Doing BI with Google

After a couple of months doing many other things, I'm back with the BI blog. This week the topic is Google.

I have an startup called Cosmetix. To analyze different KPI´s and campaigns we're using google analytics.

There is really BI. With the tool that they're giving to us for free, and on the other hand with the tool to analyze our adwords campaings, we have very valuable information, that can allow us to make decisions faster, in an assertive way and optimizing a lot our focus and budget.

My point is. Google is the leader on Internet. Now, they are offering a vast range of tools for companies and individuals (from translators to maps). How many time we've to wait to see a BI tool from Google ?

In fact, many tools like Penthao or Microstrategy are using Google Maps to show geographical information into their dashboards.

What do you think about the future of BI if a company like google make the decision to be in ?.

Open Source BI

In the last year, many people begin talking about the open source software for BI. In fact, if you see the trends in source forge you'll see a huge increase in the downloads from this site.

But the question is not about “traditional BI” vs. “open source BI”, to me the question is about the maturity of the open source solutions.

We're clear that if we buy a traditional BI solution, we're buying software from a company with a reputation, several projects developed in different markets and the two most important things, a clear roadmap of the platform and an support service free for the first year and for subscription for the next years.

In the last two year, many BI companies were sold. Oracle bought Siebel and Hyperion (and many other little companies), SAP bought Business Objects, IBM bought Cognos and so on.

In most of the cases, the company who buy say that the want to continue with the development of the products they've bought, but nobody really know if it will be true. In fact Oracle is going to integrate Siebel and Hyperion to their suite.

But now, going to the open source BI, the big question is about the maturity of the software and on the other hand if the companies can survive only with the software subscription programs.

When a company make the decision to select a open source BI, must have in mind that if is true that open source means “for free”, they've to invest in a good consulting firm to drive the project in order to assure the success of the implementation. Also, is good to know the success stories of the software house, if they have certified partners in your country, the product roadmap and if the software house was consistent in their growing.

Most of the open source BI companies have a complete suite going for operational reporting to dashboards and from data mining features to ETL tools. However, you've to be careful because some times some solutions are not strong and you can experience some issues (data mining tools is an example).

Another important point, is the software subscription. If is so that open source means that anybody can modify the code to have the right solution for your company, its highly recommended to take the software subscription and let the software house to deliver the right version and use the internal consultants to maintain the application.

Most of the open source BI software are made in Java, that is very easy to modify and configure in order to obtain the right product for your company, but always talking to modify the products features but not the product kernel.

To finish, I strong recommend to see the solutions from Penthao, Jasper Software and Green Plum, that are the most mature and professional companies on open source BI.

The best BI tool

As BI consultant, every week I have meetings with customers that have the need to analyze information to make better decisions.

In the 99,9% of the meetings, in any moment appear the same question; “which is the best BI tool for my company”. The years of experience, the common sense and the commitment to provide solutions with value added and that can be used for the companies to be more successful, to earn more money and to be better than competitors, makes me to answer always in the same way.

The best BI tool for your company, is the tool that after a detailed analysis of the processes, after to understand the needs, what information the company want to analyze, which is the frequency, level of significance, who will be the users, etc, fit better the company needs.

A mistake that unfortunately happened several times, is that companies select tools, thinking that are a Microsoft Office box, and at the end of the day, that situation produce a big frustration in the end user, because most of the times the user was not heard for the evaluation. The trends, show us that more than 80% of BI projects fail before the first year of the kick off day. When we’re talking about fail, we understand that users are not acceding to the solution and remain using the old tool that were installed before the installation of the BI project, screaming something like “I don’t trust in the new solution, is difficult to access, I don’t understand how to read the data”. If a person, access one time to the new tool (usually sold as the solution to solve any company problem), and didn’t see what is expecting, surely will access a second time and never more.

To finish, a BI tool could be a little bit nice, a little bit ugly, cheaper or expensive, with any additional feature, but if the process to analyze information is wrong, everything will be wrong. The tool is a commodity.

For this reason, before to make a decision about which BI tool will be selected and make a decision about which is the best, the users have to analyze the whole process that will be automated, and as I said before, this analysis evolve, from KPI’s, data sources, users needs, expectations, vision in a mid term and many things more.

Key Performance Indicators and Data Mining

The terms KPI and data mining are often used to discuss the benefits of BPM and the ways in which BPM drives business decisions. Knowing what those terms mean, however, does not alone guarantee business success. Instead, organizations should identify appropriate ways to apply KPI and data mining, in order to determine the metrics required for making the right strategic decisions.

KPIs are defined as the critical metrics set by an organization to reflect its financial or nonfinancial success. They help organizations identify and monitor factors that are quantifiable, measurable, and important to the organization's overall success. Although KPIs can help drive business decisions, they are only beneficial if they are set properly and reach the right people at the right time. For example, with traditional BI online analytical processing (OLAP) cubes, sales data can be reflected multidimensionally with rolling sales data over a three-year period. This, however, pales in comparison to dashboard functionality, which allows a sales manager to see up-to-date sales figures in real time, and to compare them against predefined metrics. The sales manager can then drill down on the data to access and analyze operational data, in order to determine a plan of action.

KPIs vary, depending on the function of an organization. A nonprofit company may want to measure the ratio of graduates to overall participation for a specific volunteer training course, in order to identify the success of a program. However, a sales-oriented corporation may want to set metrics to identify the amount of revenue generated by return customers. To increase sales, a KPI measuring customer satisfaction and repeat sales might be implemented. For a financial institution, it may be important to set KPIs to identify potential risk management issues, such as meeting regulatory requirements or minimizing the potential credit risks of clients. This differs from a manufacturing organization that needs to monitor parts delivered from suppliers, or from a government body that wants to measure and improve employee performance. Setting the right KPI and providing that information to the right people can make the difference between implementing a successful BPM tool and a total failure.

Data mining, also called knowledge discovery in databases (KDD), uncovers data patterns within databases. It is used as a tool to discover patterns among large amounts of data. Data mining allows organizations to identify why things happen, and helps them make connections between seemingly unrelated items. For example, if an organization wants to increase sales, identifying customer buying patterns with intuitive software saves time, and allows decision makers to focus on developing strategies based on those patterns (as opposed to spending their time identifying what those patterns are). A credit card company may want to identify buying patterns and spending habits of customers, and an organization in the pharmaceuticals industry may choose to create a KPI to improve their manufacturing process and inventory control.

Data mining can also be used to find patterns among multiple tables within relational databases. This is advantageous because data centralization (and having one view of corporate data) enables organizations to set the appropriate metrics. Data mining also allows them to measure these metrics more easily, find patterns which enable proactive corporate planning, and target customers based on pattern recognition.

Before identifying more ways in which BPM can benefit an organization, it is important to identify the user interface components, to show how organizations are using performance management software

This article was taken from Technology Evaluation Center

BI as a Service

A huge problem that several BI projects have is the return of investment (ROI). As we’ve talked in a previous post, many times this ROI is difficult to calculate because it’s not tangible.

I think that a feasible alternative for this projects is to provide BI as a service, where the company can contract a service to consulting firm including Software, Hardware and Consulting. In the fist two topics, company will pay an outsourcing cost, that could vary according the last of the contract and in the consulting topic, the project will have a setup cost and an outsorcing cost in the same way than software and hardware topics.

To finish, the company should not invest in goods that are neither strategic nor core business, which have a positive impact at the time to justify projects with the finance department. With this scheme, the objective is to build a relationship with a high level of commitment and a reliable society between both parts in the long term.

Data Warehouses and Data Mart

Because data can come from various sources, like OLTP, ERP, CRM, legacy applications, and external data sources, data can be stored in a diversified database, in different formats and structures. As a result, a data warehouse (DW), is the most important and expensive player in the whole BI system because it captures data from these diverse sources, and unifies them. The data is then ready to be accessed by BI system. As a central repository of business, DW contains data used for decision support systems (DSS) which focuses on the lower and middle management and makes it possible to look at and analyze data in different ways. Such data also used for executive information systems (EIS). Data extracted from the DW by departments are collected and put into smaller repository for easy and fast access are called data mart. Like data mart for marketing data, sales, production etc.

Five Steps to Business Intelligence Project Success

This is an interesting article that was published in the Technology Evaluation Center site.

Five Steps to Business Intelligence Project Success
Lyndsay Wise - June 18, 2007

Successful business intelligence (BI) projects encompass more than implementation of a solution on time and within budget. True success should be measured by how the BI solution improves the organization's overall performance through increased efficiency in reporting, planning, financial functions, and performance measurements. This will help ensure organizations' BI projects fall into the estimated 30 percent success rate.

Much has been written about measuring return on investment (ROI) for BI, and the general conclusion is that gaining tangible insight into the initial benefits is not easy. Identifying long-term benefits becomes more practical as planning and analysis, compliancy, and forward-looking approaches become more mainstream within organizations. To gain insight into how to implement a BI solution successfully, organizations should benchmark the success of other organizations, including their implementations and use of BI, against their own current initiatives. It is equally important that organizations learn from other organizations' failures—and avoid repeating them.

This article identifies and explores five steps that organizations should take to avoid the common pitfalls encountered by many businesses when implementing a BI solution. These areas include the identification of the business problem, BI tool use, the delivery of data, training initiatives, and development of a framework toward choosing the proper solution for the organization. These five areas provide an overview of items that should be identified before implementing BI within an organization or business unit.

Step One: Identifying the Business Problem

Identifying the BI business problem is the first step to ensuring a successful project. Once an organization knows what is broken, not only can it start to find ways to fix the problem, but it can also identify the proper resources, create buy-in, and prioritize how to tackle the project. To produce an ROI, a BI solution needs to correspond to an organizational business problem; otherwise, implementing an ad hoc query tool, online analytical process (OLAP) cube, or dashboard will not create lasting benefits.

Unfortunately, it is common for BI solutions to be pushed onto a business unit in order to meet an information technology (IT) objective rather than an organizational need. Sometimes, organizations get caught up with general initiatives and lose sight of the actual benefits BI provides in terms of performance management, collaboration, workflow, process improvement, etc. To attain buy-in, the user community should be a part of the problem identification process. An implementation decision that comes from management still requires input from users as to what their requirements are, and this information can make the difference between the implementation of a tool that works as a value proposition and an implementation that may be seen as useless.

Step Two: Determining Expectations of Use

Once BI is implemented within an organization, its usage usually grows beyond initial expectations. For example, an organization may assume that its BI implementation will involve 10 to 20 users, when in reality, over 400 users query data on a monthly basis. Based on the initial design of the platform, the system may not be built to sustain such a high number of queries, and will most likely "crash" (fail), causing users to lose faith in the new system and potentially revert to their pre-BI environment for stability.

In addition to a lack of confidence, getting an unstable system up and running may not seem worth the effort, delays, or time it would involve. With unrealistic expectations, frustration may cause the organization to rethink its use of BI. Generally, once BI adoption occurs within one part of the organization, other departments or business units see the benefits, and adoption begins to spread throughout the organization. For an organization to meet these needs, the anticipated use of BI should be determined beforehand.

Another consideration is the type of BI tool use. For example, if a sales manager needs to increase sales and wants to analyze trends, product distribution, and sales performance, creating a set of static reports will not produce the appropriate results. A data visualization tool to manage these items and develop a plan based on trend analysis will more likely produce the appropriate results.

Step Three: Understanding Delivery of Data

The collection of the right information for reporting and analysis becomes essential to deliver value to organizations. Although the identification of required data is time consuming, it is the backbone of BI. Additionally, identifying how data will be delivered, what the appropriate data cleansing activities are, and whether data is delivered in batch or in real time should be defined in advance. If data is not cleansed or delivered when needed, then the front-end BI tools will not provide the proper value to the organization. BI solutions impart value through the analysis of data, so it is essential for data to arrive when required, in the proper format, and at the right time. In addition to extract, transform, and load (ETL) tools, data quality and data cleansing need to be inherent aspects of the delivery of BI within the organization. In reality, short of an organization-wide master data management (MDM) initiative, the responsibility of providing accurate data will fall on the shoulders of the business units implementing BI.

Some organizations are misguided and think that their BI solution will provide them with the tools to fix their data problems. BI solutions can provide ongoing data quality processes, but these are not innate to software offerings. Some BI tools include enhanced data quality and integration features, and some vendors assume this responsibility should fall to the organization. Organizations should implement data management structures to minimize frustrations that result from data issues.

Step Four: Rolling Out Training Initiatives

Deciding when to roll out training contributes to project success. Training initiatives should begin right before or during the implementation phase. However, in many organizations, training is rolled out months before actual implementation, creating hype among the employees about the new system and what they will be able to do with it. By the time implementation actually occurs—months later—the initial excitement and buy-in has subsided, and more importantly, users have forgotten their newfound skills. To build momentum again, training needs to be repeated—wasting time and money.

Buy-in related to change is never easily achieved within organizations. Users become attached to their current processes whether or not those processes are productive. Buy-in does not occur immediately upon showing users the inherent value of BI, as their whole way of doing business will change. Creating a training program and delivering that training in a timely fashion helps users apply their newfound skills immediately, and helps to increase user buy-in.

Step Five: Choosing a Vertical- or Horizontal-based Solution

Organizations should identify whether more value will be provided by a vertical solution that is built specifically for the organization's industry or department, or by a horizontal solution that can grow with the organization. For example, does the organization need a generic reporting, querying, and analysis tool that will extend across the organization, or does the organization need to develop a process and compliancy that will adhere to the US Sarbanes-Oxley Act (SOX) or Health Insurance Portability and Accountability Act (HIPAA) standards? The answer will help the organization define which type of solution will best meet its needs.

In addition, the use of BI in the future and its anticipated usage may help determine whether a horizontal or a vertical solution will best meet the organization's needs. Organizations requiring compliance should take advantage of vertical-based solutions because vendors have developed solutions to meet specific compliance requirements. Horizontal solutions need intense customization to bring them up to par, leading to extra time and money spent on developing the solutions.

Organizations in key vertical industries should strongly consider vertical-based solutions that will meet their needs out of the box. Vertical-based solutions are likely to meet the general requirements of a specific industry or department, but since horizontal BI solutions do not base themselves on specified data models, they may be more versatile to the changing demands of the organization. Therefore, if an organization anticipates rapid BI growth across the organization, having the ability to develop solutions based on individual needs may be more beneficial. This relates to identifying the business problem and anticipating the future needs of the organization.

Conclusion

All too often, BI projects fail to meet an organization's expectations. But with research, planning, and a solid methodology, such failure can be avoided. To help ensure BI project success, organizations should work through these five essential steps: identifying their business problems, determining how they will use their BI solutions, knowing how and when data is delivered, rolling out user training initiatives at appropriate times, and developing a framework for selecting the type of solution that will best fit their organizations' needs.