Role of Business Intelligence in Process Improvement

Bill Gassman spoke on the role of Business Intelligence – BI – in process improvement. Bill means “big BI” – everything to do with intelligence about your business, the discipline of BI and analytics, not just a “BI” product. The road to intelligent operations he says has “haves” and “have nots” – some have BI and BPM automated, monitored and managed but many do not. Some will therefore be able to bring BI and BPM together effectively, some will not – they will struggle to find the data, integrate it, understand it and apply it to their processes.

Bill Gassman spoke on the role of Business Intelligence – BI – in process improvement. Bill means “big BI” – everything to do with intelligence about your business, the discipline of BI and analytics, not just a “BI” product. The road to intelligent operations he says has “haves” and “have nots” – some have BI and BPM automated, monitored and managed but many do not. Some will therefore be able to bring BI and BPM together effectively, some will not – they will struggle to find the data, integrate it, understand it and apply it to their processes. Three things to discuss then – what technologies are converging to build intelligent operations, how are companies using BI to improve processes and what are some of the best practices for doing this?

First, what technologies are converging to build intelligent operations

One of Gartner’s focus areas is pattern based strategies that allow you seek, model and adapt and so deliver agility. This agility is dependent on productivity, awareness, adaptability (reacting to unexpected change) and flexibility (options for expected change). But the first step is an ability to seek and see the changes that are impacting our business – both expected or knowable changes and the real outliers or “black swans”. Agility requires an ability to respond to both. Specific technologies:

• Bill talks about operationally focused BI that is a spectrum from historical BI (what happened) to in-progress BI (dashboards and performance monitoring, business activity monitoring) and predictive BI (analytics really like data mining and predictive analytics).
• Event processing is another relevant technology in Bill’s taxonomy. Obviously as the number of sensors grows plus the growth in event generating processes the number of events expands dramatically. Making sense of all these events is valuable and drives demand for event correlation engines that can generate more complex (business) events, kick off processes or call decision services.
• High performance operational queries – an ability to do complex queries against operational data quickly – is essential and is something that is being addressed by a new wave of analytic technologies (sometimes labeled Big Data technologies) including in-memory databases and in-database analytics.
• Operational visualization is also a growing topic with the ability to rapidly visualize large amounts of data in a real-time enough way to be useful in operations. Of course I would say that if you really want to respond to data in real-time, showing it to someone cannot work as people just aren’t reliable as a real-time component because phones ring, bathroom breaks are essential etc.
• Bill drew a distinction between BI within a process and BI about a process. BI within a process is about improving decisions, adding understanding and determining more appropriate actions. Embedding a decision in the process explicitly rather than just displaying an “approve/deny” step and hoping that they understand the decision. Classic description of Decision Management – make the decision and its rules explicit so you can apply analytics and continually improve. This is different, of course, from using BI about a process to see how the process is going and look for ways to improve the process.

How are companies using BI to improve processes?

Analytics are everywhere in logistics, transportation, customer contact, websites, fraud detection and more. But many companies have an uncoordinated analytics approach because the applications they buy (or build) have analytics embedded that are separate and not linked. He also pointed out that situation awareness means more than a dashboard, it means a decision support environment with collaboration, events, deviation from forecast, risk factors and more.

Bill sees a maturity curve :

1. Live dashboards but no action taking – an airplane map
2. Real time console but human decision makers – a manual cockpit
3. Prescriptive advice – a proximity alarm say
4. Autopilot feedback loops – business rules controlling decisions so the system can (mostly) fly itself
5. Autonomous – simply telling the system what your goals are and letting it manage the behavior

Examples of intelligent operations include recommendation or ad engines especially online, call center operations, supply chain logistics, airline operations, casinos like Harrahs and public venues like Disneyland. While real-time visualization, monitoring and event handling are critical to the various scenarios described, none of them work unless some of the systems involved are able to make decisions on users’ behalf – there often isn’t time to ask a person to make the decision.

Technologies are evolving as mega vendors acquire BPM and everyone with a BPM product acquires BI or analytic technologies – IBM with SPSS/Cognos/Coremetrics, Oracle acquiring Siebel/Hyperion, SAP acquired BusinessObjects, Tibco acquired Spotfire. Interestingly all of them also have business rules management systems. The stacks are not integrating all that rapidly, however, so short term projects may need to take a “best of breed” approach and/or some custom development.

What are some of the best practices for doing this?

• Build a business case based on the value of reduced delay, improved risk tolerance/management, new opportunities. Offset this value with technology costs, infrastructure, governance, organization change/acceptance (sharing information can be culturally difficult) and failures. Remember to describe where it will be used, who will be benefiting, how will things change and how will you measure the value of the solution. Remember, though, you can (and should) start small
• Network competency centers – make sure the BPM, Application Development and BI competency centers collaborate.
• Collaborate on metrics across the functional groups impacted so that there is a shared view of how success will be measured
• Adopt pattern based strategies – gather information to discover new patterns, model solutions and deploy a monitoring system based on the model to spot the patterns and do something about it (tell someone or have the system do something using Decision Management) while keeping a continuous improvement loop going.

To put it all together you need to seek new patterns in your data and event streams, analyze this using BI and analytic tools, model your decisions (my words) using business rules, define your responses as business processes and maintain a continuous improvement loop.

Recommendations:

• Combine CEP, BI and BPM to develop intelligent operations
• Smart small but start now as the cultural learning curve is steep regardless of products
• Define monitoring points, metrics and triggers when designing processes – put the decisions explicitly into your processes I would say
• Address governance through collaborating competency centers
• Measure and market the benefits of adding intelligent to business and IT operations