Domino's delivers

Every day, people across the country order thousands of Domino's pizzas. Part of Taylor Bond's job is making sure all of those pies get counted properly. Bond is market analysis director at the Ann Arbor, Mich. headquarters of Domino's Pizza, and he says that though the Domino's market research department receives sales figures from 5,000 stores weekly, it has no problem manipulating the numbers-thanks to a powerful PC network.

"We have very large databases by PC standards-most of our hard drives go from 100 megabytes to 800 megabytes-and we have very, very good PCs that allow us to process data quickly and efficiently. Domino's has made sure we have the right equipment to do what we need to do. They take research seriously and they've provided us with good equipment to work with.

"We're not as mainframe-oriented as some other companies. We have mainframes here, but our market research department is more PC network-oriented. We prefer PC's because we believe they're easier to deal with; they're easier to get, and easier to maintain. If you try to make your mainframe bigger, it's an enormous expense."

In-house analysis

Unlike the research departments in other similar-sized companies, the Domino's research department handles all data analysis in-house.

"We use our suppliers to gather information for us and then we do the analysis. The sizes of the problems we tackle are probably larger than the average market research departments in other fast food companies. Domino's has a good-sized, well-qualified and experienced market research department. We can come up with decisions quickly because the information is here."

Bond says that the two most important types of information they receive from the stores are sales figures and pie counts (which indicate the quantities of each type and size of pizza sold). This information is combined with other internal tracking sources, marketing information, and syndicated sources that report on the restaurant industry, such as NPD's CREST (Consumer Reports on Eating Share Trends), to answer the many questions the department is asked by management.

Field information

Key to that process, Bond says, is information from the field, supplied by everyone from store managers to regional managers.

"They may not know as much about how their trends are changing, but they know more about what's going on in the store, and that's the part that we don't have available. We use the statistical information and combine it with what we know is actually going on in the field, because you can't just churn out a bunch of numbers and say, 'This is what's happening.'

"If you're asking questions at the regional level, you want to find out what the region was doing to promote the product, because if you're trying to find out how effective a promotion was, you want to be aware of what differences exist across the country. For example, I always try to get a feel for whether or not they were doing local marketing to supplement the national effort."

75 variables

Stores are tracked by 75 variables, including average weekly sales, location, when it opened, whether it is corporate owned or a franchise, etc.

"When I'm doing an analysis, I may code the stores for various pieces of information that will help me describe each store individually. The variables are designed to sift the information, because sales figures alone mean nothing; you have to explain them."

Pie count, for example, is very important because it can indicate shifts in customer preference-wanted and unwanted. "How many you sell and what you sell makes a big difference. If you're promoting a particular size of pizza you want to know if people are buying it, and if so, where are they coming from? If you promote a small pizza, are you pulling away your large pizza. eaters? And if so, are you pulling away too many?"

The kinds of questions the Domino's research department is called upon to answer typically deal with sales on a national level, the effects of policy decisions, and results from test markets, and the success of new promotions.

"One of the ways we evaluate the success or failure of a sales promotion involves looking at sales prior to, during, and after the promotion. We look at what sales are versus a year ago, versus short-term trends, and those weeks immediately prior to the promotion."

Sales increases are examined in relation to goals set-did they meet or exceed the goals? Did the promotion attract new customers? Competitors' promotional efforts during those time periods are also examined.

For example, last May Domino's introduced pan pizza nationally. Bond says that prior to the introduction a substantial amount of research was done to test its viability, looking at two key areas: the competition and consumer demand.

"We wanted to find out if, by introducing pan, we would be taking our own customers and converting them to pan pizza from other products. Or would we be attracting new customers? We found that after the introduction, we were attracting new customers; there were a lot of people trying our product who have continued to eat Domino's pan pizza. So from that standpoint, it's been a success."

An analysis of sales figures immediately before the national introduction and for a six-week period afterward show that pan pizza is also a success in terms of overall sales increases.

Consider promotions

When analyzing sales figures, Bond says it's also important to take into consideration any promotions your own company might have been doing during the same time period.

"If you run several promotions together it's difficult sometimes to get a clear picture of 'before.' If you have a clean prior period where nothing else was going on promotionally then you can get a better read on the success or failure of the promotion than if you're running two promotions with only one or two weeks between them.

"We usually use a four to six week prior period and the importance of the 'after period' depends on what the promotion was designed to do. For example, with pan pizza the after period is very important because the idea was to introduce the product, sustain the sales increases, and to increase our customer base and maintain it, and we have done that."

Specific software needs

In performing many specialized analysis tasks with large quantities of data, the Domino's research department has developed specific needs from the software packages it uses.

"From a statistical standpoint, it has to be able to do a lot of things. It has to be able to process huge data files quickly and efficiently. I've run into a lot of problems with upper limits on the size of the data (some packages) can handle. It has to be flexible, with a command language that allows you to manipulate your data very quickly without having to program every step of the way.

"Processing speed is very important. Analysis is relatively easy for the most part when you compare it to the time it takes to process the data, to get it down to a form that you can use it in. When you've got a 10 or 20 megabyte file and you have to summarize it in a form you can use, that normally takes longer than the analysis."

Bond says that Systat is the package he finds most useful. "I have not been able to find a database that Systat can't handle. It has a very fast sort routine which helps us out when we want to break things out by different variables. It's more of a query system. We can ask questions and write a program to determine what the sales trends were-because trends are the big thing. Did the promotion shift a trend? Is it helping sustain a trend? Using the ANOVA packages or the statistical descriptive information that it gives you, you can begin answering those questions. "

Testing computer system

In the future, Domino's may have even more information at its fingertips, thanks to a computer system-now being tested at the company's headquarters-which would retain information on customer ordering habits and preferences for up to a year. The files would include information on everything from pizza topping choices to delivery instructions and payment methods.

"Unlike packaged goods firms-who would kill to know their who their customers are-we already know who our customers are. They give us their name, address, and phone number! The computer system is a means by which we can service our customers better. We feel that if we know who they are, and we have an idea of what they like and what they don't like, we can provide faster and more personal service. We think it will improve the overall Domino's Pizza experience."