How to Put the Intelligence Back in BI with Analytics
The idea behind Business Intelligence (BI) was to provide quick access to meaningful data so that executives could take action to cut a loss or capture an opportunity. The reality, however, saw data faithfully rendered but often without context or history which left executives still groping for meaning.
"BI has traditionally been about reporting while analytics has been about understanding," explains Melanie Murphy, director of Customer & Marketing Analytics at MindTree, ranked 19th amongst IT Services companies by IAOP in their annual list of the Top 100 Global Outsourcing Companies. "Therefore, we must first recognize the integration of business intelligence and analytics as a best practice in and of itself."
Unfortunately, the term analytics covers a lot of ground and not all of it is necessarily the terrain you need to map. "If you don't design your analytics correctly, it will tell you stuff you already know and may not be very helpful to your business," says Tripp Micou, CEO of Practical Computer Applications (PCA), a custom Internet database application consulting firm. "For example, 'it would be cheaper to ship my goods to my customers if they were closer to me' — so what, unless you plan to build out a new location?!"
Look for Meaning, Not Just Metrics
Summarized or silo'd analytics provide a number of metrics, but no real diagnostics as they may show a problem but provide no insight into the root cause. "It's like a doctor measuring a patient's heart rate at 150 beats per minute without noticing that she is running full-speed on a treadmill," says Dr. Christopher Houck, senior vice president of Product Marketing at OpenConnect, a leading provider of process intelligence and analytics enabled solutions.
"In a contact center, the number of customers served or phone calls dispatched are interesting metrics but, in reality, give no information on the effectiveness of the service operation in providing customer value," he said. "If quality and customer satisfaction are not measured, customer value and profits go down while efficiency metrics go up."
But before you can get to the validity and usefulness of analytics, and prioritize the analytics that deliver the biggest ROI, it's important to make sure the foundation of your efforts is sound. Make sure your data is scrubbed and current, or the analytics will always deliver faulty intelligence.
Useful Analytics for Marketing
It cannot be said loud or often enough that data is not the same as information. Knowing, for example, that Mrs. Marple buys a NFL team licensed shirt in a size extra large every fall may not mean that she's a diehard fan of that team or that she'll buy a shirt every fall for the several years. Knowing that her husband is a diehard fan who recently died, well that tells you to stop pitching NFL licensed goods to the grieving widow and present other offers instead.
According to Ian Fyfe, Chief Technology Evangelist at Pentaho, an open source business intelligence provider, the most useful BI analytics when it comes to marketing CRM are:
- customer segmentation - which customers have the highest propensity to purchase a particular product and which show signs of becoming non-customers?
- profitability analysis - which customers are most profitable over time; which ones are we losing money with and would we be better off without?
- personalization - using analytics to market to individual customers based on what we know about them.
When looking to implement BI in the context of CRM, he says, you should be looking for an offering with the ability to:
- easily embed or integrate BI as a seamless part of the CRM software application,
- perform operational BI directly against the live CRM application,
- extract data for offline analysis and tracking of trends and behaviors over time,
- use data mining and predictive analytic algorithms to enable effective customer segmentation and forecasting
Visualize Trends and Blind Spots
Not only must the data be converted into information but the presentation of that information should be delivered in such a way as to render a comprehensive view of the situation at a glance. Ideally, this is achieved via a data mining component, a data visualization component and with a solid multi-dimensional cube for analytical processing.
"The problem is getting these three components together at the same time," says Micou. "For example, Microsoft Analytics services does well for a solid multi-dimensional cube database. But, to get good data visualization, you need an additional BI product like TibCo SpotFire, Visokio, QlikView, Tableaux, Excel, or a custom built application."
"Unfortunately, most of the data visualization tools do not connect easily to the multi-dimensional cube databases, or data mining, which currently leaves a big void in the marketplace," he said. "That void has to be filled with either Excel-based data visualization, including the inherent problems of Excel, or with a custom-built OLAP (online analytical processing) application, or customization of these tools."
But technology is changing quickly and it's wise to stay attuned to developments.
"Tools are evolving to allow you to combine traditional reporting and visualizations with more analytical processes such as segmentation and predictive modeling, and they are becoming more intuitive and interactive," says Murphy.
Make no mistake, visualization isn't mere glitz for your Customer Relationship Management system; it's an indispensable tool that enables the human brain to derive meaning from otherwise overwhelming data dumps.
"As marketers struggle to comprehend and use the wealth of data that is now available across a variety of channels, practitioners need to be able to deliver reporting and visualizations with understanding and insights," explains Murphy. "Going forward, practitioners will need to be able to integrate multiple data sources, identify trends and patterns all while understanding how one channel affects another."