When considering strategic business intelligence needs, would you put your faith in the cloud? The idea made sense for Distribution Market Advantage. For Creativity Inc., not so much.
The profiles of these two companies draw a clear line between where the nascent BI-as-a-service offerings fit — and don’t fit — in business today. At one end of the spectrum, Schaumburg, Ill.-based DMA has no in-house BI expertise and needed to rapidly develop a Web-accessible BI dashboard for a narrowly defined purpose. On the other, Van Nuys, Calif.-based Creativity has developed its own in-house data warehouse and business analytics expertise. It mines that data to develop highly customized metrics that provide a competitive edge in developing and marketing new products for the consumer crafts market.
But even Creativity’s vice president of IT, Jim Mulholland, who decided to pass on BI software as a service (SaaS), sees the game-changing potential of BI as a service: its ability to rapidly deliver a standardized suite of analytics tools that give users most of what they need without the time, expense and hassle of developing a BI infrastructure internally. BI in the cloud could be “the next killer SaaS application,” he says.
However, the technologies — and the business models behind them — are still evolving. “It’s still an embryonic market,” says Jeffrey Kaplan, managing director for on-demand services consultancy ThinkStrategies Inc.
BI-as-a-service offerings typically import business data in a common format (such as an XML or comma-delimited file) put a structure around it, apply the appropriate data models and generate a Web-based user interface that allows for some analysis and the creation and distribution of standardized reports and dashboards. Some services can also query data in place, either behind the corporate firewall or from other SaaS applications, such as Salesforce.com’s system. And some providers offer professional services, such as data integration, ETL (extract, transform and load) and data transformation services that organize, clean and normalize data for organizations that can’t do it themselves.
DMA needed those services. The supply chain logistics organization supports 50 regional food service distributors and has no in-house BI capabilities. “We are not a technology company. Our core competency is supply chain management,” says Jim Szatkowski, vice president of technical and data services. With no in-house expertise, putting BI in the cloud made sense. But DMA’s needs also fit the hosted model in two other ways: BI-as-a-service offerings tend to play well with other popular SaaS products, and they often have easy-to-use Web-based interfaces that facilitate collaboration with entities beyond the corporate firewall.
BI SaaS at a Glance
What You’ll Pay
- Most customers spend US$20,000 to $50,000 annually, says Brad Peters, CEO of Birst. PivotLink charges $3,000 per month for 100 million rows of data and 50 users. But users can get a small pilot project started for $100 or less per month.
Consider BI SaaS If…
- You don’t have in-house resources to do the job or BI isn’t a core competency.
- You need something that can be built, adjusted and scaled quickly, such as customizable sales reports.
- You need it fast.
- You can get 80 per cent of what you need with a few customizable templates.
- You’re already using SaaS in operational areas such as CRM and HR and need to add an analytics component.
Think Twice If…
- You’re uncomfortable processing business intelligence data outside of the corporate firewall, regardless of security assurances.
- You have a complex project or one that requires a high degree of customization.
- You have a very large data set.
- The data on which you need to perform analytics changes every day.
- BI tools will be used primarily within the organization.
- It doesn’t fit the broader business model or culture. Balancing departmental and enterprise needs is key. A myopic view of BI needs can lead to application-specific silos of BI data that might be difficult to integrate in the future. Services might not have an API or support standards that would allow you to easily bring data back in-house.
How to Get Started
- Consider a small pilot project. Service providers offer low-cost or even free trials where you can take advantage of a limited set of capabilities and reports for a small number of users.
DMA wanted to roll up invoice transactions, inventory and other data that its distributors log in a SaaS order entry application, and provide a dashboard through which each could analyze operational metrics and create forecasts. It used a service from PivotLink Corp., and Szatkowski says he had it up and running in about two weeks — a far shorter time than what would typically be required for a similar, in-house project. PivotLink’s service restricted access, allowing each distributor to see only its own organization’s data.
Evolving Models
Even when resources are available in-house, BI as a service may be preferable when time to market is an issue. RBC Wealth Management already had a data warehouse and had BusinessObjects expertise in-house, but its IT organization had a three-year backlog of mergers-and-acquisitions work. So Shawn Spott, vice president and manager of marketing research and strategic analysis, hired SaaS provider Birst Inc. to deliver a BI dashboard to RBC’s 2,300 brokers. Although the tool is still being rolled out, Spott says the company has already seen “an appreciable increase in revenue” from its users. The project worked in part because RBC had an extremely focused goal in mind, he says.
But most organizations are far from sold on the idea of hosted BI. BI as a service is still a nascent market — less than 10 per cent of enterprisesare using the services today, according to Gartner Inc. (A recent survey of Computerworld readers put that figure at about 8 per cent.)
Template-based functions and a shared, multitenant architecture on the back end are what make BI SaaS economical and easy to deploy. But such services typically can’t handle as much complexity or customization as in-house projects can. Nonetheless, business decision-makers will trade complexity for simplification if it means faster time to market and greater utilization of their applications, suggests Kaplan.
The top concerns for business users are security, availability and the potential for bandwidth bottlenecks when transporting data. While vendors have made progress in these areas, particularly with security, Kaplan says, none of the concerns has been fully resolved.
At a minimum, the SaaS provider should be compliant with Statement on Auditing Standards (SAS) No. 70, which, among other things, establishes processes and procedures for proper security when using third-party service organizations, Kaplan says.
Users are also concerned about whether BI services will meet promised performance and availability levels, since they run in multitenant environments. Most vendors say they will provide service-level assurances, but the key is to have a detailed, measurable service-level agreement. “If your SLA is not well defined, you’re probably heading into trouble,” says Bill Hostmann, an analyst at Gartner.
At Allstate Insurance Co., Anthony Abbattista, vice president of technology solutions, oversaw the build-out of a sophisticated data warehouse and self-service BI infrastructure. Hosted solutions are “pretty cool,” he says, referring to them as “good-enough BI.” But he cautions that the key to good BI lies in how you pull the data together, linking schemas, tools and access strategies. “Just because you load the data into someone’s BI environment doesn’t mean you get value from it,” he says.
At DMA, Szatkowski has no regrets. “We’re getting tremendous results,” he says. “It’s a workhorse for us.”