Beyond Statistics: SAS Takes On Mainstream BI

Mention SAS and many people think of statistics or perhaps a company renowned for a worker-friendly environment. Few associate SAS with BI, even though it has long had a presence in the data warehouse market with its ETL and technical analysis tools. Aiming at the broader BI market — and at changing prevailing perceptions — the company released the SAS Web Report Studio in 2004. The company recently repackaged its BI products into a suite, SAS Enterprise BI Server, and is more aggressively pursuing the business intelligence market. This review takes a closer look at two of the Enterprise BI Server's most significant components, SAS Web Report Studio and the SAS Add-In for Microsoft Office, features specifically targeted at mainstream business users.

Like many leading BI suites today, the SAS Enterprise BI Server spans the breadth of user needs, including OLAP, query and reporting, a portal interface for users and Microsoft Office integration. SAS, however, is different from BI heavyweights in a number of ways. First, the SAS engine doesn't rely on pure SQL as many other BI engines do and uses its own language to provide deeper analytics such as rank, forecasts and statistics — an area where SQL can come up short. Second, it can leverage established predictive models and SAS stored processes that many companies have relied on for years to run their businesses. In the past, many of these models may have been trapped on individual users' desktops. Web Report Studio and its Add-In for Microsoft Office present these powerful processes in easy-to-use tools, making them accessible to many more users.

Finally, while the industry saw a convergence of query and reporting tools in the past two years, highly formatted reporting isn't part of the SAS BI suite. Instead, the company seems to be betting that deeper analytics — moderately formatted — are what users are really after.

Spotlight on Business Users

Web Report Studio defies SAS's reputation for complexity. All reports are authored from browsers; no other program must be installed on the user's machine. Users can design their own reports or use wizards to guide them. This is SAS's first mainstream BI attempt, so I was surprised to see a few unique capabilities (even in the product's first release) that users want but competitors have been slow to provide. One such capability is report interactivity via simple mouse clicks. In Web Report Studio, a business report consumer can easily rank, re-sort and filter a report without having to enter a separate report design mode.

Users can also control how measures are aggregated without having to create a complex calculation themselves. For example, users may want to analyze revenue from different dimensions and aggregate by sum; SAS lets users specify different aggregations (such as min or max) at run time. End users will appreciate the flexibility, and the administrator can still control which aggregation techniques are permitted. Most other BI tools require an administrator to uniquely define such metrics at the metadata level, even for a one-time analysis. The aggregations go beyond standard SQL aggregations to include such choices as standard deviation, variance or probability.

Web Report Studio is easy to use, but it lacks the visual appeal found in most other BI products. As shown in the screenshot below, many actions are performed via pull-down menus rather than in the toolbar. In some cases, SAS has sacrificed capability for ease of use. For example, report authors are limited in how they filter data: Nested conditions, absent in the product's first version, are now available but still cumbersome to build. Report parameters are basic, and prompts can't be cascaded.

Data Anywhere ... Almost

Users want to analyze data no matter if it's in a data warehouse, OLAP database or departmental spreadsheet. Web Report Studio can access these data sources in multiple ways. It can access relational and OLAP data via the same user interface — something other leading BI vendors have only begun doing. Important to note, however, is that Web Report Studio and the BI Server's OLAP Viewer (see the screenshot above) can access only the SAS OLAP Server and won't work with third-party OLAP databases such as Microsoft Analysis Services or Hyperion Essbase. So while the reporting and OLAP integration is good for existing or new SAS customers, it's less relevant for deployments with a mixture of OLAP solutions.

SAS also supports multiple data sources: One report can access numerous Information Maps (SAS's term for its metadata layer). In addition, report tabs (similar to those in Excel workbooks) let users create a "briefing book" that contains all key indicators in one file. Lastly, an Information Map can access multiple data sources, including SAS OLAP Server and stored processes. For example, an Information Map can access customer classifications in a departmental Microsoft Access database, sales details in an Oracle data warehouse and a stored process that runs a complex forecasting model. This capability makes the Information Maps quite powerful. Lacking, however, is multipass SQL or the ability of one Information Map to access multiple fact tables and star schemas. This capability would let users dynamically calculate Days Sales Inventory using monthly inventory and daily sales data that reside in different fact tables. For such an analysis, users could turn to the SAS Add-In for Microsoft Office.

Excel Add-In: A Killer App

For existing SAS customers, the Add-In for Microsoft Office will easily become a must-have, but it's not currently sold separately from the BI Suite. For new BI customers, this product could tip the scales in SAS's favor. As the name implies, the add-in supports not only Excel, but also PowerPoint and Word. Reflecting its analytic roots, the add-in handles large data sets, eliminating users' concerns about Excel's native limit of 64,000 rows. The add-in leverages the SAS Enterprise BI Server so users can scroll through data chunks before bringing too much data back to Excel. Also, any data residing on the server is available for users to chart in Excel; it doesn't have to be pulled into the spreadsheet first, which lets users chart data sets with more than 64,000 data points. Furthermore, the add-in brings additional charting abilities (Pareto charts, maps and random samples, among others) into Excel through ActiveX controls.

The best thing about the add-in is that "users can ask and answer their own questions," says one SAS telecommunications customer. While most BI add-ins can only access prebuilt reports, the SAS add-in lets users create new queries off a single table or a SAS stored process. This is a differentiator, but for a broader-based solution, I'd like to see SAS also provide access either to the Information Maps or to Web Report Studio reports. SAS says it plans to add support for Information Maps in a release due out at year end. The add-in works well when accessing SAS data sets, stored processes or single tables, but without this additional integration, it's not suitable for dimensional models that require joins to dimension tables.

Hit or Miss?

The SAS solution is by no means perfect. Some of its workflows, such as adding data sources to Information Maps or deleting columns from cross tabs, are awkward. While its reporting and OLAP integration are good, the interfaces within the Enterprise BI Server can't yet access third-party OLAP databases, a limitation for a company looking for an open BI solution. Report bursting and scheduling, missing in the first release of Web Report Studio, are still relatively basic, as are report formatting options.

Despite these limitations, SAS has hit the mark on a number of key business requirements. Its subscription-based pricing model, while criticized by some analysts, seems yet another selling point for customers: It reduces the risk of shelfware, lowers the upfront investment and ensures SAS has a stake in customers' successful use of its products.

• For pricing information on Enterprise BI Server, go to www.sas.com.

Cindi Howson is the president of ASK, a BI consultancy. She teaches the Data Warehousing Institute's "Evaluating BI Toolsets" and publishes independent BI research on BIScorecard.com. Write to her at [email protected].