The Evolution and Future of Business Intelligence

Register now

Discussing the evolution of business intelligence can be confusing due to terminology and ambiguous buzz words. To avoid discussions around the much abused ‘2.0’ moniker, Roman numerals will be used to simply denote subsequent maturity levels of BI. It is left to the reader to decide where BI 2.0 fits in this progression. Also, to make this article less of a theoretical exercise and to be able to more credibly identify ROI opportunities, let’s map the BI maturity levels (there will be five) to a hypothetical company: a large mobile telecom provider that communicates with its customers across multiple channels in multiple areas (service, sales and marketing, retention, etc). This is not to say the capabilities described are not relevant to other industries and other business areas. The telco is just an example.BI Gen I is about reporting, usually long after the fact. Its power has been in a technology (online analytical processing, or OLAP) that allows for very efficient slice-and-dice querying of data. With so much automation going on – just consider the massive customer relationship management system space – there’s ample data available (in data marts or warehouses) to manually hunt for and discover patterns, like underperforming departments, products, channels, or customer segments. Previously this wasn’t possible – certainly not at the speed of thought necessary for effective investigation and analysis – because data was stored (in normalized form) to support transactional systems, not analysis. Speed-of-thought analysis works much better on denormalized, dimensionally organized data that is easy to slice-and-dice. Another popular use of early BI is the so-called balanced scorecard, a dashboard that tracks key performance indicators. With a well-designed BI Gen I solution the hypothetical telco introduced above could readily answer questions about average age of the customers that leave, their lifetime value, in what channels are selling best and how that differs for age segment and geographical areas. Those questions wouldn’t take the IT department weeks of data collection, cleansing and querying, but could be answered in a brief session behind a PC (talking to a data warehouse).The power of BI Gen I should not be underestimated, because for the first time in the history of information management it allowed business-oriented staff to play with data themselves, rather than waiting for IT to produce a more conventional, noninteractive report. Still, it offered intelligence derived from historic data, and any insight is, therefore, not nearly as actionable as it could be. That put pressure on BI to move to the next level.BI Gen II is pretty much the same thing, yet this time it brings a real-time capability. This is not a feature to be easily dismissed because it implies that every step within the end-to-end process – from extraction, transformation, loading to the slicing and dicing end user tooling – needed to become at least somewhat real time, too. Suddenly, BI technology could display workflow issues in call centers, monitor point-of-sale data versus stock levels, or show as-it-happens transaction volumes on a geographical map. Now, our BI Gen II-enabled telco can track the acceptance rate of offers to customers in real time and narrow it down to specific customer segments, lines of business, channels or a combination of all three.Of course, the added capability to obtain business insight from real-time processes adds to the frustration of the business community, because spotting problems is not the same as fixing them. The telco now has real-time dials and gauges, but no levers. If a proposition has a lower-than-expected accept rate, it will require traditional business and IT processes to make sure it won’t be offered in all channels (or will be offered differently). The revised strategy is likely to take some time to implement, time in which money is lost. It’s time for Gen III BI.The next-generation BI essentially means that a report (or, more generally, insight) becomes actionable. For this to be a near real-time feature and to change a dashboard into a cockpit, our telco will require an underlying decisioning infrastructure based on analytics and rules. Let’s consider whether near real-time is actually worth having. After all, companies have survived with paper processes and email directives for a long time. What could possibly require an instant change of company direction? Despite the airplane analogy, a company, is not a fly-by-wire jet fighter – but should it be? If analysis at the speed of thought is considered a BI breakthrough, shouldn’t actions at the speed of thought be valuable too? What good is a real-time radar screen without real-time air defense? Consider the value of an instant change to a failing marketing campaign (as evidenced by a BI Gen II dashboard). And what’s the value of an immediate response to a competitor’s discount action, rebranding, or product launch? How costly is it to persist with a credit risk strategy that is getting tired in light of new market dynamics? Or consider a Web site that isn’t generating the expected number of clickthroughs? Or an agent (or team of agents) who can’t handle certain types of products, certain types of customers or the combination? It seems that the fly-by-wire principle, by which a plane can swiftly be controlled through a joystick despite the obvious complexity of flying, has its commercial benefits.To become a fly-by-wire company, many business decisions need to be automated. It’s outside the scope of this article to go into decision management, but very powerful decisioning solutions are available that can virtualize a company. Where it’s now common to see business processes defined in an agile way, not implicit or static like they used to be, it’s equally possible to express the intelligence (i.e., what a process should do) in an agile, business owned manner. Forward looking companies can modify decisions on the fly to adapt to circumstances or outmaneuver the competition. With business virtualization in place and with decision strategies that can safely be owned, monitored (using a BI Gen II solution), and changed by business, a company can now turn insight into action. This makes BI Gen III slightly less general-purpose (as you can’t monitor and control everything) but very, very powerful. The result is not only the BI I ability to slice and dice data to check out business performance from the highest consolidated level to the most detailed level of individual decisions (BI Gen I) and not just real-time access to this data as BI Gen II provides. Instead, it adds levers and controls to dials and gauges.Still, two more levels remain to BI, and the associated ROI is far too significant to stop here. The fourth level is already available in the market, the fifth soon will be. BI, as of generation IV, leverages the virtualization of the business to a higher level. Beyond controlling the enterprise, the abstraction of business insight into explicit, automated decisions allows a company to start simulating strategies before they go in production or consider modifications of strategies that already are. To remain airborne, this capability could be compared to a flight simulator. If BI at this level already captures all data and all decisions, how difficult can it be to run hypothetical strategies against real data and consider the effects using first or second generation solutions as if the results were real? The answer is: very difficult, but the more important answer is that this technology is now available. With this, the hypothetical telco can try out different, extensive what-if scenarios such as reducing the number of call center agents by 10 percent, making a proposition more often (cannibalizing others), changing an upgrade policy, doubling the retention budget or increasing the price of data services. So, finally, we reach the highest BI maturity level: Gen V. This is where the flight simulator becomes the autopilot. At this level of BI, using well-understood artificial intelligence techniques as well as more traditional statistical analyses will actively optimize business parameters. Arguably, level V could be restricted to the optimization itself, in an offline manner, and level VI would then take that to real time. The distinction is still a bit academic. In both cases, BI will calibrate business parameters for optimal results.The innovative telco in our example will now not only be able to see that a product is not selling because it’s too expensive (BI Gen I); although it’s useful, they also won’t stop at discovering this issue as soon as it happens (BI Gen II); it won’t even be sufficient for them to be able to reduce the price or boost some product features, and instantly apply that change in all channels that sell the product (BI Gen III); it’s not even enough for them to first try out those changes in a real life what-if simulation (BI Gen IV). No, the telco will use BI Gen V technology to actually find the right price for them, in the proper context of all their other strategies. This is not science fiction; it’s not even a revolution. It’s just an evolution of BI technology and a small step from BI IV capabilities that are already available on the market. Our BI Gen V-enabled telco will be a reality, and will be making very good money because of it.

For reprint and licensing requests for this article, click here.