I got an update on GDS Link recently, having last written about them in 2010 (see this First Look on GDS Link). Since 2006 they have been helping banks and other credit issuers to build custom, tailored, customer-centric risk applications.
DataView 360, the core product, was designed to address what they regard as the key problem – that of pulling data from lots of sources to push coherent information to a decision engine. With the acquisition of Modellica in 2011 they added a decision engine of their own and they have recently added functionality for building screens and workflows. They currently have just over 40 direct clients globally and hundreds in total thanks to a number of companies that are using the product to deliver solutions to others. While some of these customers only use Dataview 360, GDS Link says that over 3/4 use both DataView 360 and the DataView360 Decision Engine. GDS Link does business in the USA, Europe, Argentina, South Africa, Russia, Turkey and the Philippines – in some cases through partnerships with local credit bureaus. Most of their customers are in financial services, traditional banking and micro-lending are a particular focus, but they also have retail, utilities, leasing and other commercial/merchant services. Focus areas include originations, customer management, fraud and analytics. Three distinct components make up their product set:
- Data access, data aggregation and process flow (DataView 360)
- Decision Engine (DataView360 Decision Engine)
- Case Management and user interfaces
GDS Link helps its clients build very custom solutions – not just configuring templates but something truly unique. While GDS Link builds many of these solutions for customers, customers are also licensing the platform and developing their own solutions.
One core aspect of the product continues to be their data connector library for US credit bureau data, assets and financial data, income verification and id verification, international data sources and more. They offer a pre-built Attribute Set that has 1,000 plus standardized tri-bureau attributes for inquiry, trades, and public records across multiple industries, account types and time periods. They also support the industry standard MISMO variables for mortgage and other lending where the ability to merge credit reports is desirable. Somewhat uniquely the definition of these attributes is completely transparent to clients so they can manage and change them as needed.
The Decision Engine is similar to my last review and supports process or decision flows, segmentation decision trees, policy rules, decision tables or matrices and predictive scorecards. The tools have some improved what if analysis and can be configured to support champion-challenger test and learn set ups.
A new reporting package – Decision Intelligence – has been added. This allows clients to feed in performance data – what happened after the decision – and then integrates this with analysis of the decision execution data collected automatically by the decision engine and stored when data is accessed through the data feeds. Clients can then compare results, check scorecard stability, see what the impact of score overrides was, evaluate population stability and score performance and much more.
Future plans include an optimization capability that will allow constraints to be defined across decisions such as a total marketing budget. Data from previous real transactions will be run through the optimizer to determine what would have been the optimal action in each case and then a decision tree will be generated that mimics these results. This tree can then be deployed to the decision engine to apply the new, optimal approach. They are also doing some work to help clients move from an account and product-centric view to true customer-centric decisions.
The toolset has a nice development sandbox across the tools with unlimited test/offline process flows. These can access live data or previous real data run through for testing (including defining links to live data in the logic and then having the testing environment use the data that had been retrieved previously). Results can be reviewed in Excel and a debug facility allows a client to walk through single transactions. Results can also be integrated into the decision intelligence layer and reviewed using the standard reports defined there.
GDS Link has been more focused on building the analytics involved in these kinds of systems since the acquisition of Modellica – building custom attributes, custom scorecards, doing portfolio analysis and helping with segmentation design. Decision Lab is an internal product acquired with Modellica and this supports analytic modeling, though most customers use existing predictive analytic modeling tools. Models are manual entered into the decision engine.
Once built, applications can be deployed to local or a SaaS environment managed by GDS Link.
revised 1/31/13
Comments on this entry are closed.