A statistican on every desktop or in every service?

Interesting article in DM Review A Statistician on Every Desktop: Deploying Predictive Analytics Throughout the Organization by David Smith of Insightful Corporation. David makes some great points about the need to make predictive analytics more available in the enterprise.

David and I have a slightly different perspective on where you should make predictive analytics available. David is talking about "every desktop":

Companies that find ways to deliver predictive analytics and reporting solutions to the desktops of their key decision-makers will immediately see greater returns on their investments in both technology and personnel.

Now me, I would like to think about how we deliver predictive analytics into every service you have in your enterprise, at least every decision service. This is not about visualization or reporting but about executable analytic models embedded in the key decision services that power your transactional systems. Examples:

• Integrating models that predict insurance risk into an automated underwriting process so that the call center and the website, as well as agents and underwriters, can make informed decisions based on those models
• Integrating models that predict fraud risk into a healthcare claims process before the claim is paid to help eliminate fraudulent pay outs before anyone manually reviews the claim
• Integrating models that predict likelihood to accept a cross-sell or up-sell offer into the next marketing action engine that can then be used to ensure that call centers, tellers, agents, website and batch printing processes all make the best possible offer to grow customers.

If you limit your thinking to how to make your decision-makers better able to consume predictive analytic models you will miss a huge opportunity to improve the way your systems make decisions.