It took seven months and countless hand-coded customized scripts, but Foxwoods Resort Casino is finally starting...
to reap the benefits of its new data warehouse appliance.
Still, the work is far from finished, and the Ledyard, Conn.-based casino is simultaneously updating its Oracle data warehouse with data quality and matching tools to improve its customer profiles.
Foxwoods, which for years had been collecting and analyzing player gaming data in an IBM Red Brick data warehouse, wanted to add hotel and retail data to the mix to improve its marketing analytics and campaign management capabilities, according Michael Kutia, director of hospitality systems at Foxwoods.
When IBM announced in 2006 that it would no longer make enhancements to Red Brick, the casino decided to take the opportunity to make a change.
Kutia formed a steering committee, and the group ultimately decided on a Netezza data warehouse appliance, in part because it supported SAS Institute analytics tools, which Foxwoods finance and marketing workers were already familiar with. Kutia was also impressed by the appliance's speed at producing reports, its ability to create even more refined data marts, and its data compression capabilities, "which allow you to double the amount of data you can put in it."
Making the choice to go with Marlborough, Mass.-based Netezza turned out to be the simple part, however. Formatting and loading the data into the appliance was not as easy.
"We looked at how the data was going into Red Brick, and we knew we had to make some changes," Kutia said. "Part of the reason was we had a new MGM property" – Foxwoods opened the new MGM Grand Casino this past May – "and that wasn't part of the data warehouse."
"[We had to write] customized scripts to load the data into the warehouse," he said, because it was in a different format in the Red Brick warehouse. "It wasn't a normal ETL process."
In fact, loading just the casino's player gaming data into the new data warehouse appliance took Foxwoods' internal programmers, with the help of Netezza and Boston-based IISi consultants, more than half a year – from November 2007 to July 2008.
Separate data warehouse needed for customer profiles
Making matters more complicated, Kutia said, the data warehouse appliance -- while significantly improving the casino's marketing department's campaign management capabilities -- is less than ideal for delving into individual customer profiles.
For that, Foxwoods maintains a separate Oracle data warehouse for transactional-level data because it is significantly faster than the Netezza warehouse appliance for accessing customer data, according to Kutia.
"If I want to know everything about [a customer], I don't go to this [Netezza] data warehouse and find that, because of the way the data is stored in there," he said. "I go over to the Oracle database where we put just your customer reference stuff and look at it because it's so fast. It comes back in seconds."
Maintaining data quality and avoiding duplicate customer entries in the Oracle warehouse has been a challenge, however. "The biggest issue when you're doing something like this is a unique identifier," Kutia said.
A unique identifier is an assigned number tagged to each customer profile to help determine its identity. Foxwoods' customers are issued gaming rewards cards, making it easy to identify them at a slot machine or craps table, but the cards aren't used to make purchases at the casino's many retail shops or for checking out of its hotels. So customers are not assigned unique identifiers.
Instead, Kutia and his team are implementing Data Flux data quality tools to help the casino's Oracle data warehouse communicate and identify customers with its transactional systems. So when a person pays a hotel bill, for example, the DataFlux tools will cleanse the customer data and match it with names and other identifiers in the Oracle warehouse so the customer's transaction history is recorded accurately.
Kutia said the casino may at some point issue player cards that customers can use for all transactions, which would make it easier to maintain accurate customer profiles. That hasn't been done yet, he said, mainly because it would mean involving Foxwoods' operations staff.
"To get operations involved and explain something like this to them so they fully understand it, that's tough," Kutia said. "When you get into all that technical mumbo-jumbo, they get lost."
Hotel data, executive dashboards next
As for the Netezza appliance, with player gaming data now loaded, Foxwoods' marketing and finance departments have been using SAS Patron Value Optimization analytics tools for several months to tap the warehouse to run campaigns. "The feedback has been great," Kutia said. "People love it."
Kutia is also nearly finished transferring into the Netezza appliance three years' worth of hotel data that workers will also use to refine and analyze marketing campaigns, a process that involved custom scripts and took about a month, he said.
In 2009, Kutia plans to use the warehouse appliance to develop executive dashboards so Foxwoods' executives can monitor marketing campaigns themselves.
And despite the lengthy data-loading process and the need for a separate warehouse for customer profiles, Kutia has had few complaints from his colleagues. "In the end," he said, "if you ask anybody around here, we probably wouldn't do it any other way. It worked out great."