Every IT project requires some level of input and backing from the business side of the organization, but some are more difficult than others.
Data modeling is amongst the more difficult.
"That's because it's rare that a business person -- even one with technical savvy -- would need to access application data directly and without help from a technical person," said Philip Russom, senior manager of research with the Data Warehousing Institute. "In fact, with most normalized data models, the assumption is that no human will ever see the data model, except the original application programmer."
Yet Russom and other data modelers agree that business users often need to be convinced that data modeling is important and something they should care about.
Dave Hay, senior consultant with the Cutter Consortium and author of Data Model Patterns: Conventions of Thought, is a data modeling consultant often brought in when corporate systems aren't living up to expectations -- when a data warehouse can't provide the right consolidated sales report, for example. It's not business executives bringing him in. IT needs his help to speak with the business users in their company.
"Now, as often as not, I get hired by IT departments," Hay said. "They have made themselves so
unpopular, [business users] don't want to hear from a data modeler.
That's a serious flaw. The more businesspeople understand the nature of the data, the more they will be in a position to deal with the IT department."
Business starting the data modeling conversation
Yet, increasingly, business users are taking a greater interest in data models. They're starting to realize that proper modeling can improve data quality and save hundreds of thousands of dollars on all those catalogues going out to the same address multiple times -- or that data modeling can save them from jail.
"With 'audit paranoia' at an all-time high due to SOX and the Patriot Act, businesspeople -- even high-placed executives -- are checking their data sources themselves far more than in previous times," Russom said. "An increasing trend is to keep audit trail information in the data warehouse, so such users can get a lot of detail about which source systems analytic data came from and how it was transformed."
Russom works with analytic data models that are optimized to get data out of the data warehouse at a high speed, rather than operational data models that are optimized for getting data into a system quickly with high data integrity. Business users are affected more by analytic data models, he said.
"More and more, businesspeople are practicing 'self -service' BI [business intelligence], where they create some or all of their reports instead of depending so heavily on a technical person to do all this work," Russom said. "For businesspeople -- even technical ones -- to do this, the data model of a data warehouse or similar database, like a data mart or operational data store, must be relatively simple and have structural names that describe the data they represent with intuitive, business-friendly terms -- like Financial Customer Identification instead of FCID07."
As compliance has forced business users to take greater accountability for their data, they're showing a greater interest in where it's coming from as well.
"Even when businesspeople don't want or need to create their own reports, they often need to check where the data of a report came from before basing an important decision on the data," Russom said. "In these cases, they may explore the data warehouse to answer this common question."
There's another, more straightforward reason businesspeople should care about data modeling. They can learn from it.
"When I do data modeling, I learn things about companies that the people in the company didn't know," Hay said. "The term data modeling means a bunch of different things. When I learned it 20 years ago, it meant modeling the semantics of the business. What is its fundamental structure, the things of significance? We were basically modeling the business."
Rethinking the way the business operates and uses data gives business users a chance to re-evaluate the organization. Hay conducts lengthy interviews with business leaders to understand what they're doing and how to adapt the data model. In one instance, he had been working with a glass company for only two months but was able to discover things others in the industry hadn't, he said. The client contact even asked him how long he'd been in the glass business.
"The idea is to back up and talk to businesspeople," he said. "Only they can tell you the nature of what they do and what they need to do. Then you present it back in a data model to say whether we captured or not."
Speaking their language
But that requires speaking to businesspeople "in their own language," often a difficult task for IT. Terry Halpin, a professor and vice president of conceptual modeling at Neumont University in South Jordan, Utah, uses a fact-oriented approach to data modeling called object role modeling, presenting the world in terms of objects (facts or values) that play roles (parts in relationships).
"Businesspeople really like working with a fact-oriented approach because it's immediately understandable to them," Halpin said. "They feel like they're in control."
There are data modeling tools available to help put the concept in perspective for business users. Donna Burbank, director of enterprise modeling and architecture solutions with San Francisco-based Embarcadero Technologies, which sells these kinds of tools, said they are often brought into an organization by data architects or data analysts who have some understanding of business processes but still have difficulty explaining data models. Tools like wikis, business rules detailed in a spreadsheet or PowerPoint-like forms make the conversation easier, she said.
"Talk to businesspeople like businesspeople; don't show them the data model," Burbank said. "Make it quick for them and make it something they want to see. We need to get the business users; they're the ones that drive how your data model is designed. There's a horrible techie habit of 'I'm smart and I understand; you should understand.'"