Earlier this year I wrote about the trend of companies supplementing their on-premise business intelligence deployments with software-as-a-service in order to fill vertical niches that their traditional software doesn't adequately address.
A recent TDWI article written by David Frankel, VP of business development for (vendor alert!) 1010data, offers some additional examples of when a SaaS delivery model for BI makes a lot of sense.
Sharing data with partners and suppliers in a traditional scenario can be complicated, for instance, due to issues with data access, governance and security. But a rules-enabled SaaS BI environment can create "a neutral haven" for sharing data, writes Frankel. An example: Retailers can offer access to point-of-sale data to their suppliers so the suppliers can compare sales of their products with those of others in the store.
SaaS is also becoming increasingly popular for analyzing data purchased from outside companies. Unlike internal data, there are few security concerns with purchased data. So companies can use SaaS for highly granular analyses of large volumes of third-party syndicated data. For example, writes Frankel, financial services companies could examine historical prepayment, default, delinquency, and loss-severity rates for different types of mortgages.
Of course, there are caveats. SaaS is not the best option when a high degree of customization is desired, and this is true of many BI deployments. SaaS is also a challenge when using BI solutions comprised of multiple tools from different vendors. However, these issues are becoming less of a problem with the increasing maturity of the market, Frankel points out.
He also offers what I think is a smart compromise for companies in industries with stringent security requirements, such as financial services or health care: semi-managed data repositories in which a company maintains the hardware in its own data center while the SaaS provider actually manages the data.