Breakthrough Analytics: turning data into predictions

We know that analytics is becoming increasingly important because we are told that this is the case every week. Many of the most visible topics in IT — cloud, social media, mobility, and Big Data — derive some or all of their business relevance from tie-ins to effective analytics. But how is the analytics link connecting IT trends with business value forged? In a presentation in Kitchener, Ontario this February entitled “Breakthrough analytics: hot trends that remove barriers and accelerate business goals,” Dan Grady, social analytics and enterprise search sales manager with leading BI supplier Information Builders, identified the means by which analytics enables users to extract value from mobile, social, remote and Big Data.

Grady began his presentation by introducing the “three I’s” of these data sources: Integration, (data) Integrity, and (business) Intelligence. It’s likely that many of those in the room understood, at least to some extent, the importance of integration, or the need to connect diverse data sources to create a foundation for advanced analysis. As Grady explained “nobody…has all the data they need for analytics in one consolidated data source. It doesn’t exist” — and cloud and cloud-resident data only increases this challenge. It is also likely that many in the room understood that intelligence is the key outcome of effective BI, and appreciated Grady’s insistence that the tools that extract intelligence from data should not be limited to specialists (such as those working on developing the open source R language, and corporate users of similar tools from vendors like SAS), but built into applications that support different types of business users across the organization.

The third “I”, though — data integrity — is a topic that is rarely acknowledged in the rush to discuss the potential uses of the ever-higher mountains of data that surround us. To help position its importance, Grady introduced Gartner research on the annual business cost of data quality issues: a Gartner survey found that over 30% of respondents reported that data quality issues cost their firms more than $1 million annually (with 5% reporting annual data quality-related losses of $25 million or more). Grady focused on the largest respondent group — the 38% who stated that they “don’t know” what their annual data quality cost is — asking, “is ‘don’t know’ an acceptable answer for you?” As he pointed out, this issue becomes more critical in respect to data that is used to produce public reports, and where (as is the case with nearly 50% of Information Builders clients) the data is available “outside the firewall” to customers.

Through the course of the presentation, Grady covered the analytics issues connected with the ‘hot trends’ of social media, Big Data, cloud and mobility. He began the social media section by identifying two important questions: “number one, how effectively are you listening to the voice of your customer? That’s what social listening is. But more importantly, he asked how is what they’re saying impacting your business? Grady stated (supported by research from the Altimeter Group) that the biggest gap in effective business use of social analytics has been this inability to tie social media to business outcomes, and introduced a model that he used to divide the business maturity of social into three broad groups: engagement, listening level and business metrics. The lowest level, engagement, is tracked by counting posts, likes and similar metrics through public platforms, and is not typically used as the basis of business systems. The second ‘listening’ level, which includes metrics like sentiment, share of voice and  influence, and as a method of supporting customer responses, provides quantification of public attitudes stemming from announcements, new products, ad campaigns, or similar stimuli, can be applied to understanding internal and/or competitive positioning.

Source: Information Builders. Click on image to enlarge
Source: Information Builders. Click on image to enlarge

 

 

The third level, ‘business metrics,’ extends these data collection and analysis tools deeper into operational systems. Here, organizations are using social data to manage revenue-related issues like reputation and customer satisfaction. This seems like an esoteric future possibility, but Grady used examples from companies in industries as diverse as fast food (Wendy’s) and travel/leisure (Carnival) to show how these systems are used today to effect changes in business activities.

The session’s cloud discussion took a similar approach, applying BI principles to real-world challenges. The highlight of this session was a discussion of Salesforce.com — and of the analytics challenges that Grady claimed are experienced by 86% of Salesforce.com users. To illustrate the issue, Grady talked to the elements needed to support a “campaign to cash” analysis. Some of the data needed for this analysis — campaign data, the resulting leads, information on the opportunity funnel — resides in Salesforce. Other data needed for the analysis, though — including the order (often booked in an internal order entry system), the invoice (in the billing system) and the revenue (logged in the financial system) does not typically reside in Salesforce. Other related data, including commissions, expenses, call centre data and other information, may also enrich the application, and is also likely outside the Salesforce application. Cloud, according to Grady, creates “hundreds of integration challenges.”

The centrepiece of the session was a discussion of predictive analytics, which has been rated (according to research presented in the session) the BI technology of most interest to CIOs. Grady introduced a graphic positioning the potential value of different BI use cases on two dimensions: the degree of intelligence embedded in an application on one axis, and the competitive intelligence delivered by the application on the other.

breakthrough analytics graphic
Source: Information Builders. Click on image to enlarge

 

The slide demonstrates that the most basic use of BI is reporting — standard reports (answering the question ‘what happened?’), ad-hoc reports (‘how many, how often, where?’), and query drilldown (‘where exactly is the problem?’). The next level up — the one which dominates much of today’s discussion of BI use — involves alerts (‘what actions are needed?’) and statistical analysis (‘why is this happening?’). Forecasts (‘what if these trends continue?’) are often viewed as the ultimate destination of BI, but Grady supplied two additional steps that he believes should be part of the BI planning horizon: predictive modelling (‘what will happen next?’) and optimization (‘what’s the best that can happen?’). Revisiting a key theme of his presentation, Grady repeated his belief that this advanced capability isn’t, and should not be, the preserve of data scientists, but rather, should be an attribute of systems used to support actions across the organization. “That’s the trick,” he said at the event, “embedding predictive analytics inside of operational applications. Historically, predictive analytics has been the purview of the back office. You get a bunch of really smart guys, they build some crazy models, they print out a spreadsheet, they bring it to a boardroom, and that’s where the predictive analytics value dies. If you want to maximize the value of predictive analytics, embed the results inside of operational applications.” These applications, Grady believes, can be used in many different contexts — in call centres and other traditional niches, but also in areas ranging from child welfare to fraud detection to crime rate reduction, and to areas as wide-ranging as reducing attrition and supporting hospital administration.

InsightaaS summary: The “Breakthrough Analytics” event concluded with a slide entitled “Why Information Builders?” explaining why the company is a preferred supplier of analytics systems. At an abstract level, though, the key issues highlighted on the slide — Ease of Use & Flexibility, Scalability & Reuse, Data Access, and Low Cost of Ownership” — can be fairly understood to be success criteria for BI generally. Grady is correct in his belief that moving BI beyond the Yeti-like data scientists and into the hands of users across an organization is important to realizing business benefit from BI; as he says on the slide, “wide adoption is the greatest measure of [analytics] success.” The ability to reuse both data and code flexibly, and to scale applications to accommodate ever-larger data sources, will be a critical attribute for organizations looking to extend BI into different areas of their business operations. Seamless data access (including access from hard-to-serve — at least from a BI perspective — mobile devices) will dictate, to an important degree, the breadth of utility of BI-based systems. And low cost of ownership is critical: if BI systems require multi-million dollar platforms run by very high cost specialists, BI will not attain the ubiquity that was predicted in the session, and is predicted as well in the stream of reports and articles that testify to broad interest in the subject. In short, while we believe that there are still many chapters to be written in the BI story, we expect these themes will be prominent as we scroll across their pages.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.