Where there's OLTP, can OLAP be far behind? It shouldn't be. In fact, the latest
generation of multidimensional analytical software is a great reason to revisit
our database and data-warehouse clients. Here's why and how.
Expert Advice For Analysis Tool Success
Target the nontechnical.
Datawarehouse tools speak to a different customer set than relational technologies
namely, those who appreciate complex data modeling functionality, such as top
corporate executives and business analysts.
"Integrators who try to sell warehousing to IT audiences will fail;"
warns Holly Rader, product manager for data warehousing at IBM. "Target
the functional executives who have specific business problems that these products
Speak the language of business.
Traditional OLTP system customers are IT managers, but the purchase decision
makers of OLAP and other front end data warehouse tools are business analysts,
marketing people, and corporate executives.
"Selling these folks means talking about business problems and how the
software will solve those problems," says Neil Mendelson, director of data
warehousing at Oracle. "The biggest mistake integrators make is to talk
Use existing database systems as the reason to call.
OLAP, ROLAP, and multidimensional analysis tools extend the power of existing
OLTP and other relational database systems. Many tool packages include their
own internal data cubes that let them cull data and create their own internal
warehouses to enable fast multidimensional analysis.
Present clear ROI benefits but tactfully. Multidimensional analysis can provide
clear ROI benefits. But vertical customers will want to crunch their data differently.
Learn the data models that may improve bottom lines in the 1 to 2 percent range
it takes for the data warehouse to pay for itself. But "don't be too quick
to tell your customers how these new tools can best be used," cautions Clay
Hardin, decision support manager at integrator Radiant.
Seed the client, then wait for growth.
OLAP and other data-analysis packages give users sophisticated decision making
capability they're not used to. In soma cases they'll start getting today's
information today, not next week. In other cases they'll get brand new information.
"We've learned to install the tools quickly, provide a few pre established
models, then sit back and give the client a few weeks to try them out with key
employees," Hardin says. "Then we go back with a company-wide roll
out plan. Deployment is then usually much larger and more on target than if
we'd speced it out without the trial."