The New York Times has reduced its use of vertical response lists in favor of compiled files in an effort to increase its home-delivery subscription sales. <<p> The effort began when the paper noticed its direct mail response declining. It started cutting back on volume, resulting in an increase of the cost per order, according to Harrison Sohmer, database marketing manager for the newspaper of record.
The Times has an average daily circulation of 1.1 million copies per day and 1.7 million on Sunday. Overall, it more than 1 million subscribers.
Subscriptions are also sold around the U.S., but only in 8,600 Zip Codes—our “deliverable footprint in the continental U.S.,” Sohmer said. Pricing varies as to whether it is in the home market or the national market.
The paper relied heavily on vertical lists—of book buyers, magazine subscribers, and people with “similar information needs as a loyal New York Times reader,” Sohmer said. But vertical files were costly, and this prompted a shift to compiled lists.
In addition, Sohmer’s unit licensed a major compiled database, and sought improved targeting through modeling. It had figured out that “it’s not the list, but the people in the households who are important.”
The paper built Zip code, household and block-group models, and segregated the best and worst buyers into deciles, suppressing the bottom ones.
Household models allow greater selectivity than Zip code models, but cost more, Sohmer said. Processing time is longer, and the mailer needs a good net-name arrangement for rented files since strong models can result in suppression of up to 50% of the names, he continued.
For local prospects, the Times uses variables like age, length of residence, and the percentage of college graduates in the home. For out-of-town names, it considers age, home loan amount, household size, length of residence, and mean travel time to work.
Why travel time? Because unlike New Yorkers who take the subway, people in other areas typically drive to work and can’t read the paper en route, Sohmer said.
The Times found that the city of Los Angeles did well in the highest deciles, but not the general area. Washington scored well in the top three deciles (“We should just mail to Washington, Sohmer quipped). Atlanta performed well, “but we can’t deliver every day there” Sohmer said. Las Vegas was disappointing.
One surprise was that book buyers were “a little weak in the top deciles. The arts were not all that great, and continuities not that good,” Sohmer said. “But membership organizations did better than anticipated.”
In January 2002, the subscription department was sending 97% of its mail to vertical lists, said Sohmer. By this April, that figure had been reduced to 48.8%. “It’s almost half and half,” he added.
Moreover, the Times reduced number of vertical lists by 57%. And it has decreased the number of names in overall plan.
The paper next plans to rebuild its Zip code models, refresh its household and block group models, to develop cross-sell models for other New York Times properties, and to incorporate 2000 census data.
Network
