Taking Humans Out of the BI Loop

Ann All

In business intelligence, there is a lot of emphasis on getting the "right information" into the hands of the "right people" at the "right time."


Yet BI's biggest value may be in keeping information out of people's hands. Imagine if analytical tools could pull information from real-time data streams and send actionable recommendations to a rules engine that applies them to the appropriate transactions. That's the scenario laid out by Joe McKendrick on ebizQ's BI in Action blog.


Though we tend to admire automation for its efficiency, we often don't completely trust it -- which means it could take some time before folks are comfortable with this kind of automated decision-making.


Yet James Taylor, a partner in a company called Smart (Enough) Systems, predicts that 95 percent of operational decision making could be automatically performed by software, writes McKendrick. Though changing company culture may present a challenge, the technologies for achieving this kind of automation are already in place. Service-oriented architecture, business process management and business rules management can help bridge the gap between near-automated decision-making and a fully-automated process, says Taylor.


I got a similar view from Charles Nicholls, the CEO of SeeWhy Software, in our October interview. According to SeeWhy's research, 90 percent of companies say that embedding BI into their operational processes would improve their business performance. Fifty-five percent of companies say their operational BI alerts arrive after the event and are not specific or contextual enough to be acted upon.


While BI applications are traditionally built so that data and data logic are both stored in the database, Nicholls says companies implementing SOA are moving toward a more loosely-coupled environment in which data and data logic can be separated. He says:

So analytics can trigger business process management systems and trigger the process for retaining customers, or retaining stock, or those kinds of things. Therefore the nature of what we call BI -- for the sake of a label -- is changing. It's not just about producing reports, but about creating these kinds of autopilot systems.

While some strategic decisions cannot be automated, says Nicholls, "there are lots and lots of relatively low-level operational decisions that can benefit from automation." Nicholls advises companies to focus on specific pain points where automated decision-making can give them a big operational lift, such as getting and keeping stock on store shelves.


I blogged last month about several companies that were working to bring BI closer to their business processes. One example is Fannie Mae, which uses software tools to analyze its transactional data to determine which Web pages yield high rates of abandonment from loan applicants.

Add Comment      Leave a comment on this blog post
Mar 24, 2008 4:28 AM Paul Wallis Paul Wallis  says:
Hi Ann,This has actually been going on in Oil & Gas for many years. I used to work with control and closed-loop optimisation systems. These systems would take all of the decisions needed to control manufacturing plants, day-in and day-out, even being linked to to commodity markets to make the most cost effective multi-million dollar decisions - the mother of all automation! Reply
Mar 25, 2008 4:00 AM James Taylor James Taylor  says:
AnnThanks for the shout out. The challenge, as Charles knows, is that embedding analytic insight into high-volume, transactional processes is not just BI-but-faster, it requires new techniques. The approach we discuss in our book and that Charles' company uses (more or less) combines business rules and executable analytic models to improve each transaction by improving the decisions that impact it. Data mining, analytics and BI can be used for many kinds of decisions but improving operations and operational decisions is a little different.Look forward to reading your blog in the future.JTJames TaylorAuthor, with Neil Raden, of Smart (enough) Systems Reply
Apr 2, 2008 5:38 AM Vincent Fong Vincent Fong  says:
It's not actually that the labelling of BI is changing but the realisation that operational decisions, that is, decision made by frontline, in your client/customer faces, staff may seem insignificant but the sum total of all such decisions have a dramatic impact on the organisation in terms of who/what/how those operation decisions are made and what/where/when/how they impact the organisation and its broader strategies.The recognistion that automating decisions at the operational level can limit "bad" decisions made at a huge costs (mostly hidden) to the organisation is / should be the driving point. Reply
Apr 7, 2008 12:13 PM Pat Fitzpatrick Pat Fitzpatrick  says:
Ann:Interesting comments. This is definitely the direction that business is taking, especially in manufacturing. I spent 12 years in robotic automation and found over time that the limit to getting to the "lights out" manufacturing was the lack of "real time" intelligent decision making. This provided a high risk area that required manpower to monitor or forced manufacters to go at it blind, which provided potential cost adders for quality problems that were missed in a lights out environment.My new company is involved in pattern recognition using neural networks. The neural network approach is the future in business intelligence. The limit has been the lack of algorithms that actually performed as advertised. The technology is coming of age and will dramatically alter the fortunes of the early adopters. The next 10 years are going to be very interesting. Pat Fitzpatrick Reply

Post a comment





(Maximum characters: 1200). You have 1200 characters left.



Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.