Welcome to our blog series on leveraging the power of data and analytics to maximize identity results. In my previous article, I shared how it’s important to use the right kind of data to solve identity and fraud problems. I also discussed challenges associated with multisource data and how businesses can make trade-offs to pick the best data strategy.
The next step is to focus on the journey from data to decisions as it pertains to identity and fraud. This is tied very closely to the customer experience.
Keeping the Customer Experience in Mind
In the digital age, the customer experience can be a competitive advantage for businesses. As more transactions move to digital — and at a higher frequency — customer relationships are being tested and challenged with every interaction. Thus, the interaction experience can significantly influence the customer’s perception of the brand and impact future relationships with the business. Identity trust is a key factor in shaping the customer’s perception of their experience with the business.
Think about it. To provide the customer with a seamless experience, it is imperative that a business can quickly and confidently assess its trust in the identity with which it is interacting. Without the proper trust evaluation, consumers may be subject to unnecessary friction in the form of knowledge-based authentication.
The Journey from Data to Decisions
Just as credit risk or marketing decisions, identity trust decisions occur throughout the customer lifecycle and across multiple touchpoints. For example, trust decisions are made when a customer applies for a new financial product, attempts to login, makes a high-value financial transaction, contacts the call center, makes a change to their contact address, files for unemployment claims and so on. Considering the scope and diversity of these interactions, it’s important that businesses understand how identity trust decisions can influence user flows throughout the entire consumer journey. Let’s start by taking a look at the identity trust decisions along this journey.
The Identity and Fraud Differentiator
Similar to the intricacies of working with multisource data, the ecosystem of decision making in identity and fraud is also rich, complex and very different from other analytic use cases. Let’s take a look at the identity differentiator at each touchpoint along the data-to-decision journey:
Identity and fraud decisions are non-adverse actionable and particularly use information outside of tradeline data to assist with identity verification.
Traditional credit risk features focus on payment and delinquency behaviors, unlike identity and fraud features, which tend to capture insights such as:
- Affiliation of identity attributes (e.g. name and phone number)
- Stability and consistency of identity attributes across datasets
- Velocity (e.g. number of times address is observed in data in the last 14 days, association to known frauds, high velocity of inquiries, etc.)
Most credit risk models are supervised, as they’re trained to predict specific outcomes. However, identity and fraud models are often supervised and unsupervised. The models are designed to look for unexpected or isolated signs and patterns in real time.
Similar to a credit risk score, which is the most important factor in a credit risk decision, identity and fraud model results provide insights — scores, alerts and assessments — to help inform identity trust decisions. This information can help reveal possible negative intent (e.g. phone-name affiliation, device risk, identity theft, synthetic identity, etc.).
In credit risk, underwriting strategies may vary depending on the user profile. For example, thick or thin file, existing or new user, etc. Similarly, in identity and fraud, different strategies are employed for different situations. For example, a user registering for the first time may need to be evaluated with more data points compared to a returning user. Similarly, a user logging in with a new device may need a different authentication mechanism as compared to a user who has forgotten their username or when a known user wants to make a high-value transaction.
Making the right decisions requires balancing the customer experience with fraud prevention. Underwriting decisions usually involve approval or decline of a credit application, and if approved subsequent determination of amount and terms. In identity and fraud, typical decisions are to allow, challenge or deny the identity. If challenged, the consumer should be further reviewed to prove their identity. It should occur in the most user-friendly manner with the least friction.
Performance data is a critical feedback signal that allows models to improve over time. It includes positive and negative signals such as confirmed fraud, abandonments, successful login and more, all of which are automatically fed back into the existing model to intelligently adjust and enhance model performance and precision in real-time. Compared to credit risk outcomes such as delinquencies and defaults, fraud labels can be expensive, time intensive and sometimes difficult to obtain. It may require investigation by analysts, and more effort to sift through a lot of data to corroborate and to disposition a transaction.
Each step of the journey — from data to decisions — is connected. And it has the potential to directly impact the customer experience across multiple touch points. Furthermore, it can influence their perception of your business and ultimately — your brand. Now that we’ve covered the identity trust decisions along the data-to-decision journey, our next article will delve deeper into analytic techniques designed for identity and fraud, exploring opportunities and challenges associated with the latest AI and machine learning methods.
To learn more, visit our website or read prior articles in this blog series: