Recently we were intrigued by the Network World article "Top 10 tech leaders," which discussed the results of a survey of 600 IT executives who were asked for their perceptions of various attributes of 30 major IT vendors. Attributes included the superiority of the vendor's executive management and its leadership qualities, technology vision and role as a strategic supplier. The survey also gauged buying intentions.

The results were interesting. IBM was perceived as being tops in executive management, leadership qualities and technology vision, while Cisco was the favorite in key technology leadership and as a strategic supplier. Microsoft won out in buying intentions.

After reading the article, we started wondering whether there were correlations among these attributes. To find that out, the suitable tools seemed to be Microsoft Excel's PivotTable and PivotChart.

So, we tabulated the data with the vendors' names in the first column and the attribute values in subsequent columns, and put the score data into the body. We then selected the entire table and made a PivotChart. Seemed like a good idea, but to make a long story short, it turns out PivotTables and PivotCharts aren't suitable tools for this endeavor. You can't even get as far as making a scatter pot.

Thus it was that we turned to Tableau Professional Edition Version 2.1, published by Tableau Software. This is one of the finest applications we've seen this year. It is amazingly fast; has a well-designed user interface that is remarkably intuitive; and, considering its complexity, it is stunningly bug-free!

After opening a new Tableau workbook, you connect to a data source. Your choices for the Professional Edition are Microsoft Excel, Microsoft Access, text files, MySQL, MS SQL Server, MS Analysis Services, Oracle, Hyperion Essbase and IBM DB2 OLAP Server.

Once you have opened a source, the data will be split by default into dimensions (fields containing qualitative, categorical information) and measures (fields containing numeric or quantitative data). Measures are used as axes for the rows or columns in a table, and dimensions create headers. (Note that these distinctions are far more complex when you are dealing with relational databases where you can convert from measures to dimensions, which in turn can be converted into continuous quantities.)

In our example, opening the survey data spreadsheet in Tableau gave us eight measures, one for each of the six attributes plus two created automatically -- Measure Values (a collection of all the measures of your data) and Number of Records. Two dimensions were created automatically also -- Vendor (the names from the Vendors columns of the original table) and Measure Names, a listing of all those names.