We are often asked by clients whether and how to go about purchasing contact lists. In the business-to-business market, it's especially dicey. A recent study reveals that no two lists are alike; in fact, they can be shockingly different. “Online Sources of B-to-B Data: A Comparative Analysis, 2010” compared the performance of five list compilers that agreed to take part. The study's authors are Bernice Grossman, president of DMRS Group, Inc., a marketing database consultancy in New York City, and Ruth P. Stevens, a customer acquisition & retention consultant and marketing professor at Columbia Business School.

The list compilers volunteering to participate in the study were D&B Selectory, Demandbase, Infogroup, Jigsaw and NetProspex, with holding company Infogroup representing a number of list management companies and divisions. They were asked to indicate how many companies they were able to list in 10 industries and their volume of contacts in those industries. They also were asked to report on “complete” contacts — defined as full name, address, title, phone, fax and email — of 10 specified businesspeople each within separate industries. Those 10 served as the study's control because their complete data was known to the researchers.

"Overall, the results bring us to the same conclusion as last year: The business data available from vendors tends to be relatively accurate, but coverage is extremely spotty," wrote the study authors.

The authors provide three good options for conducting a comparative test before you buy:

1. Send each potential vendor a list of 5000 records from your house file and asking them to add data fields. Include a few dozen records on which you know the “truth,” to assess accuracy of what comes back.
2. Order a sample of names from a prospective vendor as per above, and then verify the accuracy of sample records by telephone.
3. Give each prospective vendor a set of instructions using very narrow criteria like a certain employee size range and sales volume range in a certain state. Ask the vendors to sort the records in ZIP sequence, and give you the first 1000 records to look at. A high incidence of identical records among the vendors will be a strong indicator of likely accuracy.