Enabling decision automation with Prescriptive Analytics

Last month Gartner published an article titled “Extend Your Portfolio of Analytics Capabilities” where they discussed the four types of analytic capabilities, their usage and the human intervention needed at each level to translate the data insights to action. The framework is captured well in the below figure – also from the same article.

Analytics-Capabilities-Framework-Gartner-September-20131

Source: Gartner (September 2013)

Many organizations have achieved fair levels of maturity with Descriptive & Diagnostic types of analytics. In these cases there is a significant time lag (days or even weeks) between the business activity happening and the managers doing the analysis to arrive at meaningful insights. When you are dealing with historic outcomes and causes, this lag is acceptable.The supporting technology is the traditional BI system that aggregates historic data at various levels.

iStock_000006198723Small-300x201

Adoption of Predictive analytics is comparatively recent. But again in most cases the analysis is a one off activity, done with historic data with the aim of detecting patterns that would help optimize future initiatives (on time delivery metrics, customer commits, customer churn, bundling recommendations etc.).

However, with the Prescriptive type the story is different. One will be hard-pressed to find an organization that has analytics driving day-to-day decisions. Ask any manager who has to deal with day-to-day operations about how s/he uses BI systems and it is likely that s/he points out two limitations that are show-stoppers in the operations world. The first one is the lack of real-time information – this is critical when your decisions affect individual active transactions (unlike the case of looking for broad patterns in rear-view mirror). The second is the dependence on IT for even simple changes. Yes there are a few Self Service BI tools – but in our experience for business users, Business Intelligence & Self-service are oxymorons.

For an operations manager the two limitations have resulted in endless extraction of data from canned transactional reports and manipulation on spreadsheet tools like Microsoft Excel. The cycle is repeated as many times as possible so that the data in the spreadsheet is as current as it can be. And soon there is the explosion of MOAS (mother of all spreadsheets) with unending pivots and groupings. Haven’t we seen these all around us?

At VSSOD, over the last 4 years we have been on a mission to convert fragile operational processes to agile ones. And a major step in the process was to rethink BI to work for day-to-day operational decision making. Falling prices of memory and the maturing of in-memory computing in the form of SAP HANA have been major blessings on the way. These two trends have been enablers for OpsVeda, a Prescriptive Analytics platform that is simplifying the way day-to-day operational decisions are made. Today the personnel taking tactical decisions on the field can have access to the relevant intelligence in real-time. And if a change is required, business configures business rules – not IT. What’s more – the system detects potential process exceptions well in advance and in many cases suggests corrective action that is needed. The user just needs to do a quick sanity check and act. Talk about intelligent context aware systems working for those who are running the show every day!!!

In our next blog post, we will discuss the components needed to create a real time operational business intelligence system that provides prescriptive insight and decision automation.

horisontal_banner91