticians for developing data analysis. This outreach to users of open source technology is designed to accelerate development efforts by allowing clients to access the best the community can offer, and to extend the reach of FICO solutions by making the solution easier for clients to adopt. And to further drive solution adoption, the company has also introduced PaaS access via the FICOAnalytic Cloud to its Decision Management portfolio, which includes software for business rules management, predictive analytics, data and text mining, optimization, visualization and reporting. A roll out over 2013, the company plans covered deployment of additional offerings on the Analytic Cloud, such as SaaS for social and mobile customer interactions (FICO Adeptra Mobile Services platform), for social campaigns (Customer Dialogue Manager), for creating extend highly personalized, targeted offers ( FICO Analytic Offer Manager), for managing delinquent accounts (FICO PlacementPlus), and for making instant credit decisions (the FICO LiquidCredit service).
FICO’s foray into delivery of SaaS tools is a relatively recent venture. The company “was born as a services company,” Flint noted, and “our success was based on our ability to translate the concepts of operations research, statistics and what you might now call data science and information technology, and marry that with an understanding of the business.” As a result, the company, unlike many other organizations does not have a shortage of ‘data scientists’, “magical analytics superheros” who were described by Flint as having “enough mathematical background and understanding of operations research and statistics to ask the right empirical questions and who also understand the computing realities and complexity of algorithms – how they scale with the size of data and the width of that data and what you can practically answer in a short enough period of time to make it interesting – in addition to an understanding of the business. It’s very hard to hire these, but at FICO we foster and develop that skill set to service our healthy services business.”
For the most part, FICO directs its arsenal of data skill and analytics technology at very large client organizations. However, Flint anticipates the entry of the SMB community into the analytics game with the breakdown of barriers like the data science skills gap, and SMB’s shortage of computing platform and software tools. FICO’s Analytics Cloud and the Decision Management platform it will house is viewed as one means to break down these barriers: by putting analytics and decision management design tools in the cloud, the company expects to make the technology more widely available. And to address the skills issue, cloud services will be bolstered through the creation of an online community of analysts that can provide ongoing support. Cloud superheros?