Ferraris don’t run on kerosene: The case for good data

Madhu Nair
Madhu Nair|April 19, 2013

Imagine a shiny, new red Ferrari in your driveway. You have splurged on it and cannot wait to rev the engine and pull ahead of the annoyingly slow cars on the highway. This is the machine that will take you to your destination with style and speed.

Would you run your Ferrari on kerosene? You would probably not even use regular unleaded gas and might even think twice before you use mid-grade fuel. Because you know fuel matters. The higher the quality of the fuel, the better the performance.

Do you do the same when trying to pull ahead of your competition?

Now, imagine you have invested in a best-of-breed business analytics solution or operational application. You are installing it, dreaming of getting that optimized recommendation to a business problem that is going to put you ahead of your competition. For example, customer satisfaction might skyrocket because your new marketing application is going to offer the best upsell product for your customers. But have you given a thought to the data that is going into the solution?

Data fuels all of your business technologies. And just like with fuel, quantity doesn't equal quality. Here are top three kinds of data that you might want to consider.

  1. Unleaded or regular data – This is basic data. You collect data and do some basic transformations before you feed it into your business solution. Sure, it works. But you don’t really trust the results of analyzing this data. With this kind of data you are the slow one on the information highway.
  2. Premium or standardized data – This is data that has been clean and enriched. You standardized the data, enriched it, and accounted for outliers. You spent time preparing the data, and it shows in the dashboards and reports. Apart from some extra time spent reconciling your reports (or double checking the recommendation of your business solution) you are a happy soul. You are cruising along the information highway.
  3. High-octane or master data – This is the path to data nirvana. Data is clean, enriched, governed and trustworthy. You have a consolidated view of every entity you deal with. Pure data is always on hand to feed your applications. Your departments and decision makers eagerly await every single report that your installed technologies churn out. Your competition is eating your dust in the information highway as you pull ahead of them.

You can't always buy a Ferrari. But when you do buy one, make sure you get the right fuel.  Similarly, you can't always get the best business application. Which is why when you invest in one, you should make sure you are feeding it the best possible data.

After all, Ferraris don’t run on kerosene.

Post a Comment

Narrowing the lead on big data early adopters

Madhu Nair
Madhu Nair|April 9, 2013

A new SAS survey uncovered a big data disconnect, with only 12 percent of organizations on board. Why weren’t more of the organizations surveyed taking steps towards a big data pay off?

Without a doubt those that are implementing big data strategies will see a competitive advantage. And the longer they build that lead, the harder it will be for others to catch up. However, for those that are waiting, all is not lost. Improved technology and processes can help them make quick progress on the path to big data mastery.

If you're still in that 88 percent waiting to get started, here are some tips.

Start with the business side when planning big data initiatives – Most CEOs and VPs know their teams are missing opportunities by not using big data. Line-of-business managers then investigate opportunities (sometimes lost or wasted) in their business decisions to understand where big data can help the most. Is it in cost reduction through identifying fraud? Or is it more important to increase profits through knowing your customers better? Those answers should guide the effort.

Share your project and data – Business management should involve IT early in this process to assess how big data technologies can augment existing technology investments and see where additional investment is needed. Then, business and IT will do well to define data ownership, data standards and project-success metrics—all critical components of data governance- to set the stage for big data project execution.

How SAS can help – SAS continues to address customers pain points when it comes to big data. Processing power and data storage are no longer prohibitively expensive. SAS provides access to big data sources such as Hadoop so organizations can easily bring big data into existing environments for management and analysis. SAS also delivers big data processing products that leverage in-memory and parallelization to increase performance – an increasingly popular capability. Interestingly, with these advances, data sampling is becoming obsolete because organizations are no longer limited by ability to handle massive data volume.

I especially like how Hong Kong Efficiency Unit uses SAS to decode citizen messages. The government can better understand the voice of the people and help government departments improve service delivery, make informed decisions and develop smart strategies. I have never been to Hong Kong, but I have heard great stories about the city’s efficiency. No wonder.

To the many of you who might just be getting started all I can say is: don’t wait. Choosing the right big data management vendor makes this an interesting and rewarding investment for your company.

Click here for the full big data report I mentioned earlier.  And tell me about one of your own big data challenges or successes. What's working for your rganization?

Post a Comment

Everything is "big data" these days – but where is the data management?

Daniel Teachey
Daniel Teachey|March 28, 2013

When I first began working in the data management industry in 2003, I interviewed a VP of IT for a Fortune 500 manufacturer about their data quality and data integration initiatives. The executive was excited to talk about their rather novel approach to a massive ERP system integration. She mentioned that instead of just dumping data into a new ERP system, they were taking 18 months to investigate the quality of their information, write some rules to guide the management of data over time and monitor those rules within the ERP system.

That doesn't seem very groundbreaking, does it? They simply wanted to make sure they didn't saddle end users and analysts with unreliable and inaccurate data. Yet, at the time, the idea that IT and business would agree to delay the implementation to address data issues was not a very common concept. Data management was too often an afterthought, but this executive – and other key stakeholders within the organization – realized that the success or failure of that ERP project was the data within the system.

Today, if you read IT publications and websites, the concept of a "simple" ERP implementation seems quaint. Everything is "big data" these days, with information flooding the enterprise from social channels, trading partners, radio tags, etc. As Jill Dyche pointed out in a recent Information Management article, "Big Data's Three-Legged Stool," the amount and complexity of data may have changed, but data is still data. I'll let Jill take it from here:

Acquiring specialized technology and maturing analytics behaviors aren’t easy. But what people don’t know — at least, not yet — is that the hard part of big data is managing it. The challenges of identifying and sourcing the data, applying data correction rules, circumscribing usage, access, and storage policies, and provisioning the data to other platforms and applications requires its own set of rigor. Regulatory requirements mandate that your bank mask Social Security numbers before availing half a billion credit card transactions to hungry data scientists hoping to fortify themselves on fraud indicators. Simply applying a new file system and some statistics to the problem without first applying business rules to the data can mean large fines and, maybe worse, additional regulatory scrutiny.

Sounds very similar to my conversation 10 years ago. This time, instead of looking at enterprise applications, we're talking about pedabytes of data from a host of sources stored in more powerful appliances designed to store mind-bending amounts of data.  But will a big data platform do any better than an ERP or CRM system filled with questionable data? Are we creating a bigger mess here? Jill?

The promise of big data analytics is as expansive as our imaginations. But I’ve also seen the garbage-in, garbage-out phenomenon writ large on the balance sheets of naïve executive teams. Solid data governance and data management processes can mean the difference between new legacy technologies and innovative business actions.

Yes, just like in 2003 (and in 1993 and 1983), data only has value when it means something. Big data is a great thing, but the phenomenon will also need all of the data quality, data integration and data governance skills that we can muster.

Soon, SAS will release the findings from a big data survey. The survey shows that while big data awareness is high, preparedness and strategies lag behind. The same happened with enterprise applications a decade ago, and many organizations paid dearly for going to fast – or moving too slow. Big data will be no different.

Post a Comment

Data management on display at SAS Global Forum

Daniel Teachey
Daniel Teachey|March 19, 2013

Years ago, I was at a user conference, listening to an attendee talk for 15 minutes during a break about a data quality job he had created. He had good reason to celebrate. The job had reduced the amount of work required to clean up product data within a data warehouse. How much of a reduction? What once took 40 percent of a data analyst’s time each week now required a 15-minute processing job. This was a true data quality "enthusiast," and his excitement was contagious.

This is why I've always loved a good users conference. These events can spur hundreds of conversations about how your technology works – and how it can improve the data that drives an organization. This information is indispensable, giving fellow technology users a chance to find new ways to solve old problems. It also gives those of us on the vendor side a unique insight into how the technology is used every day.

Each year, the SAS Global Forum provides educational and networking options for professionals focused on maximizing the value of data. In 2013, look for more conversations around data quality, data integration, data governance and master data management (MDM). These data management topics are a big focus for this year's event, being held at the Moscone West in San Francisco, April 28-May 1.

SAS Global Forum 2013 will feature hands-on workshops, demos and training focused on MDM, data quality, data governance and more. This year's agenda features a variety of options for attendees, including:

  • Pre-conference training on April 24-26, including DataFlux Data Management Studio: Basics
  • 50-minute hands-on workshops for SAS DataFlux technology
  • Data management demos

The event isn't just about training and demos, however. SAS Global Forum provides ways for users to learn the latest tactics and techniques that can maximize their data management implementations. Look for presentations on a variety of topics, including:

  • Best Practices in Enterprise Data Governance
  • What’s New in SAS Data Management
  • How to Do a Successful MDM Project in SAP Using SAS MDM Advanced
  • SAS Data Integration Studio: The 30-Day Plan

Other sessions will focus on in-database data quality, entity resolution, optimizing for performance, and many more. For all SAS DataFlux users, this will be the place to be to share ideas and find new ways to maximize your data management program.

What sessions are you looking forward to at SAS Global Forum?

Post a Comment

Data stewardship and the benefits of employee retention

Daniel Teachey
Daniel Teachey|January 16, 2013

Like most Americans (who hold an average of 11 jobs over the duration of their careers), I’ve worked for a variety of companies. Some were employee-friendly, while others, to put it kindly, thought of employees as a means to an end.  I was thinking about all those jobs today as we found out that SAS ranked No. 2 on the 2013 FORTUNE list of Best Companies to Work For in the US.

SAS has a long tradition of being employee-centric. Whether it’s a fully stocked break room or the wealth of on-campus services, it’s a terrific work environment. Naturally, there’s a method to the madness.  Happy employees stay longer, minimizing employee turnover while retaining irreplaceable knowledge. It’s a different culture than any I’ve ever experienced, but the results are impressive.

Employee retention can be a key factor in the data management world. As data environments become more complex, the knowledge of how the pieces fit together is becoming a career-defining attribute. Those employees who know how the data is aggregated and maintained are increasingly valuable in the workplace.

Recently, these individuals have been getting their fair share of attention. On Data Stewards Day, the industry celebrates the “best and brightest – the experts who not only manage your company’s data, but keep your business running behind the scenes.” People can nominate their co-workers for a “Stewie” award, and while the submissions reflect a variety of backgrounds, there is one common theme: Stewie nominees are generally established employees whose “organizational memory” allows them to solve complex data challenges.

That shouldn’t come as a shock. Data stewards are often the go-to professionals for both simple and complex data-related questions. Data stewards can tell you how a customer or product is defined in a particular data source. And why a certain field differs from system to system. And how to reconcile those views. These employees have aggregated immense knowledge of the data within an organization and how it has been mishandled in the past.

Organizational memory is one of the most valuable resources a data steward can have. No matter how much documentation you have on a dataset, nothing replaces the human memory. A long-serving data steward fought the battles to establish a data source, and they likely know the best way to maximize that information. They know the upsides and the downfalls of previous data management programs, and they can steer the ship away from disaster in the future.

Years ago, the concept of data stewardship was just starting to emerge, and it’s good that these data champions are getting their recognition. Organizations should do more than just acknowledge those employees.  They should focus on cultivating and retaining these valuable resources. Additional resources (human, financial and otherwise) focused on data management can yield a significant ROI. Perhaps the big data crunch will lead to more investment in this critical role, and here’s hoping that big data stewardship is on the horizon.

By the way, if you’re a data steward reading this, feel free to “clip and save” this for the next employee review. It may not lead to the job of your dreams, but maybe you’ll get some free snacks here and there.

Post a Comment

Keep the change: Why investing in change management matters

Analise Polsky
Analise Polsky|January 11, 2013

The always dynamic data environment makes it difficult for organizations to avoid change. In my last post, I talked about how change management can bolster data governance programs. It can also be used for other activities, including taking a portfolio-driven approach to business intelligence and adopting a new enterprise application or revising your IT architecture.

All of these activities entail some level of transformation that will affect your current processes, policies, or business strategies. Here is the reality: Without proper oversight, most of these changes will fall short of expectations or fail outright. The more an organization experiences failure, the less likely it is to make major changes in the future, even if it is vital to improving business.

The following examples are common problems that have the potential to undermine the success of your next initiative. For each, I have highlighted ways in which change management can help you address, overcome, or prevent them.

1. Failing to establish a vision

Organizations that do not establish a vision, especially one that addresses the transformation about to take place, are opening the door to confusion and missed targets. Your vision should outline the goals of your project and establish a clear direction for how the goals will be achieved. This process answers the critical questions: What is changing; how is it changing; why the change will impact your business; and what you hope to achieve as a result (targets). A vision is also what will coordinate employee’s actions by giving them a shared goal to work towards. The less uncertainty employees have the more committed they will be to the initiative.

2. Thinking too big

You may desire the enterprise wide use and adoption of analytics, and it is certainly a worthy goal; however, you are unlikely to achieve it in a single step. Many organizations skip the smaller phases and expect universal results. This underestimates the weight and complexity of most changes. How will you create enterprise wide commitment? Is there a fair way to measure the success of the program across all departments or verticals? How will you demonstrate ROI to executives?

In our four-phased approach to change management, we emphasize the importance of small, controlled projects. Pick a starting point for your initiative, so you can more easily develop the vision, adjust the process, and measure the results before you take it to the next level. You can also manage the people side of change more effectively and maintain focus on the initiative if you are working with a distinct group of people.

3. Ignoring employee morale

Change freaks people out. As morale declines, employees are less productive. A Gallup study found that employee disengagement costs the U.S. $300 billion annually.  You also don’t want to lose employees. Hiring and training a new employees can cost your organization exponentially more than keeping an existing one.

Communication and planning are fundamental components of change management. It’s important to communicate with employees about the vision for the change, the steps that will be used to enact it, and the positive impact it will have on their work day. This will keep employees motivated and engaged throughout the process. Balance conversations with insight into areas your organization is succeeding and where it can improve. Employees tend to be happier (read: more productive) and commit more to initiatives when they know what is going on, why it is happening, what their role is, and how leadership is committed to the effort.

Post a Comment

Data integration: The changing nature of tools and experiences

Charlotte Crain|December 18, 2012

When I first thought about writing a blog post about data management, picking a topic was an obvious, if daunting, first step. The challenges facing both IT and business are increasing in the era of bigger, more complex and more immediate data. My experience working with organizations across a variety of industries has taught me one thing: the management of a company’s data is the foundation for any business.

Another thing I've learned is that data integration is one of the most common and most established methods of data management. In today’s organizations, with data arriving from a multitude of sources, making sense of varied and disconnected information is a concern for both business and IT. The business side of the organization needs a cross-functional view of data to make good decisions, while IT needs to make sure a trusted, coherent view is available, safely and securely, for business users.

As a result, a comprehensive and flexible data integration strategy is necessary for any organization. For example, the need to capture marginal revenue, increase market share, and improve customer experiences result requires a company to truly “know the customer.”  And knowing your customer means knowing your data.

Data integration technologies were formerly known as ETL tools for the “extract, transform and load” processes they managed. For years, organizations viewed data integration as the process of getting information into a data warehouse or similar consolidated data target. Organizations now view data integration as a practice that doesn’t exist in a vacuum – it is part of a larger data management construct.  Data integration affects (and is affected by) a variety of initiatives, including:

  • Enterprise data governance
  • Changing business processes and data consumer requirements for data
  • More and different types of data and their sources (social media, RFID, etc.)
  • Decision support, business intelligence and advanced analytical needs
  • Real-time and near-real-time data requirements that fall outside of typical batch data integration

These initiatives are changing data integration technologies into more mature platforms.  Data integration is often part of larger implementations, such as SAS Solutions, where data integration is critical precursor to analytics and reporting efforts. Data integration is also a part of operational master data management (MDM) initiatives, helping create a more unified view of customers, products or other information types to support enterprise applications.

To accomplish this larger role, data integration suites, part of SAS DataFlux Data Management offerings, include functionality for a broader set of functionality, including:

  • Data access – Access and manage enterprise information and publish integrated data back into the IT environment
  • Data quality – Enforce data cleansing and data standardization rules to produce accurate, consistent information
  • Data governance – Manage and enact business rules to ensure data is “fit for purpose”

In the next post, we will discuss some of the interesting attributes of SAS Data Integration, including its role in high-performance computing, high-performance analytics and decision management.

Post a Comment

Planning the MDM journey: SAS DataFlux delivers phased MDM projects

Matthew Magne
Matthew Magne|December 17, 2012

2013 is rapidly approaching and everyone is focused on change, whether it's a shift in your current role, a New Year's resolution or a new business plan. Similarly, organizations are looking at their master data management (MDM) strategy with an eye on the accompanying change management. There is a shift towards a more reasonable approach to supplant the "pie in the sky" or "boil the ocean" mentality that accompanied many early MDM efforts.

David Loshin, president of Knowledge Integrity and a long-time consultant and thought leader in MDM, once wrote, "Pure and simple: The most critical factor to master data management is data quality." As more complex data sets become the norm in many enterprises, data quality will continue to increase in importance. At SAS, we have long understood that MDM is more of a journey than a destination – and it's part of an overall information management effort with many phases.  SAS Information Management provides both data quality and master data management capabilities as well as the consulting expertise at any juncture in the MDM lifecycle.

As you move through your MDM journey, you will obtain a unified view of customers, products or other enterprise assets. The project will also involve a wide variety of other disciplines including data governance, data quality, data integration, and identity resolution, all driven by business goals like optimizing revenue, reducing costs, or meeting compliance and regulatory requirements. A simple but effective planning process can make all the difference in coordinating efforts in this complex project.

MDM planning process

Data governance and MDM go hand in hand, as evidenced by the speakers at the 2012 SAS DataFlux IDEAS conference. Anytime you consolidate data from multiple data sources, it requires some rigor around defining the master entity objects, understanding the affected processes and allocating supporting resources. At SAS, we have a team of consultants with hundreds of years of combined experience in this space that can assist you in planning your MDM deployment to drive business value, reduce project risk, and improve collaboration between business and IT.

Since MDM is a large undertaking, IT and business groups require a way to manage this complicated process – and set the groundwork for more unified corporate information. Ideally, companies begin an MDM journey with the following processes:

  1. Business case generation – This phase requires you to generate and prioritize of business requirements through business stakeholder interviews. This information is then distilled into business use cases that frame how MDM can deliver business value.
  2. Source, business rule and entity definition – The second phase focuses on the translation of business requirements into technical requirements. A joint effort between IT and business, this phase identifies business rules and entity definitions through a governance process. From there, you can identify, prioritize and profile data sources to assess the level of data quality prior to MDM integration.
  3. Plan a phased implementation – After the initial data governance efforts, you have answered the questions and started to deliver the business rules and IT logic to manage master data.  Now, you can generate and implement a phased MDM project plan.
  4. Value augmentation – In this phase, you are building on prior project success, mastering new domains, or integrating additional sources into existing domains, always with an eye towards expanding the initial investment footprint into driving more business value. Internal evangelization and alignment is critical.

Where SAS fits in

As part of the mapping phase between business and IT, the Business Data Network component of the SAS Data Governance solution allows business users to agree on a common vocabulary and collaborate in the creation of an entity that can then be exported to the MDM solution. A data quality analyst or developer could then leverage SAS Data Quality or SAS Master Data Management to import that entity, profile sources, add technical information like the size and data type of the field and augment the matching rules – all while creating that entity definition. What this means is more IT and business collaboration and concrete technical artifacts output from a business process that feeds the front-end of your MDM or data quality initiative.

In my next blog post, I will explore the SAS DataFlux Master Data Management Foundations technology, a component of several of our information management offerings, and how it can assist with this phased approach to MDM. Master Data Management Foundations is unique in the marketplace in that it provides that transition between more short-term tactical data quality challenges (for example, de-duplicating a mailing list) and longer-term strategic MDM initiatives. Stay tuned, and have a happy 2013.

Post a Comment

Baking it in: Change management as part of data governance

Analise Polsky
Analise Polsky|November 28, 2012

If people are the heartbeat of an organization, data is the blood that moves with each pulse. When everything is flowing smoothly the organization thrives. Unfortunately, it is not hard to imagine instances where something malfunctions. Little by little vital components will slow, break, or shut down.

Is there a way to rejuvenate and revitalize the system? The answer is yes! Half the remedy is change management, and the other half is data governance.

Change management and data governance share similar characteristics, but they are definitely not the same. Change management helps solve business issues by aligning both people and processes to strategic initiatives that will help an organization achieve its business vision. Data governance is the oversight of the enterprise data which drives the business. It encompasses the business framework, the processes and policies surrounding the data, and the day-to-day management of that data.

Let’s say your organization recognizes that it has data quality issues and wants to establish data governance. Often, organizations struggle to kick start the initiative and employees are reluctant to participate. (If this sounds familiar, you’re not alone.) Applying the tenets of organizational change management, you can ensure understanding and buy-in of a nascent data governance project.

First, you need to plan: identify the “pain point” or problem area in your organization that is driving the need for data governance. Depending on the scope of that problem, you may carve out a subset—what we call the Small, Controlled Project (SCP)—that serves as your initial data governance initiative.

The SCP is built on achievable, incremental goals. Well-planned SCPs will provide verifiable results and repeatable processes that will help build the momentum necessary to establish and expand data governance within your organization. When planning for your SCP, seek the feedback of stakeholders, outliers, and possible saboteurs. Allowing room for feedback and input will increase employee buy-in.

Second, analyze the environment. Consider which metrics you will report back to executives and assess possible cultural challenges or constraints. From a cultural perspective, is your organization more top-down (executive-driven) or bottom-up (accustomed to more grassroots efforts)? How will this affect the decision making surrounding your SCP? Use the metrics and cultural constraints you identify to draft a delivery model. Make sure to engage business and IT in the process. If possible, empower a data steward to mediate the delivery model conversation.

Third, you need to enlist the people involved or affected by the initiative. Diverse communication channels will keep information circulating in a consistent and appealing manner. Think of alternative ways to reach out to both enthusiastic and hesitant team members.

For example, instead of a standard report based meeting or email chain, you could schedule a lunch-and-learn with business and IT members of the SCP team. Maybe there are skills that both groups need to develop around data quality that will help achieve the goal(s) of the SCP. Bringing the two groups together to focus on learning in an informal way will get everyone working toward a common goal, establish a working dialog, build trust, and promote buy-in. Uncertainty and confusion will breed frustration and disillusionment. A confused mind says NO.

Fourth, you will need to measure progress and success. Evaluate the process and figure out which components succeeded and which did not. Be ready to provide recommendations and next steps to your executives. What other business goals can be achieved data governance initiatives? Don’t forget to spread the word of your success to the rest of the organization and use your win to conquer any remaining naysayers. If your SCP is successful, the stakeholders you enlisted will help you apply the new data governance processes to other business programs.

Think of it this way: change management is like the daily vitamin of your organization. You don’t always want to take it – unless it’s the gummy kind…those are fantastic – but deep down you know it is good for you. Data governance is your daily dose of caffeine. It wakes you up, keeps you going, and without it you will have a major headache. Taken together they can give your organization a longer, healthier life.

 

Post a Comment

Data federation provides option for security and compliance efforts

Madhu Nair
Madhu Nair|November 5, 2012

This is the third in a series of posts about the topic of data federation. Click here for the full series.

In the previous post, we examined how data federation offers a faster, more agile way to bring together data from disparate sources. Another interesting use case for data federation is security and compliance.

Today’s organizations must balance regulations and data privacy issues with the need for business decision makers to access, analyze and explore data. How do you enforce complex and layered compliance efforts in an environment where the underlying data – and the users who need access to the data – have different levels of sensitivity that are always changing?

Consider the scenario of my previous post. The organization used data federation to get a new, accurate view of customers following an acquisition. With that unified view, the company gained critical insight into the newly expanded customer base that could help business staff make quick decisions on marketing to all existing customers.

However, there are likely many details from the customer records the marketing department should not be able to view (like marital status, personal bank details, and other sensitive information). Similarly, the finance department might want access to all the billing details of the customers, but they are not authorized to view other details about the customers. Neither of these departments could have access to highly sensitive information like Social Security Numbers or PINs.

Now, just imagine the scenario as you add different types of users – and the different degrees of personal information. As the number of underlying data sources and users increases, this environment can quickly get unmanageable.

Data federation can help manage this scenario by providing a robust security model, mirroring the security controls already in place. Through proper application of these security models, data federation can enforce access rights on a number of levels. Information can be masked or made available based on a user’s profile, the source of the underlying information or even at the row and column level. More importantly, when security is applied at the virtual data layer, there is now a single point of administration where access can be tracked and audited.

Data federation in this instance allows IT to enforce business requirements for security. In this way, it helps gain quick wins by integrating data together to solve a business need. But it can also ensure that this is done in a compliant and secure manner that can scale across the enterprise.

In the next post, we shall explore how data federation fits in with the rest of an organization’s data management investments.

Post a Comment