How credit scoring pioneer FICO is extending its reach into Digital Twins of Organizations

FICO pioneered data analysis tools for improving business decisions. It later invented credit scoring algorithms underpinning the $17 trillion US consumer credit market. Now, it is innovating new tools for building Digital Twins of Organizations (DTO) that promise to improve profitability, reduce risk, and create more customer value.

The idea of DTO has been around since 2018 and championed by the Process Mining industry. FICO’s entry suggests a new approach focusing on business outcomes rather than optimizing processes. For example, it will allow enterprises to build diverse models that can be integrated into business operations to reduce operational, credit, fraud, attrition, and insurance risks. 

Key components of the new platform include better tools for: 

• Data connection and ingestion.
• Applying analytics and ML to enterprise optimization.
• Lifecycle management for decision models.
• Simulating new business scenarios, pressure testing possible changes, validating strategies, and simulating KPI outcomes.

FICO Chief Product and Technology Officer Bill Waid says:

The idea of digital twins is to create a safe workspace where you can harness all the complex factors that determine business success – financial transactions, consumer behaviors, individual preferences, demographics, geopolitics, macro-economics, etc. – to replicate and simulate real world outcomes, based upon verifiable data from all your actual interactions.  And by experimenting with the parameters, you can create optimal strategies in much shorter change cycles.

For example, it could help enterprises strategize and test e-commerce offerings and adapt them to how the market receives them. In this case, the digital twin approach can simulate unexpected events, see how the business responds, and develop appropriate resiliency plans. This represents an adaptation of digital twins in civil engineering that helps teams understand the interplay between the physical variables and the structural design that cuts across multiple inter-connected dimensions. Looking at dimensions in isolation only yields a partial view. This is where modeling and math become so valuable.  Waid explains:

In many respects, that's where FICO’s approach is coming from, although we are applying it to the simulation of business operations, both real and hypothetical.  We can capture and replay the layered distributions of human behavior as well as manipulate market and economic forces.  In enterprise decisioning, we’re not dealing with a three-dimensional world, but we are dealing with an equal amount of complexity across banking, insurance, telecommunications, and other industries.

From model management to digital twins

FICO has continued to innovate new tools for modeling and optimizing complex decisions since its founding in 1956. Over the last decade, it has been integrating its sophisticated modeling tools from separate applications into an integrated platform. As a result, formerly standalone point solutions can now be addressed as part of a unified decision foundation that shares common capabilities and data assets, enabling enterprise synergies across more functional areas.

Over decades, FICO has earned a sterling reputation in analytics-driven decision management. For example, it’s credit scores business helps enterprises make smarter, faster, more profitable decisions. Over time, it’s software applications expanded into fraud, risk, decision management, and other areas. To stay competitive, it developed tools to help banks and insurers make real-time decisions on online applicants while decreasing their risk of fraud. 

Waid argues that a major part of FICO’s development effort has been to accelerate time-to-decision to increase the accuracy of predictions and to remove uncertainty about individual consumers by triangulating as many sources of alternative data as possible:

Because of our legacy, digital twins are a natural extension for FICO and an ideal showcase of the simulation and optimization capabilities within the FICO Platform.  Our clients are very focused on business outcomes. Unfortunately, business outcomes cascade, and they often influence each other.  This knock-on effect of one KPI impacting another is difficult to connect outside of a simulation because they aren’t always intuitive.  

This is something that keeps businesspeople up at night, wondering, ‘If I change X, what are my unintended consequences?’ Simulation gives them a better ability to anticipate, see, and proactively address those unforeseeable consequences before a crisis flares up in production. Digital twins help them realize that ‘Whoa, I didn't realize changing X was gonna cause that,’ and be prepared when market conditions change.

Making better decisions needs to start with developing better models for all aspects of the business and then rearranging and composing them into larger models to simulate different business ideas. These models must consider data flowing through as a composition of feature calculations, model executions, decision rendering, consumer interactions, case management, and other areas. The FICO DTO breaks decision management into four categories of capabilities: data, insights, actions, and outcomes.  

Putting this into operation requires a team effort where everyone who contributes a capability or asset, from an external data source to a new data feature to an analytic model or a decision, rigorously tests their individual contributions. Simulation is one of these capabilities, but it also acts to synthesize the interplay of all the others, to show how a given decision strategy will perform against goals, and to allow users to play with the various components of that strategy to improve outcomes. Waid explains:

Its real power is to help the business understand the implications of strategies that reflect real-world, multi-variate transactions that are representative of the circumstances in which they are likely to occur. The dialogue this stimulates across the members of the digital team and management stimulates curiosity and experimentation and streamlines business innovation.

The simulation also allows enterprises to incorporate real world data, complexity, and distributions to a part of the business and see how it performs.   As these activities are streamlined, they help business and technical users build and develop a digital company in a unified fashion.  Individuals who would otherwise be testing and working in isolated silos can come together and look at it from the business perspective. The simulation also allows all collaborators to collaborate and ideate about the simulation’s findings to form a consensus.

Beyond Process Mining

The first generation of DTOs was built on top of process mining tools that analyzed how business processes were executed to construct a more holistic model of the business. FICO’s approach has been to start with the business problem and opportunities being created and frame the technology in that context. Linking the business outcome to the decision provides the golden thread that is most important to executives in a business context. 

The horrific 2017 Grenfell Tower fire tragedy in London that resulted in hundreds of dead and missing people inspired the idea of the golden thread after it was discovered that different teams had lost visibility into the fire safety risks of building cladding materials. In the case of civil engineering, the golden thread means ensuring an accurate digital record of crucial building information from the design phase through construction and maintenance. 

In a business context, the golden thread pulls together connections across different roles, such as the Chief Risk Officer, Customer Officer, Digital Officer, and front-line staff. This ensures they are not overwhelmed with technical minutia but can manage their business and impact their markets in a language that makes sense to them. Waid says:

This is what guides and gives meaning to the rest of the process.  FICO respects excellence in each individual aspect, but we prioritize around what the business is trying to do. What are the most valuable KPIs? What is the driver? If it is shareholder value, how do we ensure it all connects?

Building the contextual glue

An essential property of digital twins, instead of just a collection of data, is a contextual glue for linking tidbits of information across different sources. This is variously called an ontology, schema, taxonomy, or semantics. In the case of a building, this might correlate information about the flammability properties of a material, how it can be safely used, and the buildings it has been built into. In a business case, it provides a scaffolding for linking together data about a customer, business operations, and outcomes across financial documents, transaction logs, databases, and third-party sources. 

FICO looks at this from two perspectives.  First is the organizational structure of the definitional data that its clients configure into the assets in the FICO platform repository.  This represents the executable domain knowledge of their business.  The FICO repository enforces a strict asset ontology.  The benefits of this are transparency, lineage, and dependency management between all assets in the repository.  

Suppose a manager needs to change a pricing model for compliance reasons. In that case, she can quickly determine where this asset is used and adjust it with full awareness of the impact and the pace of movement that this confidence imparts.  One of the repository's definitional assets is called “Business Terms.”  These are mappings that correlate ingested data to a semantic layer, which the business can define using terminology that they are familiar and comfortable with.  Waid argues:

This alone boosts the intuitive interaction between the business and the data.  Typically, what we see is that Business Terms orient around entities and concepts, like customers and accounts, and payments and merchants and devices.  What this allows is for users to bring their mental models and organic ontologies to the FICO Platform, and this greatly facilitates innovation.

The second perspective is around all the operational data that gets generated. The FICO Platform captures and formats it for digital twin simulations. Combining ontologies spanning organizational structure and operations helps enterprises wrap their arms around the connection between their digital operations and their business outcomes or KPIs.

This builds on FICO's existing tools for credit ratings, adaptive controls for revolving credit lines and better fraud models that stop fraud in two-thirds of all card payments worldwide.  The digital twins simulations build on this ontology to construct simulation models that can be saved as assets and then recomposed and reused for new scenarios. Waid explains:

Composability gives you the opportunity to easily choreograph everything and everyone into experiences for end users.  The concept of a ‘composition,’ therefore, is an ideal level for businesspeople and industry domain experts to work, innovate and flex the business.  Compositions can be thought of as the flows into which data, models, decisions, interactions, and exceptions can be woven to serve the business needs.  Often, we call them use cases, and so I frequently pair them as ‘use case compositions.’

FICO has spent the last several decades figuring out how to automate and make better business decisions. This involved everything that led up to the decision and then automating the decision. Waid believes that expanding this scope and integrating it into better and more flexible representations of executable domain knowledge is a clear and present opportunity. In the future, he believes there will be additional opportunities in the post-decision phase.  This includes understanding the quality of decisions in a way that provides formal, structured feedback. This will help guide machine learning, optimization, and other techniques to deliver more value by suggesting business improvements to managers. 

Waid is also interested in the concept of augmented management and developing the underpinnings for better collaboration between people within the organization. He says:

FICO has a very people-centric point of view. We don't believe that technology is going to solve all problems. One of the biggest challenges companies must contend with is improving synergy between people and partners.

(Pixabay)

It can be challenging to grasp the significance of digital twins, even in concrete domains like product lifecycle management and civil engineering, more so when it comes to modeling organizations. It's not just about data engineering. It also requires developing a contextual glue for tying meaningful data about things across different sources and for various types of users. 

FICO’s entry into the market, with a focus on business outcomes, offers a fresh take on the concept that promises to extend the reach of digital twins beyond more sustainable physical products and buildings to more sustainable businesses. They have a track record of bringing related innovations to the market to create more profitable credit offerings, reduce fraud, and develop adaptive business offerings. I hope innovative businesses use these tools to balance shareholder returns, a more sustainable future, and #acceleratetrust with employees, partners, and customers.