Right Time Business Optimization

October 28, 2010
132 Views

Mike Ferguson presented on Right Time Business Optimization using on-demand and event-driven analytics at the Teradata Partners conference. Business optimization, Mike says, is about continuously knowing what is the best action to take and when to take it in every business process to dynamically keep a business running optimally while remaining compliant, minimizing (or I would say managing) risk and maximizing profitability.

Mike Ferguson presented on Right Time Business Optimization using on-demand and event-driven analytics at the Teradata Partners conference. Business optimization, Mike says, is about continuously knowing what is the best action to take and when to take it in every business process to dynamically keep a business running optimally while remaining compliant, minimizing (or I would say managing) risk and maximizing profitability. And doing this in the context of a business strategy and ensuring enterprise-wide execution of this business strategy. This means taking strategic objectives and ripple these down to operational levels with suitable KPIs and objectives at every level. Mike emphasized the importance of all the small decisions – the micro or operational decisions – at the front line rolling up to match this strategy.

Lots of building blocks for business optimization being deployed like business process management, SOA, BI, Performance Management, Event processing and even MDM/data governance. Enterprise Business Optimization requires all this to be pulled together – for instance using BPM to automate and integrate a common procurement process, BI to do spend analytics to manage costs, analytics to drive automate procurement decisions within the framework laid down by executives (rules), monitor expenditure and supplier performance events, and integrate the whole thing to show how all this contributes to the strategic objectives and company KPIs.

A key then is to integrate business intelligence (in the broadest sense) into all operational processes, out to customers, onto mobile devices. This has to be integrated into business processes to guide operational decisions made by people and to drive automated decisions in those same processes. Business intelligence (in the broadest sense, including data mining and predictive analytics) being used to improve every operational decision. The objective is to build on the efficiency gains of business process approaches and technology by using intelligence to improve effectiveness. Continuously guide processes to meet objectives by improving decisions.

This means service-orienting business intelligence and, in my opinion, using analytic models and data mining that can be embedded in automated decision services. The data infrastructure has to support lots of services that provide analytic results to both people and automated systems as needed throughout a process – business intelligence or recommendations. The latter is key in operational systems because the users don’t have time to use BI tools and the systems can’t use BI tools – recommendations, actions, are what they need.

Understanding the roles of the people involved is key as the human decision makers will be at different levels. It also means integrating BI and processes so that worklists and portals, process and BI components, are all integrated into a single workspace. Mike gave some great examples, though a couple were places where scores were displayed and should probably have been replaced with recommendations based on those scores. And Mike goes on to point out that ensuring consistent, and consistently excellent, customer treatment means sharing a common sent of recommendation services (what I would call decision services) using business rules and analytics (BI) in combination.

Mike sees the availability of predictive analytics in the database or warehouse as critical – moving the analytics to the data helping ensure good enough performance. I would add that using these analytics in the context of a rules engine is also critical as analytics without actions, even fast analytics, is not enough. He also points out that if we are building a decision service then we may need analytics and thus data not in the warehouse – if we begin with the decision in mind and work back to the data we need to improve that decision we may end up at data we don’t own. This means we need a more real-time information federation environment. To reduce the time to action (Richard Hackathorn’s great idea of decision latency) we need to automate data collection and federation,automate the analysis and automate the decisions themselves.

Once companies move into this real-time mindset then Business Activity Monitoring and Complex Event Processing are top of mind. These technologies allow for continuous ingestion of data, continuous analysis and continuous decision-making for always-on business optimization. Of course the data flowing in must be trusted enough to justify this decision-making! He gave a great example of a single error, that could have been detected and alerted in an event-centric environment, that resulted in nearly $2M in costs and fines for a shipping company. Of course doing this in real-time  means a change to the standard data warehouse mindset – we need to detect an event, enrich it with data, analyze it and act on it BEFORE it gets stored in the data warehouse.

This new world needs rules engines, predictive analytics, event processing and integration with business processes. In particular, he says, it requires event processing that drives always-on operational BI, that can detect and correlate events that matter from an ever-increasing stream of events. This is data in motion and requires new tools to manage it. Lots of new tools but only the very beginnings of new standards so this remains a challenging area.

Mike repeated his key point (with which I strongly agree) that there is a critical need to link all this back to our corporate strategy and objectives so we know what a good decision is before we try and automate it or help someone make it.

At the end of the day this requires an architecture with data, and the data warehouse, at the center. Sometimes it is acting as the source of intelligence, sometimes it is storing the actions taken to events. It links operational data to the strategic measures and objectives of the organization. And it supports rule engines, predictive analytics, process management tools (both for people-based and automated processes).

Top recommendations:

  • Know your business strategy
  • Know your business processes
  • Know the roles and needs of your people
  • Understand the events of value to your business
  • Ensure you have good enough data quality for what you are doing
  • Common data definitions, a shared business vocabulary (critical for rules especially)
  • Adopt the technology you need like business rules engines and an enterprise service bus