MDM without boiling the ocean

I was thinking about Master Data Management (MDM) last night. There’s been a lot of hype over the past few years about MDM. At its simplest level, MDM is about handling common data, such as customer or product data, in a way that enables disparate IT systems and business groups to reuse the data. The concept is certainly not new. I remember working at a company twenty years ago where consultants were hired to try to get a single view of certain core information – in this case a single view of business customers.

MDM is Complex

MDM is a complex undertaking that needs to be thought about at a strategic level. A complete MDM deployment is not going to happen all at once. There are numerous technical, operational, and organizational challenges that need to be addressed when implementing an MDM solution. These include data governance, other organizational issues such as resistance to change, determining the MDM architecture, dealing with data quality issues, and the list goes on. No wonder many of the larger platform companies and system integrators have gotten into the act.

 

So, given this reality is there room for smaller players, say those that have a pragmatic approach to support and supplement MDM as well as offer an alternative to the single view challenge? I think so.

 

I recently had an interesting conversation with Robert Eve, the VP of Marketing, from Composite Software. For those of you who don’t know, Composite Software provides data integration by virtualizing data silos. The software accesses data from disparate data sources and creates a virtual data layer that can be used by various enterprise applications. Composite has done a good job in transforming itself over the past few years from providing composite views to EII and now data virtualization/real time data integration.

I wanted to understand how Composite Software supports MDM. Robert and I spoke about the various approaches to MDM including custom coding, ETL, messaging, and data virtualization. Composite provides solutions in the last category and it has positioned its software to support MDM initiatives in three ways:

  • Creating “multiple” single views, combining how ever many columns in the core data master hub with potentially thousands of columns of related data, depending on depending on “user” role and “business” requirement.
  • Helping to create the hub via better data access, especially from complex systems such as SAP or Siebel.
  • Enable data quality processes by teaming with data quality vendors.

MDM light?

While Composite is positioning itself as a complement to MDM. it does provide what can be termed “MDM light”. A popular use case (sans MDM) for the company is creating a single view of something across several data silos. This single view might be a single view of a physician, sales manager, or employee. While this isn’t an MDM hub with all of the bells and whistles it does provide a practical way for companies to get at the single view issue on a smaller scale.

Innovations in Data Visualization – Visual analytics

A wise man once told me, “Look at the data. What are the data telling you?” That was my dissertation advisor, some twenty years ago, before the term data visualization was even coined. And that’s the sensible advice I’ve followed throughout my career when analyzing all different kinds of data.

Data visualization in the form of slicing and dicing, charting and pivoting is standard for most knowledge workers performing data analysis. BI vendors provide visualization in the form of charts and tables of data cut different ways. Microsoft provides its ever-popular pivot table, but dealing with the data can be cumbersome, especially if you want to explore the data quickly across multiple dimensions.

Marcia Kaufman and I recently got a chance to meet with Christian Chabot, CEO and co-founder and Elissa Fink, VP of Marketing from Tableau Software, Seattle, Washington. They impressed us both with Tableau’s innovations in data visualization.

The visualization is the query

So, what’s so interesting about Tableau’s approach?

Consider the following typical analysis problem. You are trying to analyze sales for different categories of TV sets at ten different store locations for the first half of the year. Data include location, region, TV type (flat panel LCD, flat panel plasma, LCD projection, etc.) date sold, dollar value, sales person, as well as information about promotions and warranty plans. If you used a pivot table in Microsoft Excel, you could cross-tab and slice and dice information, you could even drag and drop various attributes onto a chart. At the end of the day, however, you are still left looking at a two-dimensional static plot, or (more likely) a bunch of static plots, trying to derive insight.

With Tableau, it’s not about slotting the data into a plot or report to examine; it’s about rapid visual analysis of the data.

Tableau reads in structured data from many sources such as Excel, Access, text files, SQL Server, Oracle, DB2, MySQL, PostgreSQL, Firebird, Netezza, SQL Server Analysis Services, and Hyperion Essbase. The columns in an Excel spreadsheet would be read into Tableau and put into something it calls a dimension (non numeric) or a metric (numeric) that are listed on the left hand side of the Tableau screen. The user then simply drags and drops as many of these dimensions and metrics as desired onto the palette and the visual representation of the data changes.

In the example above, you might first start an analysis looking at sales of TVs by category and by region.

TVs by Category by Region

Then you might drag in another “column” that further breaks this down into the type of TV in each of the categories such as flat panel LCD, flat panel plasma, etc. onto the palette. This changes the visualization to include this additional dimension.

tvs-by-category-region-and-type.png

 

 

 

By interacting with the visual in this manner, the user is querying the visual. The product makes it easy to look at the data dynamically from all different angles, thereby enabling rapid analysis and discovery.

Here are a few of the features that make the analysis quick:

  • The product makes good use of color, so for example, losses would be shown in red. There are also very nice graphical representations to work with.
  • If the underlying data permit, Tableau lets users look across any time dimension (daily, weekly, monthly, quarterly, yearly) with a simple click of a drop down menu.
  • If you don’t want a particular time dimension included in the analysis, simply select and remove it and the visual changes.
  • Tableau lets the user drill down into the visual, to see the underlying data.

The product is flexible and extremely easy to use. It’s also visually appealing – the company definitely practices what it preaches. The charts are clean and crisp and there is good use of color. The latest version of Tableau (3.5) also includes Tableau Server, a Web-based sharing and publishing solution that enablers users to share their results with others. The Personal Edition is a visual analysis and reporting solution for data stored in Excel, MS Access or Text Files with a price tag of $999.00. It’s worth looking in to.

Follow

Get every new post delivered to your Inbox.

Join 1,190 other followers