Facebook Group: 8065544817 FeedBurner: tagthink/WeWn TAGtv: http://tagtvonline.com Linked In Group: 41590 Twitter: TAGthink
Friday, August 01, 2014

Focused conversations within the Georgia technology community.

Text Size
Send
Print
Written by admin
Practical Tips to Overcome Barriers to Pervasive BI

Created 11/10/11
Author Name Sid Adelman | Dave Shrader
Author Company Teradata Corporation
Body of Topic

Table of Contents

Executive Summary
What is Pervasive BI?
What are the Barriers to Pervasive BI?
Tips for Overcoming Barriers
Successful and Unsuccessful Attempts to Adopt Pervasive BI
Best Practices – Tips for Getting Started
Summary 
Appendix A – Active Applications by Industry or Function
Appendix B – Resources and References
About the Authors

 

Executive Summary

Business runs at a much faster pace than in the past. Customers demand access to current information and faster customer service.
Society sets a higher bar for speed and content. Customers expect your company to know them and provide appropriate services and recommendations for products within minutes, not days. You need information sooner, and you need the information in greater depth – you need a 360-degree view of customer information from across your entire enterprise.

 

Many companies are reacting to these market needs by driving toward pervasive intelligence, augmenting traditional business intelligence (BI) with the ability to capture, interpret, and act on data immediately to make smarter, faster decisions at the front line. With pervasive intelligence, your enterprise data warehouse (EDW) changes into a sense-and-respond system that can proactively and reactively interact with the business eco-system, a world filled with customers, prospects, suppliers, and partners. It provides contextually appropriate decision options to help your organization build superior capabilities to respond and take action based on current integrated data, not on out-of-date data that show only one dimension of the customer, product, and supplier.

 

Many companies are reacting to these market needs by driving toward pervasive intelligence, augmenting traditional business intelligence (BI) with the ability to capture, interpret, and act on data immediately to make smarter, faster decisions at the front line. With pervasive intelligence, your enterprise data warehouse (EDW) changes into a sense-and-respond system that can proactively and reactively interact with the business eco-system, a world filled with customers, prospects, suppliers, and partners. It provides contextually appropriate decision options to help your organization build superior capabilities to respond and take action based on current integrated data, not on out-of-date data that show only one dimension of the customer, product, and supplier.    The result: an agile company that makes excellent operational business decisions and processes.

 

Three primary questions need to be answered if you are thinking
about adopting a pervasive BI approach:
1. What different actions could you take if you had current data?
2. If you had up-to-date data and insights, what kinds of measurable improvements would you see for the business?
3. Do you have the desire, power, political will, and awareness of the value to create the impetus for a pervasive BI project?

 

This paper provides answers to these questions, and also catalogs impediments to driving pervasive BI. It begins by describing why pervasive BI is important and relevant to every industry. The next sections cover many of the hurdles of adopting a pervasive BI strategy, separated into the categories of “people,” “process,” and “technology.” Another section deals with return on investment (ROI), because each major project must be cost-justified. The paper concludes with:

•Case studies of successes as well as failures.

• Best practices.

• Appendix A – Active Applications by Industry or Function – Be sure to look beyond your industry for ideas in others.

• Appendix B – Resources and References.

 

Key Takeaways:
• A variety of people, process, and technology issues can get in the way of unleashing value to front-line people and systems.
Recognizing these is the first step to resolving them.
• Some of these impediments are the same ones that get in the way of building an enterprise-wide database (such as inertia or weak leadership), but some are novel (such as smart backoffice people who may not know the front-line groups that could benefit from contextually relevant product offers).
• There is a variety of ways successful companies have overcome the obstacles.

 

What is Pervasive BI?


Pervasive BI (often also called operational BI) is the capture and use of information in near real time (sometimes called right time) to make smarter, faster, operational decisions. Often the number of front-line users accessing the data warehouse will be an order of magnitude larger than the number of traditional BI users. Pervasive BI requires intra-day data loading from front-end or external systems. The frequency depends on business requirements,
meaning the data must be current for better decision making, but not more current than required. Decisions must be made about the timeliness requirements for different data sources and applications, but pervasive BI applications typically use intra-day feeds – four times each day, hourly, or updated within minutes. Operational BI applications leverage historical context and blend this with fresh data, which requires that data from front- and backoffice systems be integrated, providing a more complete and
comprehensive end-to-end view of the customer, supplier, product, or business process to people in operations, such as contact centers. Data warehouses supporting pervasive BI – as opposed to the more expensive approach using operational data stores (ODS) – must provide workload management with priority scheduling so operational accesses with demanding service level agreements (SLAs) get the highest priority, as opposed to long-running BI queries. Finally, operational applications that are mission critical may require a high availability architecture and platform.

 

Why is Pervasive BI Important?
The value of operational BI can be huge. For starters, an organization that implements pervasive BI can make better operational decisions. For example, a customer service representative (CSR) may see an automatic pop-up for the next-best-offer for an inbound caller. Interactive Voice Response (IVR) and Automatic Call Distributor (ACD) systems can be driven to dynamically construct the right menu options for each caller and speed them to exactly the right service agent. Websites can be personalized
in the afternoon with relevant offers that reflect this morning’s transactions by each customer. Behind the scenes, front-line applications can also use the data to construct events that are pushed inside or outside the enterprise to customers or suppliers. For example, information on out-of-stock conditions could be directed to the appropriate suppliers, and impending deliveries could be accompanied with short message service (SMS) messages 15 minutes before arrival to the customer who is awaiting the delivery.

 

Pervasive BI can be a key factor in an organization’s success or failure. If your competition knows your customers better than you do, or understands the dynamic value chain and suppliers better, they will be able to negotiate for better terms and conditions from suppliers, and be more agile in customer interactions than your company. With pervasive BI, your parts, inventory, employees, and other assets will perform more effectively. Your ability to respond to this changing world will provide a more nimble information base upon which you will be able to lower your costs, increase
your revenue, and deliver an improved bottom line.


Pervasive BI enables an organization to make smarter, faster decisions, which can have dramatic impact on fiscal results.
For example:
• A five percent increase in customer satisfaction may drive one percent revenue growth because of referrals and reputation improvements.
• A five percent improvement in cross-selling activities at frontline touch points may translate into one percent revenue growth simply by recommending the right products at the right time.
• A five percent improvement in process efficiencies gained by analyzing and fixing process glitches in real time may translate into a two percent reduction in operation costs.


Yet, timing of initiatives is important. Pervasive BI adoption rates vary by industry. It may not be right or even possible for your organization right now, and it’s better to recognize this before making the effort. While near real-time insights on operational results can be valuable in identifying problems and opportunities, they will also lead to a level of transparency and change which may be uncomfortable for managers or employees.

 

What are the Typical Uses of Pervasive BI?
Customer-facing pervasive BI applications include marketing, sales, and customer service applications that deliver dynamic insights, such as relationship pricing, propensity to buy, probability to churn, customer profitability, next best activities and offers, and risk for fraud. Additional applications can make a significant difference in other areas of the business, including logistics, supply chain, and forecasting, by using current as opposed to day-old information. Finance applications include cross-organizational financial views, dynamic risk management of corporate assets such as mark-to-market value in banking, and a move to a metrics-oriented culture and a culture of accountability. Pervasive use of dashboards and scoreboards provides visibility into budgeted commitments and how much is spent to date, as well as real-time indicators of business performance. Appendix A lists pervasive BI applications by industry.


What are the Barriers to Pervasive BI?


With all these advantages, why don’t more companies adopt pervasive BI? There are many impediments. Some of these are legitimate based on an environment clogged with legacy applications and processes, and others are excuses with inertia stifling new projects. Many of the problems can be overcome by those with the power and the will to make it happen.

 

The impediments we’ve studied can be put in the categories of people, process, and technology. As you review these impediments, check the ones that seem appropriate for your organization. After you’re done reading this section, you can follow each impediment’s action recommendations to the corresponding heading in the following section in these areas: Leadership, Education, ROI, Team, Leadership, Governance, Technical Team, Cases, Steering Committee, and Applications by Industry.

 

Inventory the Excuses


People
 “There is no business sponsor for pervasive BI.”
While IT/database administrators (DBAs) may understand the capabilities and opportunities, the weak connections between IT and the business may make the business reluctant or unaware of pervasive BI opportunities. The key is finding the business sponsor with the pain or the opportunity. See Leadership and Education.


 “There is no support from IT.”
This is the flip side of the previous impediment. The business may understand the value, but some in IT may feel that their specific power base in the organization could be threatened, which may prevent IT from endorsing and planning pervasive
BI applications. See Leadership and Education.


 “The IT people don’t even know the front-line user groups and their systems because they’ve never had a contact center/website touch the data warehouse directly. The people running the contact center are unaware of what’s possible.”
There are many opportunities to raise the awareness of IT to the possibilities of using an EDW for customer contact data and to raise the awareness of the contact center or Web managers of benefits and possibilities. See Education, Team, and Leadership.


“The business doesn’t believe it can react quickly even with the real-time information, so there’s no advantage to having
pervasive BI capabilities.”
This is a legitimate concern. The business needs to be able to undertake small steps showing it how it can react quickly, and how this can demonstrate success with a positive ROI.
See Case Studies, Education, and ROI.

 

 “The internal wrangling within IT could keep any significant
project from going forward.”
Internecine disputes will never allow for agreement about technical standards for the company, including architecture and technological improvements required for pervasive BI.
Just as it’s difficult to plow through the barriers when standardizing on BI tools, you may need to provide stronger guidance when moving to Service Oriented Architectures (SOA). See Governance and IT Leadership.

 

 “Data warehousing is, by definition, the capability to perform
tactical and strategic BI analytics, not operational analytics,
so it should not be used in any way that resembles an operational
database.”
There are many old hands in IT who would find the operational use of their data warehouse to be repugnant or scary. Education, including examples of case studies in their own industry and exposure to what industry leaders are saying, should help turn around these people. See Education and Case Studies.

 

 “The new CIO had to be convinced about the benefits of a
pervasive BI approach based on our EDW, and she put our
project on hold.”
Use data from the EDW to help these groups build a first project.
Use that as the lever for getting more up-to-date information in the database. Use that project to document the value of smarter, faster decisions, then publicize the results, and build a strong game plan for the next project – and the next one. See Leadership and Education.

 

 “Our company has a mix of new and long-time employees.
The long-time employees are risk-averse, while the newcomers
are too aggressive.”
This is a difficult minefield, but co-opting the naysayers by identifying some benefits for them might bring them over from the dark side. It does seem like newer employees and the early tech adopters understand the power of pervasive BI and want to make it happen much more than the long-time employees.
Younger people have grown up in a non-batch world, so they understand the idea of instant data availability and action. See ROI and Leadership.

 

 “We are hamstrung since only anointed people in IT can
talk to the business. This not only slows us down, it leads to
miscommunication and misunderstanding of what the
business needs and hampers feedback to us about changes
and corrections that are needed.”
This problem is usually solved by a powerful sponsor who makes clear the need for open communication between the beneficiaries of operational BI systems, and intolerance for anything that might cause the project to slip or be over budget. See Team.

 

 “Our company is too big. We have a terrible time communicating,
and it’s especially difficult communicating with people in other cities and other countries.”
It’s difficult to integrate your company using up-to-date data and align the interests of business and IT when resources are remote. Companies that are “too big” have a difficult time when the Web people are in one building, the contact center in another, and IT in yet another – even on the same campus. It’s hard to marshal all the right people at the right time and get them all pointed in the same direction, but it can be done. See Leadership.

 

 “Our front-line and back-end groups are usually organizationally
separated, so there are no natural forums for them to interact and talk about using pervasive BI techniques.”
The bigger the company, the harder this problem becomes because, even if geographically co-located, people have layers of management and organizational roadblocks in the way of their finding each other and launching a mutually beneficial project. The litmus test – if you are in the BI group: Can you name the owner of your contact center? Smaller companies
have the advantage. See Team.

 

 “There are managers who feel the need to make adjustments to the inputs (“spinning the numbers”) so they look good.
These managers will want the time to either justify why their
department didn’t do very well or will want the opportunity to actually make changes in the reports. Pervasive BI will not give them the time nor the opportunity to “qualify” their results. They will resist pervasive BI because their numbers will be immediately and widely available.”
The involvement of the chief financial officer (CFO) or an independent group can dispassionately help to provide the impetus for everyone to manage by the numbers. See Team and books by Davenport in the Reference section.


Process

 “It’s too difficult for us to quantify ROI for an operational
BI project.”
ROI has been difficult to calculate for some data warehouse applications, but most pervasive BI applications are easy to justify and can demonstrate a positive ROI. See ROI.


 “In our ROI and cost justification, the payback has to come
this year.”
In an environment that focuses only on short-term results, delivering payback in the current year is probably not achievable. The potential for benefits in the long term should be explored even with this short-term requirement. It’s time to have a discussion with the CFO about this unrealistic costjustification requirement. The first pervasive BI project may
take a bit longer, but once you’ve done the work for the first one (for example, moving to intra-day loads, building a message bus for active integration using Web services), then that infrastructure is all “free” for the follow-on projects. See ROI.

 

 “Our budget has serious constraints. It was cut again this
year by 10 percent.”
Waiting for the budget hammer to descend and not being proactive, showing the monetary benefits of pervasive BI, will mean that a first project will never happen. See ROI and Case Studies.


 “We use a system integrator to do our implementations,
and they haven’t been educated, nor do they believe in this
pervasive BI idea. Anything that is real time in their view
will automatically require an operational data store.”
This common problem can be overcome with having your Steering Committee place pervasive BI applications high on the project list, and by providing education for the systems integrator by the IT group or your tech vendors. See Governance and Education.

 

 “The data quality has to be perfect.”
Data will never be perfect, so the question is, what’s good enough? The data should be at a level of quality that is sufficient to be successful and to allow for appropriate actions.
The timeliness of the data and the level of integration are the data-quality dimensions specific to pervasive applications. See Steering Committee.

 

 “There is no need for pervasive BI applications in our industry.”
Some industries are slower than others in adopting operational BI, but every industry has dozens of applications waiting to be discovered. It’s possible that those making the decisions are  unaware of what this technology advance holds in store for them. See Applications by Industry, Appendix A.

 

 “We got along without it so far. Why do we need it now?”
This is the classic case of inertia, usually espoused by people who want to avoid any additional work. See Leadership.

 

 “Our organization likes to make very big changes in our
operational processes versus small incremental changes.”
Learn as you go is a much better approach. See Steering Committee.

 


 “IT delivered some pervasive BI information to the front-line
groups, but the business is too overloaded with existing
problems to use it.”
Any business that is unable to exploit these capabilities probably needs training and additional learning-curve time to integrate new functionality. See Leadership and ROI.

 


 “We receive so many reports; it’s difficult to know what’s
important and what information is relevant and should
cause us to take appropriate actions.”
It’s time to move away from voluminous reports and shift to exception notification whereby the ADW “notices” problems and “notifies” management in real time via alerts or changing dashboard items to flash or turn red in real time (for example, to adjust staffing and other resource allocation, and to address emergencies). See books by Davenport in the Reference section.

 


 “We’ve heard of some other companies that have failed trying
to use pervasive BI technologies. We can’t afford a similar
failure either to our reputations or to our bottom line.”
For the people who are risk averse, pervasive BI applications can be frightening. These people will have to be encouraged and incented to move to a more modern IT environment. See Leadership.

 

 “Too many of our rules are embedded in old processes
(such as fraud detection). Isn’t it too hard to rework fragile
existing processes?”
In many cases, rules should be teased out of applications so they can be used across a variety of channels to detect fraud in near real time vs. after the fact.

 

 “Decision makers prefer to build their own solutions rather than looking at what’s already available to them, like the EDW. Sometimes it’s important to do a new project fast with another data store than do the work to use the EDW.”
A steering committee and the system architects would be able to enforce standards and direct a solution that would include the use of an existing EDW. See Steering Committee, System Architecture Group, and ROI.

 


Technology
 “There’s a lack of existing integration of data. We acquired
three companies in the past seven years, and they each continue
to run their original applications. We thought about consolidating into one system, but we never got around to it.
Since we can’t integrate our existing data, how can we possibly
integrate and modernize our applications using pervasive BI approaches?”
Let’s define our terms before going too far. What do we mean by
integration? It could mean a 360-degree view of the customer, having all data related to the customer – all the customer’s interactions, calls, transactions, analyzed text, shopping patterns, demographics, lifetime value, wallet share, service preferences, propensity to buy, risk of fraud, and churn risk. For legal, cost, privacy, security, and value reasons, we would never be able to or want to have this level of completeness or integration. The operative questions are: “How integrated does it need to be for pervasive BI applications to be valuable?” “What is the cost?” and “What is the value of each additional piece of data?” And so in this example, we could all agree that complete integration of customer data isn’t required for an initial implementation for customer insight. See Technical Team.

 

 “The infrastructure isn’t suitable to a pervasive BI implementation. Today’s workload already overloads the hardware, and any additional workload will mean our existing SLAs
will not be met.”
The hardware and software platform may first have to be funded and then upgraded to support pervasive computing. It’s also possible that the infrastructure isn’t overloaded, but that it’s not properly tuned and configured. See Technical Team.

 

 “The applications are so old, non-standard, and undocumented
that it would be almost impossible to integrate them. Besides,
the people who developed them are gone, and no one knows
how the applications work.”
This isn’t uncommon, but it reflects a serious problem for moving ahead, not the least of which is porting these old applications to pervasive BI. See Technical Team.

 

 “Moving to operational BI applications directly connected to the EDW means that the application will be mission critical, implying the need for a high-availability infrastructure and the need for the additional hardware to support a high-availability SLA. There might even be a 24 hours/day, 7 days/week scheduling requirement that will be even more demanding on the architecture, infrastructure, and the skills to build and maintain it.”
Not all pervasive BI applications are mission critical, nor do they require an extremely high-availability SLA. In most cases, an implementation will deliver more timely information and capability than was previously available, so requiring 99.99 percent availability would be inappropriate and not cost justified. See Case Studies and ROI.

 


 “We don’t feel we can achieve integration. It’s too difficult,
expensive, and risky.”
Not knowing how to plan for and implement a pervasive application could keep the project from getting started. You may need to hire a leader with experience in pervasive BI.  See Leadership.

 

 “We have an ODS – actually, a few of them – and they’re giving us real-time information to support our business.
Why would we need a pervasive BI solution?”
A pervasive BI solution will provide more opportunities for integration and will also minimize the redundancies, inconsistencies, and the increased resource requirements and costs of an ODS approach, especially as the number of ODSs increases. See Case Studies, ROI, and Applications.

 

 “There’s a serious risk of putting operational data in a data
warehouse. The operational data have been our crown jewels
with demanding SLAs and attention when the system is down.
The data warehouse has usually been treated as a nice-tohave
system with more lenient or no SLAs at all. Putting operational data in the data warehouse may impact or jeopardize the availability of the operational systems and cause us to miss the SLAS that are demanded by these systems.”
The data warehouse is now being used for operational purposes. Let’s be clear: in this scenario, you are connecting the operational users to the data warehouse. This scenario may also require an environment allowing both hardware and software upgrades without a diminution of service level either
for response time or availability. The organization needs to provide a higher level of data stewardship and DBA attentiveness as operational data are stored in the data warehouse. But consider that the presence and cost of an unneeded operational data store can be removed. In this situation, a 
high-redundancy architecture might be an appropriate solution. See Technical Team.

 

This concludes our discussion about the most common reasons we’ve heard for why companies do not adopt pervasive BI.

 

 

Tips for Overcoming Barriers

 

You may have thought you were alone. You may have thought there were no solutions to the impediments you checked off in the previous section. But there are. What follows are some guidelines for moving ahead with the tips to provide answers to the impediments. They include a means of calculating ROI, suggestions about how to build effective teams, proposals for business and IT education, leadership approaches (including a steering committee
that will make all the difference between success and failure), and
recommended technical roles. The cases studies will give you realworld
experiences that might be close to your situation and should give you a basis for more research and discussion. The applications in Appendix A will give you ideas about what’s possible.

 

Leadership

Pervasive BI applications require powerful champions, change agents, and sometimes visionaries. These sponsors should come from business and from IT. On the IT side, the CIO or a senior IT manager should be a primary sponsor, since pervasive BI applications require IT resources. In the most successful cases, sponsorship comes from the line of business that recognizes the need for operational decision making. That support must reside high in the business organizational hierarchy – from the senior executives who understand the value – but also be supported by people in
the lower echelons who are more focused on their immediate and
circumscribed jobs and measurements. If pervasive BI applications aren’t seen as benefiting them, they will either have no interest or will try to thwart the whole program. The higher the level of the leadership, the more likely the program will succeed.

 

For example, a Chief Marketing Officer (CMO) might realize that to properly service the organization’s customers, the data about the customer must be integrated and current so that a CSR can appropriately respond to complaints, requests, and status questions. This might drive a project to put more right-time information on the contact center agent’s screen, which often simultaneously achieves three goals: First, for the customer, interactions are quicker, and problems are likely to be resolved on the first call without agent handoffs. This is likely to improve customer satisfaction. Second, having all relevant and up-to-date information on the screen means
that average handling times drop, so each agent can handle more calls per unit time. Third, “once and done” achievement on the first call also means the number of callbacks drops off, reducing overall call volumes. The manager of the contact center or the CMO (preferably both) might be the drivers of the revised screen effort to give the CSRs the information they need to do a superior job.

 

A model organization would be a Customer Experience team, which spans marketing, sales, and customer support groups, so that the focus and responsibility is under one person, preferably a senior VP. If you don’t have the power to change the organization (which you probably don’t), then try to get a cross-organizational Customer Focus team effort going that can pull together the correct, interested people to execute a pervasive BI project.

 

At other successful implementations, a savvy CFO can be a primary driver of promoting pervasive BI since the CFO focuses on the numbers and sees all the money going in and out. Finance is aware of all the lines of business, and they see the power of integrated data, the value of broader access to the data, and the need for those data to be current for the organization to make better operational decisions. Some CFOs have driven pervasive BI applications to close the books faster, halving the number of days to close the books and doing virtual closes at any point in the quarter.

 

Steering Committee
The top-level leaders need to drive a new or existing steering committee to focus on the pervasive BI applications and to focus on the importance, opportunities, and ROI that pervasive BI brings to the organization. This group makes critical decisions about the direction of the program and the project portfolio. The committee should be composed of the people who have a vested interest in the program’s results and success. The members are the managers who either fight for the program’s funding or are able
to fund it themselves. On the business side, members should be line-of-business (LOB) managers who own the data and the applications, and are thereby accountable for the results of the program. This may be a CRM manager, a contact center manager, a marketing manager, or a manager responsible for the organization’s inventory. On the IT side, it’s possibly the CIO, but more likely a senior IT manager, a CTO, or respected senior systems architect. The committee analyzes the costs and benefits of pervasive BI applications, evaluates and vets the ROI, and provides the resources (including budget) and, most important, the people who will be dedicated to the program. The steering committee should occasionally bring in outside speakers to suggest new opportunities and should make use of outside consultants to validate their work and decisions.

 

Driving a pervasive BI applications program isn’t a one-time effort; it’s ongoing with multiple sub-projects, one per specific application. The steering committee must be continually involved and committed – with a meeting every 1-2 weeks, sometimes with new members – as long as the organization will be making decisions about the operational use and reuse of near real-time data.

 

Business Intelligence Competency Center
The BI Competency Center’s (BICC) primary goal is to ensure wide reuse of data and insights. Its members look for the best way to get information into the hands of business users. They either report to the business or, alternatively, have a very close relationship with the business users, and understand the needs of the different levels of users – executives, report consumers, power users, and users of operational data. The BICC works to establish SLAs, understand timeliness needs, including the need for intraday
data, and reconcile and drive various delivery desires (dashboards, reports, alerts, and more on various device types – PCs, Blackberries, iPhones, or iPads). Its members also monitor and encourage propensity to use various applications. This group should be the keeper of the BI insights and key performance indicators (KPIs), keeping a watch on the factors that could be
dynamically changing, and communicating them and working with the LOB people to inject more real-time insights into the front-line IT systems, such as the Web, contact centers, and executive dashboards. The team would also be involved in monitoring the effectiveness of the pervasive BI applications and making appropriate modifications (and measuring the effect of those modifications); monitoring usage and performance, including
which departments were actively running queries and which ones were not; as well as keeping an eye on the effectiveness of user training.

 

Education
It’s important to recognize that the education of IT and business users is a critical success factor for a successful pervasive BI implementation. Business users need to be made aware of the possibilities for improving their day-to-day operations. The education should make them aware of pervasive BI key concepts, as well as leading practices within their industry. This education will also ground them in what their IT organization is capable of delivering. The education will focus on their industry and the functions within the organization, such as customer support or inventory control.

 

For IT, it means training on the tools and processes they will be using with an understanding of the tools’ capabilities and how they fit in the organization’s environment. A key part of education will be about the data, what data are available, what they mean, their source, how they were cleansed and transformed, the valid values, and how the data relate to other data elements. IT education about the business is often overlooked. The knowledge of the business gives the IT people a level of understanding that will allow them to make intelligent choices about the architecture, design, and where the effort and resources should be expended in building and supporting the pervasive BI system.

 

Project Team
There needs to be a cross-organization Project Team that links and
promotes lines of communication between and among these groups:
• IT Operations
• IT data warehouse
• Business people responsible for the applications
• Front-line business groups and users
• BICC

 

This linkage is a melding of the IT back office with the front office.
This needs to be a virtual team with very open lines of communication,
frequent meetings, and, most important, common goals and incentives. This virtual team will have business liaisons within IT or business people who act as liaisons to IT.

 

Note that there are at least two kinds of IT people represented on this team – the database-centric people and the business-side IT support group that runs the website, contact center, POS system, or ATM network. Those two IT groups need to work together (including any middleware IT people if you are taking a messagebus approach). Also, the IT business modeling people familiar with the data warehouse need to speak with the business-processcentric people to augment data models to cover attributes and
relationships that may not be in the EDW data model.

 

Everyone on the team should have a reasonable understanding of the business and the problem or opportunity the pervasive BI applications will address. IT needs to learn the business terminology, be very familiar with the business data, and have a good understanding of the needs of the business and how it runs. Some organizations require new IT staff to attend orientation classes designed for employees new to the business. The business needs to understand the needs and requirements of IT and some basic technology constructs. The challenge is to create a cohesive project
team that works well together and has common and well-defined goals. Performance plans, incentives, and bonuses based on the achievement of the common goals will help assure a productive, effective team.

 

Technical Team
Data Architecture Group
The data architecture group needs a complete understanding of what the business needs, the SLAs, and they need knowledge of the data in question. The data architecture group is composed of the data architect(s), the data administrators/data modelers, and the database administrators. The data administrators/data modelers are usually the ones responsible for metadata and sometimes responsible for data quality. Since pervasive BI applications will place the data warehouse in the operational world, both data modelers and DBAs need grounding in operational events. The DBAs need to create physical database designs that support operational transactions (including judicious use of indices) and that integrate well with front-line system data feeds and serviceoriented interactions. Often initial implementations can be built that don’t require changes to the front-line systems, but having an end-state architecture in place can clarify the roadmap to the ultimate technical goal.

 

Other responsibilities of this group include:
• The use or non-use of data staging.
• Workload management including controls, exception management, audits, change control, and security.
• Control over shadow databases and departmental databases.
• Capture and use of business metrics and technical metrics.

 

Applications Architecture Group
This group will need to work closely with LOB owners, as well as data and system architects to determine which existing applications should be rewritten to take advantage of current data and context. If operational data stores have been used, then call-outs to the ODS may need to be replaced by call-outs to the EDW. 
New applications may need to be written and placed in the hands of front-line groups. For example, sales agents can instantly use mobile phone apps to look up product availability or nearest locations with in-stock merchandise.

 

System Architecture Group
The System Architecture Group should drive use of an enterprisewide
approach to architecture across all applications, tools, middleware, and databases. In many companies, a key architecture decision is to drive everyone to use a service-oriented architecture (SOA). The enterprise architecture encompasses protocols and functionality to encourage widespread use and reuse of data models, processes, methodologies, tools, and code. The System Architecture Group makes the following decisions and determinations:

• How the data flow from the data sources, where the data are transformed, filtered, cleansed and integrated.
• The use or non-use of an ODS.

• If the Web is being used for access, how that access to the EDW would be implemented.
• For high-availability systems (ones that require an availability SLA of 99.99 percent), strategies for how to achieve end-to-end system availability using redundant components with fast failover and recovery techniques (end-to-end, not just for the data warehouse).
• Use of software – which tools will be used for which functions including ETL/ELT, security, real-time data feeds, data integration, data cleansing, metadata, as well as the DBMS hardware and software platform.

 

ETL Developers
The ETL developers should employ tools such as Oracle Golden-Gate, Informatica PowerCenter, or IBM Websphere that execute trickle feeds or run mini-batches, as well as any data cleansing software that will be invoked dynamically prior to the data being loaded in the data warehouse. The primary differentiator between pervasive ETL developers and conventional ETL developers is a break from the batch mentality, understanding the design and performance implications of trickle feed or the loading of mini
batches. The developers have to be very aware of the performance,
timeliness, and, response time requirements set by the business.

 


Return On Investment
Return On Investment (ROI) is used to cost-justify potential benefits and develop a business case for each project. The requirement usually comes from the CFO and the LOB manager and is often required for next year’s budget. The steering committee will always focus on ROI and use it to make decisions and establish project priorities. Actual ROI (vs. predicted) is also extremely useful as part of post-implementation project evaluation because
it can help bolster the case for the next pervasive BI project.

 

ROI involves costs vs. benefits:          Benefits (to the Business)

Costs (to IT and Business)

 

ROI can be computed through a variety of means – simple backof-
the-envelope calculations, or with more sophisticated measures,
including payback periods (or break-even analysis), net present value (which accounts for the dollar’s value dropping over time), and rate of return (or yield on investment).

 

It’s almost always the business that calculates the ROI with input from IT for projected costs. Benefits depend on the application area – for example, putting the “next best offer” on the contact center screen or website may drive a 15 percent improvement in cross selling. A benefit of using a dashboard to show process delays in fulfillment can highlight needless expense that can be taken out of expediting charges. Business people often use Excel spreadsheets to calculate before/after business benefits (the numerator of the equation). A good, readable overview of marketing ROI, including free spreadsheets, appears in Mark Jeffery’s new book Data-Driven Marketing: The 15 Metrics Everyone in Marketing Should Know.
Read this book, and give it to your business allies.

 

We focus here on the costs, the denominator in the equation. For many organizations, the costs involve a comparison of a quickand-dirty ODS-based approach versus a more disciplined EDW approach. A pervasive BI implementation using an EDW may be perceived as expensive but if you have already:
• purchased the hardware and software
• worked out the data models
• built good ETL processes with data quality controls in place
• bought BI tools

 

… then you probably are 90 percent of the way to pervasive BI, and can unleash new value by reusing what’s already in place. You may need to change from batch to mini-batch or trickle ETL feeds, but this can be done over a period of time. The first step may be the biggest (for example, loading three times per day instead of once overnight) but then your business users will start driving you to get hourly data, hourly updates to scoreboards, etc. Most companies are just trying to get to “yesterday’s data today,” so the state of the art is still behind where it could be. But companies that are already using yesterday’s data today are now poised to go to the next step.

 

For companies that do not yet have an EDW, the comparison of ODS vs. EDW is much like the calculations involved in using a data mart approach versus an EDW. Initially, first-project costs are lower than the more comprehensive EDW approach, but the more people who use and share the data and the more projects generating ODSs, the more costly an ODS solution becomes compared to an EDW over time. As Figure 1 shows, the cutover point for an EDW becoming less expensive can be as quick as the
fourth project.

 

For the front-line IT groups, another cost is modifying applications, for example the Web-rendering engine’s scripts or the contact center’s screen sequences. In many cases, injecting analytic insights into the application just adds more if-then-else statements into existing apps. For example, a business unit programmer might add code such as “if High-Value-Customer, then X else Y where X can be ‘give complimentary upgrade to Business class,’ and Y can be “apologize for non-availability of Economy seats, and recommend next flight with seats.”

 

Another cost will be adopting or adjusting policies for workload management to accommodate the SLA requirements of pervasive front-line applications. If the DBAs are already using workload management tools for various flavors of BI workloads, all that is required is adding new classes of operational workloads and giving those top priority.

 

 

The bottom line on costs: if you have an EDW, a big sunk cost investment has been made already. Moving to pervasive BI leverages past investments and provides new, timely data and insights to the front lines at a very modest cost. For organizations that already have a suitable platform in place with the required level of integration and data quality, a majority of the work might have already been accomplished and most cost already taken on. The
cost of additional pervasive BI applications is a decreasing function once you are over the hurdle of doing the first one.

 

Sometimes pervasive BI applications require data availability that is much more demanding than those in a typical data warehouse environment. This may require high-availability designs, models, architectures, and implementations, and so in these cases, it’s imperative that pervasive BI applications be associated with a cost justification and a well-defined ROI.

 

 

Successful and Unsuccessful Attempts to Adopt Pervasive BI

 

The following cases provide insights into what some organizations have accomplished with successful pervasive BI projects. Each case delivered a different benefit – such as improved customer loyalty, more efficient employee staffing, next-best-offer, cost reductions, and operational effectiveness – and all are based on real pervasive BI implementations.

 

Case 1: Retailer
The impetus for using pervasive BI at a large retailer stemmed from three sources: an IT leader who understood the business, the business group’s perspective that it’s easier to grow revenue from the installed base than to acquire new customers, and a CFO willing to provide sponsorship. The immediate motivation came from a new effort to add a Web channel for sales. The first instinct was to add one more data mart to support the Web channel, but then a variety of people, including a director on the Board,
realized that what they needed was a single repository of all customer information. As a consequence, they refocused on their loyalty program and used their data warehouse to put more information at the fingertips of employees and customers. They now support more than one million queries to the EDW coming from their website, their stores, and their contact centers each day, with a 0.10 second average response time. Customers on the  eb, store employees through the POS system, and contact center agents can instantly find any customer’s number of loyalty points, so the customer knows how far away she is from free merchandise.
The company uses trickle feed to provide near real-time access to their data. They link content with inventory for their contact centers and their Web channels. The EDW supported an extremely successful e-commerce initiative.

 

In the post-implementation review of the project, the project manager said, “This was an IT-led project. Our CIO was a visionary. The good news is that our steering committee and CFO understand the power we now have; we have a strong PMO office, and we are launching and completing projects 30 percent faster than before.”

 

This company also took several other process approaches that worked. For example, the company has a single point for project requests called the Customer Information Management (CIM) steering committee. It is led by the CFO, and all requests go to the BI/Apps or Customer Service/POS teams for sizing. All projects requiring more than US$100,000 in internal and external spending must have CIM and IT steering committee approval.

 

In response to the question “Why should the CFO drive this effort?”, the answers are:
• “The customer data is the most strategic corporate asset.”
• “It gets down to who owns the customer.”
• “Because it’s quite common for the CFO organization to have to balance requests vs. resources for hard assets, we decided to do it the same for customer-related projects.”

 

Case 2: Bank
A large bank needed to create consistent, personalized banking offers for its customers at all its touchpoints, which include its contact center, website, and face-to-face teller and banker point of interactions. This required a single view of the customer. At any time, marketing and sales have 50 banking offer possibilities available to sell to clients. Predictive propensity models were built to generate the next best offer for each customer,
independent of what channel(s) the customer uses. Contact center screens were updated with portlets that show the CSR all recent banking activities, the next best offer(s), as well as the likelihood to buy. This last indicator was used to let the agent know if he or she should spend extra time selling the bank’s services. The next best offer is also available to the website, teller,
and banker screens. At interaction time, the front-line apps make a call-out to the EDW to see which offer to present based on a good understanding of each customer and the customer’s most recent activities with the bank.

 

The result was a 122 percent increase in sales to high-value customers making inbound calls. It also resulted in a 24 percent decrease in the average handling time for low-value customers or Successful and Unsuccessful Attempts to Adopt Pervasive BI customers with low potential for sales with no decrease in customer satisfaction. In addition, outbound sales calls showed a 15 percent increase in conversion rates. Customized Web advertising using the same recommendation engine achieved five to 27 times better results than those achieved by undifferentiated Web offers.

 

Case 3: Retailer
The impetus for operational BI at this well-known company was “managing locally by the metrics,” with expansion of access to information down to the store level. This company has a culture of actionable operational metrics, looking to the numbers to drive local store management to take specific steps to improve its store’s business performance, as activity in the store is happening “right now.” The first instinct was to add an ODS. The goal was to
build an application to dynamically shift staff to high traffic areas with the most demand and sales opportunity, and support a “floating staff” model, all based on POS data captured in real time. They soon realized that much of the information needed to do this application was already being loaded in near real time to their EDW. In addition, the EDW approach allows the company to identify the departments within the store that are and aren’t
meeting specific goals such as cross-selling services for their products in near real time. Finally, the more global approach allows management to compare stores in their relative effectiveness in shifting staff in real time.

 

Case 4: Media
This all-digital TV service provider has the requirement to provide near real-time operational information to their contact center agents, as well as to provide current and quality data for their field services to improve the responsiveness of the field technicians who provide installations and field support. Near real time means they need to perform analysis and decision
support on data with a latency of less than 15 minutes. Data volumes are substantial, handling 150 to 200 million records each day. The data warehouse supports more than 1,500 users generating about 8,000 requests per day.

 

The benefits have been substantial, allowing the company to manage churn (customers leaving or attempting to leave) in real time. The CSRs can manage service calls using intraday and upto-the-minute information. When a customer calls to request discontinuation of services, the contact center takes the order, but that kicks off an analysis of whether or not the company wants to try to turn the customer around. Sixty percent of the time, this is the case, so the system queues up an outbound call back to
the customer within two or three hours with an inducement to maintain service. The inducement is based on customer predicted value, as well as past service orders, like pay-per-view. Twenty-five percent of the time, the customer takes the offer and continues their service, resulting in the lowest churn rate in the industry.

 

Case 5: Railroad
This large railroad company runs “by the numbers,” and their employees are held to their performance metrics. It’s a wellmanaged company that values information and makes it widely available throughout the organization. As a logistics company, tracking boxcars, monitoring yard operations, and managing arrivals of raw materials for factories are all core uses of information. The EDW “runs the business” and so has tight SLAs. A
primary application of the EDW was empowering customers with self-service ability to track and analyze the movement of their freight and their railroad cars, which was a serious differentiator from their competition. A side effect of more customer self service was a subsequent significant reduction in contact center staff and the savings associated with that reduction. Major reasons for the company’s success are the level of sophistication of their DBAs – their motivation, initiative, knowledge of the railroad business, the responsibility and pride they take for their work product, as well as the technical sophistication of the business people in marketing, finance, and operations. Another key ingredient of success was IT personnel’s knowledge of how the corporate business strategy translates to the jobs they perform as many of them came from the operations side of the business.

 

In addition to studying successes, it’s often very instructive, albeit
rare, to look at failures.

 

Case 6: Telecom
This company in the telecomm space has stalled in its attempts to
adopt pervasive BI. The DBA and IT people studied and learned
to appreciate the pervasive BI approach two years ago. As a result,
they believe they understand what they need to do. However, the
business liaison people assigned to help groups use the data
warehouse and drive new projects (and uses of data) haven’t
bothered to learn much about pervasive BI applications. The
business liaisons are all BI-tool-smart people who are proficient in
gathering and conveying requirements for traditional BI projects
(for example, they recently completed a Customer Lifetime Value
project). But they don’t have connections with the front-line
groups and haven’t sought out any projects to apply Lifetime
Value beyond the direct old-style marketing function. They have
not scheduled any meetings or workshops with the LOB people
(either director/VP levels or the managers of the Customer Care
Centers or Web groups) to educate them about better application
possibilities. Their work queue is filled with more BI projects.

 

Worse, these people currently perceive the IT group as a threat;
in their words, “IT is trying to push pervasive BI” and “IT is going
around us to the business.” Without a business need and sponsor,
no project will happen at this company, and there will be no
funding for the IT people to do the ground work to adopt pervasive
BI as a strategy. With no visionary leader who can break the
inertia to do anything different, it’s likely that this group will
continue in this state and will fall behind their more aggressive
competitors and squander the opportunity to use real-time
data. Our prognosis, sadly: Calls to their care center will still be
routed as usual; direct marketing interactions will continue to be
mostly irrelevant; customer interactions and the website will
remain “one size fits all;” and this company will continue to be
susceptible to customer poaching by their more agile competitors.

 

Case 7: Retailer
Another failing case example involves a major retailer. Their business groups, in particular the Web team, understand the value of pervasive BI for the website. They would like to, and often do, put up personalized offers based on the previous browsing behaviors. They analyze dropped market baskets after each session. When a customer returns, they put offers for products placed in carts but not purchased on the home page. But they
also know that this approach isn’t working out because sometimes the customers browse on the Web, but buy in the store – and this lack of an integrated up-to-date view causes them to field complaints. For example, when they put a product on the Web on sale, a customer may call to complain that she just bought it at a store for a higher price. Another major problem is reverse logistics – someone may buy a product on the Web, but want to return it to the store. Not only are there accounting problems for the stores and inventory problems when this happens because they don’t have integrated, up-to-date data, this also creates return issues with upstream vendors and the distribution center.
So there’s a clear need for a 360-degree view of the customer and products, and a holistic, up-to-date view of the business processes and accounting.

 

The problem stemmed from a decision when the Web group was formed to organize it as though it was another store, not a channel, with its own profit/loss statement. Coordination with corporate marketing was weak, and because the marketing/merchandising groups were still doing independent
product-line-push (vs. a customer-centric approach across multiple channels), the Web group felt like it couldn’t make much progress on resolving its process problems with physical stores, even though they saw the need and value to customers.

 

Best Practices – Tips for Getting Started

 

Put a Plan in Place
1. Educate yourself. All change happens with one individual. The more you can learn about pervasive BI, the better armed you are to educate others. Order and read the leading-edge thinkers’ books referenced in Appendix B. Download and read the best articles. Seek out best practice cases that remedy some of your company’s pain points, and avoid practices that would
provide opportunities to your competitors. Focus on areas where your company has had issues – for example, loss of market share because of bad customer service, or supply chain problems because of lack of visibility in the value chain. Get to know the corporate strategy team and the company strategy, and then find areas where pervasive applications could make a difference.


2. Find a personal sponsor or mentor. Someone who has more power than you can help you and provide guidance. Educate this person. Have lunch. Talk about the opportunities, and try to establish a high priority for the pervasive program.

 

3. Educate others about pervasive applications. Build a list of areas where other groups should have interest. A sample exercise is to take your organizational chart and highlight which groups have the need. Then talk to these groups to also rate them on interest. Engage groups with broader oversight, such as the finance organization, if you think they might be an ally. Engage the BICC if you have one.


4. Pick one area for further exploration. Hold a kickoff meeting
and explore the need for more current information and where measurable process improvements could be made. Mock up ROI.


5. Determine which applications and processes have common
characteristics. Inspect the process workflows of the business, and identify where added real-time insights and injections of intelligence can improve business results.


6. Create plans to calculate and store analytical results, such as
customer lifetime value, lifestyle, and churn probability, in the data warehouse for ready access by the processes.


7. Analyze the costs and benefits of the top pervasive BI proposals,
and calculate the ROI before any additional monies and efforts are spent.


8. Roughly size the project.What assets do you already have in
place that could be used? For example, maybe you already have
next-best-offer analytics done for direct marketing purposes,
but you haven’t applied those ideas to the Web.


9. Mobilize resources and support for a first project. Consider
getting business people on the team, full time. Provide experienced people who can transfer skills to those in the organization who will have responsibility for building, enhancing, and maintaining the environment. This might require a temporary transfer of personnel or hiring experienced people from the outside.

 

10.Launch the project. Monitor progress, and stay on top of cross-group issues that may cause it to go awry. Institute justin-time training for impacted front-line groups. The timing of the training is very important, and there should be no lag between when the student returns from class and when he or she starts to work with the application.

 

11.Build a culture of reuse. If you do not have one, establish a BICC that is on the constant lookout for interesting, reusable insights. Precalculate and store analytical results, such as customer lifetime value, channel use predictors, and churn probabilities in the database for wide use. Raise the semantics of the data warehouse from “data” warehouse to “insight”
warehouse, with a focus on reuse.

 

 

Summary

 

The value of pervasive BI applications can be exceptional. The changes resulting from use of up-to-date data and better operational actions can vault an organization into the top tier of its industry based on customer satisfaction, cost reductions, improved operational controls, revenue growth, and reputation. The need and the value are industry-dependent. The ease and difficulty are functions of how much the organization has already accomplished, including its level of data integration (customer, supplier, and more) and data quality, as well as the skills and resources of the people who will be called on to implement the project. The key to success is organizational leadership, in particular business leadership, backed by an IT organization that fundamentally understands how to rapidly support the business people and systems with right-time data and insights.

 


The steps to adopting pervasive BI are straightforward; this paper showed how to inventory the excuses, anticipate the obstacles, and put a plan in place. Good luck!

 

 

Appendix A – Active Applications by Industry or Function

 

Banking
Identify potential loan defaults – Use pervasive BI applications to identify changes in credit worthiness to minimize defaults.
Make daily or hourly “mark-to-market” calculations of risk exposure – Use pervasive BI applications to identify underwater home and commercial mortgages and take appropriate risk-reducing actions.
Identify indications of leaving the bank – Use pervasive BI applications to identify a potential churn, learning of the reason, and providing options to retain the customer.
Process loan applications more quickly – Should result in fewer abandoned loan applications.
Identify credit card fraud – The sooner you know, the sooner the card can be deactivated.
Provide up-to-date consumer portal access to banking activity.
Monitor intraday actuals vs. goals on scoreboards for balances, customer acquisition, retention, expenses, and more.

 

Distribution
Dynamic Pricing – pervasive BI applications give the information to  change pricing based on distribution network dynamics (for example, excess  inventory, partial loads, geo-proximities) or create a new offering reacting  to a competitor’s offering or price change.
Monitor and change inventories – updating data can trigger manual or automatic orders to minimize stock-outs and to minimize bloated inventories.
Make near real-time transportation decisions – weather, catastrophes, and strikes require very timely decisions.
Optimize the supply chain:
– Detect and remove persistent problems in supply chain, such as a weak supplier or intermediary transportation companies not fulfilling SLAs.
– Six Sigma on the distribution process, real-time geodashboarding
on distribution status.

 

Government
Terrorism intelligence – Use a pervasive front-line check-in application to identify no-fly information for the airlines.
Immigration – provide an application for agents at points of entry, as well as pervasive apps accessible for employment verification checks.
Customs – provide pervasive apps for cargo shippers and transportation companies on customs clearance status and rules.
Taxes – provide pervasive apps for retailers that can be used to calculate the right tax, by geography.
Department of Defense – Purchase optimization, consolidation through visibility of orders in progress, fewer redundant orders. Tracking shipments. Availability of spare parts in other locations.
State and local government – pervasive BI applications minimize mistakes from decisions based on stale data.
Disaster analysis and actions (for example, oil spill impacts on wetlands and coastal regions, based on today’s drift patterns).
Food stamp fraud detection.
Drug and doctor monitoring – faster controlled substance reporting and follow-ups. Over-prescribing doctors and patient/doctor fraud.
Weather and road conditions.
Corrections – tracking perpetrators, inmates.
Resource deployment – Fire, police, utilities.

 

Insurance
Insurance underwriting – a broker can provide a quote and issue a policy in front of the insured without delays.
Insurance policy application status (broker and customer) – knowing the status of an application will minimize the chances that the policy applicant will cancel.
Property and casualty insurance (status of my claim) – up-todate
information about the status of a claim is a major factor in
customer satisfaction.

 

Manufacturing
Manufacturing process analytics.
Inventory vs. demand – Knowing the current status of the inventory and making appropriate decisions.
Supply chain, just-in-time made-to-order process optimization.
Demand chain visibility – see changes in demand as they are happening.
Plant capacity monitoring.
Manufacturing problems – earlier detection of warning conditions, real-time flow of instrumentation, test results.
Quicker detection of supplier problems, product quality.
Warranty, warranty reserves – automobile dealers have been known to perform warranty work where none was needed.
Dynamic prioritization of supply chain activities – determine which supply truck to unload first, which outbound products to ship first.

 

Medical
Procedure checklists – did we follow all the steps in critical medical procedures?
Medical claims – faster, more thorough processes, with in-line fraud detection.
Drug interactions, improper dosages, and incorrect medications.
Preliminary diagnosis especially in emergency situations.
Staffing – hospitals have the challenge of having the right staff with the right credentials at the right time without incurring the costs of overstaffing.
Optimizing communication with patients impacting mortality and morbidity – drug interactions, change in medical status, real-time device data feed monitoring.
Checking adverse reaction and contraindication possibilities – real-time nursing follow-ups.
Eliminate unnecessary procedures – If the data show that the patient has had a prostatectomy, and the consulting doctor orders a digital prostate exam, the system would flag that the exam is both inappropriate, as well as uncomfortable.

 

Retail
Target marketing – Marketing the right products to the people most likely to buy.
Pricing – especially where dynamic pricing can help smooth
supply/demand curves.
Demand alerts – especially for fast-moving new merchandise or fashions (for example, the latest iPad), knowing what to order and what to expedite.
Supply alerts, inventory, ordering – visibility and alerts.
Transportation – transportation modes (air, surface) and carrier charges.
Problems in supply chain – early out-of-stock alerts.
Return fraud.

 

Retail (continued)
Checker fraud.
Wrong pricing alerts.
Offers for special events.
Reduction in cycle time – packaging, collateral development,
advertising, marketing campaigns.
Marketing response rates – advertising effectiveness.
Speedy identification of tainted food, recalls.

 

Telecom
Put history of dropped calls on the CSR screen – along with indicators of how angry the customer has been on previous calls through emotion analytics.
Status Information – use pervasive BI applications to identify service request and repair status.
Fraud – use real-time credit checks and fraud history to vet orders and services.
Churn potential – encourage CSRs to spend more time with high-value customers at risk of churn.
Cross sell/up sell/bundles (phone, internet, TV, movies on
demand) – use pervasive BI to identify customers for up-sell service bundles.

 

Transportation
Tracking shipments (internal and external customers)
up-to-date location status.
Truck breakdowns – immediately see impacted customers,
SLA misses.
Scheduling – and rescheduling.
Routing – normal as well as dealing with weather or congestion
issues in cities.

 

Travel and Hospitality
Dynamic customer preferences – no seating next to crying babies, extradited psychopathic murderers, the morbidly obese, or anyone about to set any body parts or clothing on fire.
Fraud prevention – do not issue tickets paid with bad credit cards.
Service recovery – smarter responses for customers about delays, canceled flights, weather, and baggage problems.
In-flight report to pilots and flight attendants – delays, weather, airport problems, impacts on selected customers.
Monitor online check-ins and bags checked, which helps with dynamic staffing for counter and baggage.
Complaints:
– Avoid over-rewarding chronic complainers.
– Quickly identify source of problem.
Marketing – use detailed customer-specific information (past behavior, origin, destinations, travel interests), create:
– Personalized real-time travel Web offers.
– Focused emails.

 

Cross-Industry Applications
These applications are relevant for most industries:
Customer Management/Customer Support/Contact Centers
Fast authorization of sale or service – this includes highlighting the next best offer for a customer website or a CSR interaction with the customer.
More targeted offers – the targeted offers would be based on the demographics, buying patterns, service preferences, and all that is known about the customer.
Customer lifetime value – knowing an approximate value, you can provide an appropriate level of support. Put lifetime value calculations on the contact center agent’s screen, along with training to spend more time with high-value customers and less time with low potential value customers.

Next Best Offer – put next best offer on the contact center agent’s screen, along with training to provide the CSR with an offer the customer is likely to accept. This facilitates up-sell and cross-sell promotions.

Previous call experiences, complaints, services provided,
purchases, services, channel preferences – minimizes offers that have been turned down in the past. Put more context on the screen effort so contact center agents can immediately see the state of the customer.
Who is gaming the system – some customers have learned how to request multiple cost reductions, multiple promotions (only one per customer), and additional services for free. Without knowing previous interactions, CSRs can’t identify the gamers.
Customer differentiation, channel preferences – don’t treat them all the same. Use the contact center to help drive the capture of preferences, or capture missing information (for example, cell phone has changed, or emails are bouncing).
Customer self service – empower customers to access and use the website or the IVR.
Text analytics on call notes by CSRs.

Complaint resolution – current information about the customer, especially when they are angry, gives the CSR the best opportunity to resolve the problem and not lose the customer.
Service providers capturing all the data needed to perform the service or submit the request for service – once and done.
Customer support staffing – balancing out the workload to reduce wait time and hang ups and lower the cost of over staffing. Use pervasive BI applications to balance loads within and across multiple contact centers.
Reduce CSR costs – move answers to common questions to the IVR (“When is my repair guy going to arrive?”) as the first button push, or drive people to the Web (often a self-service project – measure all typical customer iterations, then do a project to move those to self-service channels).
Information about external/contract providers/servicers – especially important when customer support is outsourced.

 

Active Situational Awareness
Alerts – this could be any heads up that would require immediate
notification, attention, and possibly action. The alert could go to a person or could trigger an automatic response such as ordering a part in short supply.
Scoreboarding and dashboarding – in real time.
Event analytics, such as RFID scans, stock trading events and
flight logistics – accompanied by “event detectives” to screen for only “interesting” business events.
Ability to support governmental compliance – Important when non-compliance could result in fines, bad public relations, or decertification.


Process Instrumentation and Improvements
Predictive analytics leading to operational decisions – changing some processes, offers, prices.
Capturing intellectual capital to be used in pervasive BI applications – the knowledge of experienced personnel should be able to be incorporated in process changes near real time.
Measuring performance/results, ability to monitor effects of
operational decisions, make better decisions each year
compared to the last.
When there are bottlenecks in manual reviews – the reviews might be holding up loan approvals, licenses, visa applications, foreclosure servicing, or service requests.
Improved referral notification – when approvals need to be done ASAP.
Actions eliminated – no follow-up call is necessary – we’ve already spoken to the customer.

 

Market Planning and Analysis
Faster insights into 360-degree behavioral changes.
Customer segmentation – for example, yuppies, bohemians, landed gentry.
Customer profitability including customer’s lifetime value.
Customer tiers (novice, experienced) – propensity to use technology.
Next best offer to customer.
Personalize interactions.
Personalized notification – The new Harry Potter book will be on the shelves next Wednesday; the alligator pumps in your size just arrived.
Preferred channels – no phone calls, email only.
Dynamic pricing, reaction to competitors’ pricing and promotions, optimizing markdowns.
Sales forecasting.
Advertising effectiveness and planning – this is especially important with the rapid changes available with online advertising and online advertising bidding.

 

Finance/Credit/Risk
Offering credit terms to customers/borrowers who are
behind on their payments.
Fraud detection.
Financial revenue scenario management, yield, price
optimization – modifying prices based on supply (number of
empty seats), customer’s price sensitivity, cost, channels.

 

Books We Recommend
• Thomas H. Davenport and Jeanne G. Harris, Competing on Analytics: The New Science of Winning, Harvard Business Press, 2007.
• Thomas H. Davenport, Jeanne G. Harris, and Robert Morison, Analytics at Work: Smarter Decisions, Better Results, Harvard Business Press, 2010.
• Mark Jeffery, Data-Driven Marketing: The 15 Metrics Everyone in Marketing Should Know, John Wiley and Sons, 2010.
(http://www.amazon.com/Data-Driven-Marketing-Metrics-Everyone-Should/dp/0470504544 )
• James Taylor with Neil Raden, SMART (ENOUGH) SYSTEMS:
How to Deliver Competitive Advantage by Automating Hidden Decisions, Prentice Hall, 2007.


Articles We Recommend
• Judith R. Davis, Claudia Imhoff, and Colin White, “Operational Business Intelligence: The State of the Art,” BeyeNETWORK Research Report, 2009. A good overview of the state of the art, with a focus on pervasive customer management.
• David Garrett, “Moving toward Operational Intelligence: A Guide to Navigating the Active Enterprise Intelligence Project,” Teradata Magazine, June, 2007. A second good overview.
http://www.teradata.com/tdmo/v07n02/Features/SpecialFocus/OperationalIntelligence.aspx • Dave Schrader, “Build a Better, Faster Value Chain,” Teradata Magazine, May, 2010. Highlights five opportunities in the value chain area where pervasive BI can help.
http://www.teradata.com/tdmo/Article.aspx?id=14207
• Dave Schrader, “Think Fast,” Teradata Magazine, December, 2009. Highlights areas where pervasive dashboarding/ scoreboarding is useful.
http://www.teradata.com/tdmo/Article.aspx?id=12626

 

About the Authors
Sid Adelman is a principal in Sid Adelman & Associates, an organization specializing in planning and implementing data warehouses, in data warehouse and BI assessments and in establishing effective data strategies. He co-authored Data Warehouse Project Management, Data Strategy, and is the principal author of Impossible Data Warehouse Situations with Solutions from the Experts. His web site is www.sidadelman.com.


Dave Schrader is the Director of Strategy and Marketing for Active Enterprise Intelligence™ strategy at Teradata Corporation, the world’s largest company solely focused on creating enterprise agility through database software, enterprise data warehousing, data
warehouse appliances, and analytics. He focuses on customer best practices, case studies, and how to apply intelligence to operational touchpoints like contact centers, web sites, and mobile devices.


logo
TAG’s article library is optimized by search engine optimization company Vayu Media, provider of SEO Services to technology companies nationwide. Vayu Media develops online strategies to grow businesses through increased sales, leads and internet brand awareness.

 


 

Member Status

Facebook Fans