Building a Data Warehouse

Building a data warehouse for 100 percent of their customers and those customers’ data has proved to be very expensive for many companies. Most large companies that have built such a warehouse have spent at least $20 million on the project. Why is it so expensive? Because a data warehouse is designed by a committee. It has to include all the data that an organization ever conceivably uses: data on customers, prospects, leads, transactions, promotions, responses, products, prices, models, and employees. Since the accountants are involved, the data in the warehouse have to balance to the penny. Since IT is involved, the warehouse is built by in-house programmers who have never built a warehouse before; they are very excited by the idea, but they have no experience at all.

As most people who have studied data warehouses will tell you, these warehouses almost never pay for themselves in terms of increased profits. Why not? Because the benefits of using a warehouse are incremental. The warehouse will enable you to do better targeting of your communications. Instead of getting a 2 percent response rate, you will get a 2.1 percent response rate. The question is, will the profits from the 5 percent increase in the response rate pay for the cost of the warehouse?

In most cases, they won’t, and you can prove it to yourself with a simple Excel spreadsheet before you begin to build your warehouse.

Suppose your company is taking in $156 million per year in direct sales to customers. The benefits of a warehouse are shown in Table 10-1. That’s very nice. Your warehouse has brought in 30,000 more sales, totaling $7.8 million more per year. But let’s look at the incremental profits. Suppose you are making a 30 percent profit on direct sales. Table 10-2 shows what would happen.

Table 10-1 : Warehouse Costs

	Current	Warehouse	Increase
Number of offers	100,000,000	100,000,000
Response rate	2.0%	2.1%
Number of responses	2,000,000	2,100,000	100,000
Sales conversion	30%	30%
Number of sales	600,000	630,000	30,000
Average sale	$260	$260
Sales	$156,000,000	$163,800,000	$163,800,000

Table 10-2 : Increased Profits from Complete Data Warehouse

	Increased sales	Profit rate	Increased profits
Annual warehouse benefits	$7,800,000	30%	$2,340,000
Annual warehouse costs			($3,500,000)
Loss from warehouse			($1,160,000)

Your warehouse will lose you about $1.1 million per year. So what should you do? Use the 80 percent rule. Don’t build a data warehouse that will retain all the data you could ever possibly need. Build a simple data mart that meets 80 percent of your marketing needs. Even the largest corporation in America can build such a data mart for less than $1.5 million total, with an annual cost of about $800,000. Table 10-3 shows the result of the 80 percent solution. You have converted a $1 million loss into a $1 million profit increase by using an 80 percent solution.

Table 10-3 : Increased Profits from Data Mart

	80% increased sales	Profit rate	Increased profits
Data mart benefits	$6,240,000	30%	$1,872,000
Data mart costs			$ 800,000
Gain from data mart			$1,072,000

Where did the $3,500,000 annual cost for a warehouse come from? Here is a simple way to measure data warehouse costs. If it costs you $6 million to build the warehouse, you can amortize these costs over 3 years, for $2 million per year. Maintenance of the warehouse will cost you at least $500,000 per year. Once the warehouse is built, you need to use it to communicate with your customers—otherwise, what is the warehouse for? Let’s say that you spend only $1 million on customer communications per year. Then the warehouse will cost you at least $3,500,000 per year, as shown in Table 10-4.

Table 10-4 : Warehouse Costs

Step	Cost
Build warehouse	$6,000,000
Amortize over 3 years	$2,000,000
Annual maintenance	$ 500,000
Customer communications	$1,000,000
Total	$3,500,000

Whenever you or any of your colleagues at work come up with a proposal for a large direct response project, Web site, or warehouse, ask this question: How much could we save by concentrating on the easiest 80 percent and postponing the rest for later? You will be amazed at the profitability of the answers.