To clarify, Microsoft BI is by no means free in that you have to have enterprise licenses of Microsoft Office, Microsoft SQL Server, and Microsoft SharePoint. But those licenses give them access to Reporting Services, Integration Services and Analysis Services delivered through SQL Server, Office Excel as an all-purpose interface for crunching and visualizing data, and SharePoint as a place to publish and collaborate on analyses. That's Microsoft BI, and the bundling strategy has catapulted the company to the top of the BI heap (in terms of numbers of users) in little more than five years. (IDC ranks Microsoft fifth in terms of BI software license revenue through a mysterious calculation method I've stopped trying to understand.)

Many find Microsoft BI to be more like a developer's toolkit than what they want in a BI suite, since it's up to developers to build out query, reporting and application interfaces that other suites include. So needs vary and one size definitely does not fit all, but Microsoft's success shows cost matters. Want more proof? When we asked about factors driving interest in software-as-a-service/cloud-computing-based BI/analytics," the number-one answer (cited by 55%) is "low overall cost" and the number three answer (cited by 39%) is "low initial cost."

Make sure it's easy to administer. "Self-service BI" and improvements in systems admin and management have been familiar themes in plenty of recent BI product upgrades, but there are signs vendors haven't done enough. "Minimal need for IT staff/support" draws the second highest response (cited by 46% of respondents) on the list of factors driving interest in SaaS/cloud-based BI. What's more, the hegemony of corporate-standard BI appears to be slipping, with only 41% of this year's respondents agreeing that their firms have standardized on "one or a few products," down from 47% last year. That hints that departments and line-of-business units aren't waiting for IT. Instead they're embracing fleet-footed alternatives including SaaS services and hosted solutions that require minimal IT support.

[ Want more on business intelligence? Read More BI Packages Add Collaboration: Who Needs It?. ]

Here again, an ounce of research before you buy could save you a pound of support burden. Reference customers can tell you about IT staff requirements and the time and effort required to develop and deploy new applications, dashboards and reporst. You'll want to know for sure just how much self is in those self-service capabilities.

Make sure it's embeddable. It's important to remember that BI emerged as a separate category in the '80s and '90s only because the reporting and analysis capabilities built into applications were so inadequate. Separate warehouses and analytical processing power also kept mission-critical transactional systems from browning out under the added burden of analytical loads.

Times are changing, processing power is now dirt cheap, and applications vendors are increasingly embedding BI into their software. SAP is even talking about doing away with the separation between BI and apps with the aid of in-memory technology. With this move, BI blends back into the application domain.

Yes, companies tend to have many applications and data sources that might beg for an independent BI platform. Even so, organizations are looking to embed dashboards, key performance indicators and other forms or decision-support directly into their business applications and workflows. Our survey respondents rank "exception management" second and "embedded BI" fourth on a list of 11 leading-edge BI capabilities. Exception management means spotting problem transactions, workflow bottlenecks or emerging risk, so it's as much about blending of BI/analytics and applications as is embedded BI.

To support embedded BI, look for rich BI/analytics services capabilities and event-aware integration capabilities. An event might be the initiation or completion of an order, an alert that a process is taking longer than expected or a trigger when inventories have reached a critical threshold. These are capabilities that Microsoft, Oracle and SAP in particular have talked up because they bridge the gap between applications and BI.

Measure Based On Business Results.

It's time to stop measuring BI success by the seat. BI started out as a Band-Aid, and there's no reason it should be enshrined as something everybody must aspire to use. Decision support should be delivered in the context of the decision, and that might be within an enterprise application, an email inbox, a process workflow, a well-integrated dashboard, a collaborative interface, a mobile interface or even a social network interaction.

The ultimate measure of BI success is not how many people touch it, but to what degree it optimizes stocking, ordering, fulfillment, sales forecasting, days sales outstanding, financial planning, customer satisfaction and so on. Business value delivered is the difference between BI success and failure.