SEATTLE — Forget thin crust versus cheese-stuffed: the pizza wars are being waged by mining the database, and one chain is not only scoring customer loyalty, but also a return on its investment.
Pizza Hut’s customer relationship management programs are built on a data warehouse of millions of
customer records gleaned from point of sale transactions at its restaurants. They know your favourite toppings, what you ordered last and whether you like salad with your meat lover’s pie. Much of that has to do with data mining, a technology that converts details from customer data into competitive intelligence that companies use to predict trends and behaviours.
Pizza Hut installed Teradata Warehouse Miner and after a year of using it to better manage direct mail campaigns, the chain and its parent company, Yum Brands (Pizza Hut is one of eight restaurants in the group that includes Kentucky Fried Chicken and Taco Bell) are starting to see results.
The chain first got into the CRM game by using the more than 10 years of order history in its database, consisting of all purchases delivered to households that ordered by telephone.
Pizza Hut claims to have the largest fast food customer data warehouse in the world with 40 million U.S. households or between 40 and 50 per cent of the U.S market, according to Keith Jones, the man responsible for analysis and management of direct marketing analytics for Pizza Hut International. For about six years the company did analysis on a small sampling of the 40 million households it knew about, but nothing it could really call data mining, says Jones.
“”There was no prediction or analysis, really,”” he said, speaking at Partners 2003, Teradata’s user group meeting being held in Seattle last month. “”There was a lot of error in the data because people moved around a lot and the phone numbers weren’t correct. We had to clean up the data.””
The existence of duplicate households in the warehouse (same family, different phone number) made it difficult to successful target direct mail campaigns.
In the first year of using the Teradata Warehouse Miner product, Jones said Pizza Hut was able to recover the cost of licensing, integrating and training staff to use the product, and it made money for the company in the first quarter of use.
“”We made so much money this year, we’re afraid our competition will start using it,”” said Jones.
Using Warehouse Miner meant an improvement in “”household uniqueness”” from 80 per cent to 95 per cent. While direct mail has traditionally been a break-even cost centre, Pizza Hut has turned it into a profit centre.
“”We use it to do target marketing and find the best coupon offer for that household. We can segment customer households for groupings according to patterns of past buying behaviours, offer preferences and price points,”” he said. “”We can also use it to predict the success of a campaign.””
Pizza Hut now tracks not only phone orders, but online orders, and it can track in real time what commercials people are watching and responding to.
By using Teradata Warehouse Miner, all data mining functions run directly inside the database.
“”A lot of data mining can be sloppy as a result of situations where they are sampling from too small a source and it comes up with conclusions that are too preliminary,”” Jones says.
Keith Gonzales, president of El Paso, Tex.-based The Focus Group, Ltd., a firm that offers clients information on managing data warehouses, says in-database mining is good for medium and large data volumes, as it eliminates the cost of data movement and allows for a collaborative environment.
“”When you look at the needs of organizations, you can’t do (data mining) using traditional architectures,”” said Gonzales. “”You have to marry together data mining into warehouse architecture.””
He said database vendors are pushing mining into the warehouse and they are doing it so companies can exploit all the technology of data mining from the database. “”Pizza Hut is taking advantage of this,”” he said.
The added benefits of in-database mining include minimizing data redundancy, reduced proprietary data structures and simplified data and system management.
