Skip to main content
Live from InterACT: Decisions On Call: How the Decision Service Provider (DSP) Model Works

Posted by Guest Blogger, Ian Turvill
From Fair Isaac's InterACT Conference, San Francisco, CA

Well, as promised, I'm plugging the gap on a description of the "Decision Services Provider" model.  James just delivered a presentation on this topic at InterACT, in collaboration with Craig Dillon, who runs ScoreNet, Fair Isaac's Decision Services Provider offering. 

63_decisionsoncall_dillontaylor_mIn short, a Decision Services Provider is "on demand delivery of analytic business decisions, through data, software, analytics, services and consulting."

James and Craig highlighted nine specific traits which together either defined what a DSP is, or represented particular conditions or requirements that the DSP had to meet, namely:

  1. Builds and maintains analytic models and business rules
  2. Provides network and processing to move data in and out of operational systems
  3. Pools necessary 3rd party and customer data to build models
  4. Accesses 3rd party and customer information on each transaction
  5. Builds and maintains application software that enables analytic models
  6. Provides necessary support services (call center, implementation, etc.)
  7. Provides consulting to determine  business value for the business owner
  8. Distributes through channel / sells to client / runs in ASP
  9. Ensures regulations and industry guidelines are followed

If I were to summarize the reasons why the DSP model is rapidly emerging, they would appear to be as follows:

  • There are a very large number of data sources on which decisions are made, and it is impossible for the typical firm, such as a bank or retailer to orchestrate access to them all.  Instead a DSP can present efficiencies to its many clients by acting as a common hub for hundreds of other data providers.
  • DSPs are consistent with the overall trend towards Business Process Outsourcing, which is often seen as a way of quickly realizing economies of scale and scope, without having to make significant upfront capital investments.  BPO is projected by Gartner DataQuest to grow to $110 Billion in North America by 2009.
  • Analytics and decision-making can be based on pooled data and expertise, rather than on the understanding of a company's own (potentially limited) experience.  This approach is already common in areas such as fraud, but it could become important in areas such as marketing, where shared data sources (e.g., customer panel data) can play a role in retailers and CPG firms' decision-making.

As an example of an existing DSP implementation, James and Craig pointed to a diverse set of organizations, including the California Department of Motor Vehicles, First American Real Estate Services, and Egg, a "non-traditional" UK bank.

Egg faced a number of challenges:  the costs of manual intervention in credit decisions were higher, its customer experience was below its expectations, it lacked consistency across channels, and it struggled to make the right data available for analytics.

After implementing a DSP as a shared service for its call center and website, Egg reduced the number of manual referrals 35% to 9%, reduced bad debt provision by 1%, and lifted overall income by 1.5%.  And, oh yes, they've had 0% downtime since February 2003!  Not a bad result at all!

related posts