The Data Route Delivers

One thing The New York Times has never had a problem with is brand recognition. But it did need a more cost- effective way of finding the prospects most interested in spending each morning with the “Old Gray Lady.”

A data mining and modeling initiative did just that, letting the newspaper use more compiled files for direct mail prospecting, cutting its new customer acquisition costs in half in the process.

Harrison Sohmer, database marketing manager for the Times, says that direct mail typically has much better retention rates than other channels such as telemarketing. “It’s like a prompted channel, but the subscriber has to take action. That’s really what works.”

Modeling has helped make mail more efficient, dropping the cost per order from about $400 to slightly less than $200. Sohmer says the Times has always used modeling to prioritize call center leads, but it’s a fairly new tactic on the direct mail side. In the past, less than 10% of the lists used were compiled. Now, the mix of compiled files modeled by region, and vertical lists — subscriber files for publications like The Nation, Time, and Newsweek, and arts groups such as The Shubert Organization — is a more economical 50/50 split. With the help of Direct StatSoft, not only was the paper able to increase its use of compiled names, but work with list brokers to get better net names from the vertical lists.

“It’s really helped us lower our cost per order,” he says.

The Times currently has about 1.1 million active subscribers, half of whom receive the paper seven days a week. Approximately half of the subscriber base is in the New York metro area. The remainder is spread across the country, with concentrations in areas such as San Francisco, Washington, Boston, Philadelphia, Los Angeles and Chicago.

A typical subscriber is in his or her late 40s, well educated and lives in an area with high home values. Not surprisingly, subscribers like to read, and enjoy hobbies such as skiing, antiquing, art and jogging. Sohmer notes that modeling has shown certain other leisure pursuits — such as hunting, fishing, car repair or sewing — indicate someone isn’t a likely Times reader.

The base rate for a subscription is $9.25 per week in the New York metro area, and $11.50 for the national edition. The cost means “it’s something more than an impulse buy,” notes Sohmer. The paper has a “core group” of subscribers who have taken the Times for over two years, as well as a “super core” segment who have been around for over five years. Then, says Sohmer, there’s a “froth” of people who have subscribed for less than a year. There’s a lot of churn in the last segment, especially for people who came in through telemarketing, or choose to drop the paper once the promotional rate ends.

The Times drops about four big mail campaigns annually, including 2 million prospecting pieces, as well as additional mail to expires and other house files. The direct mail control is a folded-over BRC. In marketing efforts, Sohmer says the Times plays up the domestic and international news coverage as well as the arts section, positioning the paper as a second read for a full range of news to the national audience.

The financial investment in the database efforts is in the “tens of millions” when you take into account the data warehousing efforts, business intelligence tools and call center applications across the Times’ units, says Sohmer. ROI is judged by cost per order, as well as new subscriber starts and how long they remain in the fold.

Sohmer says the Times is moving toward developing more of a monthly mailing program, targeting new residents and new homeowners. One factor that always has to be considered is that because the Times is a newspaper rather than a magazine, delivery is done by local carriers and not the USPS. This means the paper is only available in about 8,600 ZIP codes across the country. “One of the challenges is mailing more efficiently to those areas we can deliver to,” he notes.

Also in the works is a customer relationship management initiative, where the paper looks at households’ subscriber history. This will allow the Times to analyze behavior — such as an ongoing cycle of subscription starts and stops — over time, and use random effects models to target back expires, selecting them for targeted offers by news and information segments.
— BNV

Harrison Sohmer of The New York Times and Richard Deere of Direct StatSoft Inc. will present “New Data Mining Systems for Prospecting: Higher Profits With Multiple Models” on Monday, June 2 at 3:50 p.m.