This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.
You can now buy analytics as an add-on module for your BI suite, but you'll need expertise to make the leap into predictive analytics.
It's one thing to clearly understand where your business stands today — something plenty of organizations have achieved with the help of business intelligence (BI) query, reporting and analysis tools. It's quite another thing to predict where your business is going — something far fewer organizations achieve with the aid of advanced analytics. Now mainstream BI vendors including Business Objects, Cognos, and Information Builders are hoping to tap into growing demand for prediction by introducing their own advanced analytics modules. The question is, can analytic newbies handle statistical analysis and predictive modeling without hiring experienced hands?
Just how common is analytics expertise? As one indication, the $5 billion BI market is nearly four times larger than the $1.4 billion advanced analytics market, according to IDC. Until recently, BI vendors left analytics to specialists such as SAS and SPSS, but with IDC expecting 10-percent-per-year growth in that market through 2011, there have been a spate of recent announcements:
- Business Objects, an SAP company, last week introduced the BusinessObjects Predictive Workbench, a rebranded version of the SPSS Clementine data mining workbench that has been integrated with the BusinessObjects XI platform.
- Information Builders on June 3 announced it will add an analytics module called RStat to its WebFOCUS BI platform. To be released by year end, RStat will be built on the R open source project, which is popular in academic circles and used by some one million practitioners worldwide.
- Cognos, an IBM company, announced in March a joint marketing deal under which both companies will promote existing integrations between SPSS Clementine and Cognos 8 while also adding Cognos-compatible blueprints for predictive applications.
It's not likely that organizations that are already analytics practitioners will toss out incumbent vendors, so the immediate opportunity for these BI players is to find existing customers that are interested but not yet invested in analytics. Business Objects says at least a third of its customers have expressed an interest in predictive analytics.
"This is a great time for customers to look at predictive capabilities because you have to be super efficient in times of economic pressure," says Franz Aman, VP of Business Intelligence Platform Product Marketing at Business Objects. "You have to determine your best products, spot the best customers and reduce customer churn by segmenting customers properly and having specific offers that keep those customers engaged."
But can new customers move right into prediction? Predictive analytics projects typically require two phases of development, according to David Hatch, an analyst with Aberdeen Group who wrote the May 2008 report "Predictive Analysts: The BI Crystal Ball." The first phase involves developing an understanding of the business measures that the organization is attempting to predict, such as customer churn or best cross-sell or up-sell opportunities by customer segment.
"You don't need a PhD to determine how to measure or mitigate churn," says Hatch, "but the PhD-level experience becomes more necessary as organizations enter the second phase — defining the analytical modeling and mathematics involved to accurately predict future performance."
While software is available that can automate intricate modeling processes and ease MBA-level business analysts through common predictive analyses, knowing which model to apply and understanding the mathematics involved is important, says Hatch.
"You can employ smart, automatic default settings and have the software automatically recognize some patterns and give you answers from some algorithms under the covers," agrees Mary Crissey, analytics strategist at SAS, "but there's more inquiry that only a person who has done data analysis would think of. It might be asking, 'what if I look at just this part of the data,' or 'what if I try a different break in the customer groups,' or 'what if we ask for additional information we don't currently have in the data warehouse?'"
And then there's the question as to whether business can afford and find experienced talent. Data miners are among the most highly paid IT professionals, with median incomes of around $100,000 to $110,000 in the US and Canada, according to a 2006 poll by the data mining site KDnuggets.
Crissey of SAS says "welcome to the club" to BI vendors who want to take predictive analytics mainstream, but, she adds, "it's a challenge to find advanced analytics talent. They're being called to go into retail. They're being called by the telecoms to go into cross-sell/up-sell. They're being called for life sciences… [The need for talent] is one of the reasons there's debate about immigration and work visas, and it's one of the reasons we opened an R&D shop in India."
It's clear that the need is great and analytics tools are clearly becoming more accessible and integrated with information management and BI platforms. But unless they're ready to pay for experienced talent, would-be analytics practitioners will have to learn how to crawl and then walk before they can leap into advanced analytics.
We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.