Insights Are the Meat; Data Is the Relish
The problem with so many BI tools, says Chasney, is that they're no different from the standard corporate reporting tools that have been around for years, which churn out old data like curdled butter and don't provide information that executives can chew on. If companies really want to get value from BI, he says, they need a system that provides them with insights, not just mountains of data. "There's nothing worse, in my opinion, than a business intelligence system that reports changes on a weekly basis," he says, because those systems don't provide any context as to what factors are influencing those changes. Without that context, you don't know whether the data is good or bad; it's just useless.
When charting a course for BI, Chasney advises companies to first analyze the way they make decisions and to consider the information that executives need to facilitate more confident and more rapid decision making, as well as how they'd like that information presented to them (for example, as a report, a chart, online, hard copy). Discussions of decision making will drive what information companies need to collect, analyze and publish in their BI systems.
When Chasney started building CPR in 2000, he asked the company's CEO and the chief operating officers of CKE's three restaurant chains - Hardee's, Carl's Jr and La Salsa Fresh Mexican Grill - what information is most important in their efforts to run the company. The CEO wanted to know what caused changes in sales. The COOs wanted something that would indicate business opportunities they could pursue as well as clear indicators as to which restaurants were underperforming. The discussions taught Chasney that BI systems need to focus on a company's most important performance indicators - including sales and cost of sales; exceptions, such as those areas of the business that are outperforming or underperforming other segments; and historical and forward-looking business trends - if they're to provide the company with any value.
Good BI systems also need to give context. It's not enough that they report sales were X yesterday and Y a year ago that same day, says Chasney. They need to explain what factors influencing the business caused sales to be X one day and Y on the same date the previous year. CPR uses econometric models, which the company reviews and refines each month, to provide context and to explain performance. The econometric models take into consideration 44 factors, including the weather, holidays, coupon activity, discounting, free giveaways and new products. If the CEO wants to find out why sales were down on any given day at Hardee's, all he has to do is click the "explain" button on his computer screen, and the model performs its magic. The CEO will see, for example, that 5 percent of the 8 percent decrease was due to torrential rain in the US Northeast and 2 percent was due to free giveaways.
"If your business intelligence system is not going to improve your decision making and find problem areas to correct and new directions to take, nobody's going to bother to look at it," says Chasney.
Start with the Freshest Ingredients
The key to getting accurate insights from BI systems is standard data. "Data quality remains a very overlooked issue in business intelligence, but a massive one," says Gartner's Friedman. "I continue to see failures due to a lack of attention to data quality." Data is the most fundamental component of any BI endeavour. It's the building blocks for insight. Companies have to get their data stores and data warehouses in good working order before they can begin extracting and acting on insights. If not, they'll be operating based on flawed information.
Ruby Tuesday's Ibrahim advises companies to develop plans that outline what they're going to do with data once they get it, practices for preventing redundant data and methods for organizing it in a way that makes sense to the business. For instance, Ruby Tuesday organizes its data around three categories - sales, labour and food costs - that happen to be the key drivers of its business. Those three categories are tracked in a database and put into separate table spaces for ease of reporting and processing, Ibrahim says. That way, information on what products are selling does not get mixed up with information on labour and vice versa.
Knowing that the key to using information to improve decision making is ensuring that the transactional data collected at the point of sale is consistent and accurate, Ibrahim standardized all of the company's restaurants (700 at the time), including those run by franchisees, on a common technology platform in 2001. He also moved the company onto a Microsoft SQL server and open-architecture databases, which makes it easier for business analysts to get to the data they need. The open architecture lets analysts run specific queries against databases when they're looking to find out, say, how many margaritas the company sold on Cinco de Mayo, rather than having to sift through mountains of data to get the answer.
Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.