Packaged Analytic Applications: Can We Make Them Work?
Kalido last week announced an all-in-one analytic system—from hardware to software to services—targeted at the pharmaceutical and insurance industries. This newfangled packaged analytics application got me thinking about the long quest by BI vendors to become “the SAP of BI.” Let me explain.
SAP became a $16 billion company by offering packaged transactional applications to replace legions of outdated, non-integrated, homegrown applications. For the past 15 years, many BI vendors have believed that a similar opportunity exists to sell packaged analytical applications to replace antiquated, homegrown reporting systems or consolidate a myriad of legacy BI tools and reports. The idea of racking up billions selling cookie cutter applications had an irresistible appeal.
But like Odysseus and the Sirens, the lure of packaged analytic applications has caused a lot of shipwrecks and near misses. The problem is that packaging analytics is like trying to grab a greased pig. Every customer sources data from different types of systems and each thinks that they have unique processes which require customizing the default application model and reports.
Very quickly, these “load and go” analytic packages morph into colossal custom consulting projects. At the end of the process, the customer realizes that it would have been cheaper and quicker to build the application from scratch. And to add insult to injury, the customer has no clear path for upgrading their customizations to the next version of the package.
On the vendor side, it takes deep pockets to support a packaged analytics strategy. You need to hire domain experts to flesh out a credible business model and reports. Then, you need to build a suite of cross-departmental applications because few companies will commit to one packaged application while other its others are custom built. Then, you need to verticalize the functional applications to ensure the models and reports cleave as close to customer requirements as possible without customization. You also have to port the applications to run on multiple platforms and support multiple languages. Finally, you need to provide predefined ETL mappings between all major source systems and applications and your target data models, and keep all the mappings up to date with each new release of the source system or application. Phew!!
The Few and the Many
Few vendors have the appetite for all this. To date, Oracle leads the pack, having made a genuine commitment to expand the packaged analytic applications it inherited from Siebel Systems. It now offers more than 50 applications across seven functional areas and is in the process of verticalizing them. It has more than 2,000 customers and generates more than $100 million a year from the business.
SAS also has made a big commitment to packaged analytic applications, offering a variety of solutions across dozens of industries that embed complex analytics, including fraud, anti-money laundering, supply chain intelligence, sustainability metrics, and so on. Chief technology officer Keith Collins told me recently that SAS probably couldn’t have made the broad and sustained commitment to offering packaged analytic applications if it was not a private company that had the luxury of making long-term investments without short-term payback. Part of the problem with domain-specific packages is that they narrow the market of potential customers, which extends the payback period significantly, he commented.
IBM Cognos has focused on solving the customization conundrum by creating model-driven applications that can be modified via metadata. However, it only offers a handful of these adaptable applications. Teradata also offers packaged analytic applications for CRM and supply chain. And of course, SAP makes all of its transaction content available as data marts, but I suspect most of these are used largely for operational reporting and aren’t easily extendable to non-SAP data sources.
The Newest Entrant – Kalido
Despite the odds, hope springs eternal. Kalido's announcement last week of KONA, an all-in-one packaged analytic applications priced at $450,000 that comes bundled with Kalido’s data warehousing software, BI tools, reports and dashboards, hardware, and services, has breathed new life into the notion of a packaged analytic application.
By embedding the application inside a data warehousing appliance, Kalido offers a new twist on the old story. It's goal is to simplify and speed the deployment of data mart applications and minimize the risk by offering everything you need at a fixed price with consulting services to ensure rapid and successful deployment.
Specifically, KONA embeds Kalido’s data warehousing and MDM engine and QlikTech’s QlikView BI tools within Netezza’s new TwinFin appliance. It offers canned solutions (i.e. ETL, reports, and dashboards) for two vertical markets: sales and marketing in the pharmaceutical industry and sales and distribution in the insurance industry. It will be interesting to see how quickly Kalido can ramp up sales.
The sweet spot for packaged analytic applications is small- and medium-sized businesses and departments in large companies with little to no IT support. With business increasingly beating the “faster, better, cheaper” mantra, the time may finally be right for the era of packaged analytic applications. Certainly, a few veteran BI vendors and some upstarts are determined to get the formula correct.
Posted by Wayne Eckerson on September 14, 2009