Can Software as a Service work for Business Intelligence?

April 25, 2007

Can Software as a Service work for Business Intelligence?

Software as a Service (SaaS) has become an important option for many business applications.  The model is subscription based; users don’t pay a license fee and the service provider hosts, runs, and manages the software in a one-to-many model. Usually, the software can be implemented very quickly – much faster than a purchase software package.  The model is especially appealing for small to mid-sized companies who don’t have the resident skills to run and manage software applications.  Over the past year, vendors have been offering up solutions, but can this model work for business intelligence applications? 

BI Vendors have been quick to jump on the SaaS bandwagon.  Over the past year, Business Objects, SAS, and Cognos have launched their versions of SaaS.  Business Objects provides a “BI on Demand” service via CrystalReports.com. It enables users to share their reports online.  SAS offers five specific on demand services covering business intelligence, supplier relationship management, marketing automation, marketing relationship management and anti-money laundering.

New entrants to the SaaS BI market have emerged. These include Oco and LucidEra.  Oco’s business model is to take a company’s data, and deliver access both to their large repository of pre-designed reports, and to a dashboard which monitors key data against specific thresholds. Oco claims to be able to implement this in 6 weeks. LucidEra recently entered the market with a business visibility service.  For about $3000 a month (up to 100 users and includes connectors to three systems) LucidEra offers a pre-built application called Forecast-to-Billing.  The service takes a company’s sales and financial data and enables end users to slice and dice the information in various ways and generate reports. LucidEra boasts a 2-3 week implementation period.

But does the SaaS model work well for BI applications?  We are positive about the model and it’s clear that some early adopters are giving it a whirl, but we are also wary of some issues that will inevitably arise and will need to be addressed. These include:

Data Quality.  The perennial issue of data quality must inevitably emerge in the BI SaaS model.  Some data for the service may be coming from sources such as Excel spreadsheets or small Access databases, generated by people who gave no thought to the possibility of such data being used for analysis.  So, the data might include all sorts of information, strung together in an ad-hoc fashion.  One mid sized company I was speaking with recently talked about how some of their data had name, address, and even revenue related information in the same cell of an Excel spreadsheet.  Cleaning up the data can be a time consuming nightmare.  Some vendors, like LucidEra, have thought this issue through. While they provide a level of data cleansing, they will re-direct a client to an outside provider if their data needs a lot of scrubbing. Others claim to provide a complete cleaning capability as part of their service.

Metadata Coherence. Hand-in-hand with data quality goes metadata quality. If a company runs just a few packages from a single vendor that it is possible, even likely that the metadata issue will not be onerous, but where there are multiple applications with some written in-house it is likely that there will be metadata problems – including some that will be difficult to resolve.

The Limited Nature of Solutions.  This is a blessing and a curse.  No vendor provides a full BI capability on a SaaS basis, so that small and medium sized companies will be happy to generate comprehensive reports from their information. But any needs they have beyond that will be difficult to meet. BI is an interactive iterative process.  As users happily slice and dice their information, they will discover the power of BI and may want more.  SaaS providers need to think about how they to accommodate such needs.

Solution Scalability.  The technology of the SaaS model for BI is scalable, but is the support capability scalable? With transactional systems, the level of support needed is reasonably predictable. But with BI, even if setup activity can be kept to a minimum – and this is doubtful where you have a significant data cleansing problem – the amount of help desk support that may be necessary to deliver a satisfactory service by phone (or in person, if necessary, depending on the scope) could be very high. It may be difficult to arrive at a viable monthly rental figure – especially for start-up SaaS operations. Providers need to focus heavily on maximizing the self-service aspects of their offerings.

Software as a Service (SaaS) has become an important option for many business applications.  The model is subscription based; users don’t pay a license fee and the service provider hosts, runs, and manages the software in a one-to-many model. Usually, the software can be implemented very quickly – much faster than a purchase software package.  The model is especially appealing for small to mid-sized companies who don’t have the resident skills to run and manage software applications.  Over the past year, vendors have been offering up solutions, but can this model work for business intelligence applications? 

Newsletters 2007
About Fern Halper

Leave a Reply

Your email address will not be published.