This time, Sulzberger engaged many people in the discussion. In early 2009, he kicked off a wide-ranging internal investigation into how to generate revenue from NYTimes.com. In February, a steering committee began to meet regularly and the newsroom hosted a series of brainstorming lunches. On March 30, Sulzberger and Nisenholtz went on a five-day trip to meet with business leaders on the West Coast including executives at Google and Amazon. The company also commissioned outside consultants, including McKinsey & Co., to do research, says Keller.

The brainstorming lunches ensured that people from the newsroom were part of the discussion. While we honor and respect the separation of the people who make the money and the people who make the news, we had to have the newsroom involved, says Keller. One lesson of TimesSelect was that the design of it was not inclusive. One of the reasons the columnists were upset was that they hadnt been in on the decision in the first place.

The eight lunch meetings ran from February 3 to July 27. Masthead editors met with a new group of one or two dozen newsroom staffers each time. Ways to generate revenue from web content was a major focus; charging for apps, a membership model, or donations got frequent mentions. Several newsroom staffers suggested that the Times consider micro-payments, asking customers to make very small paymentson the order of a few centsfor each article read. Science writer Ken Chang even prepared a detailed presentation on the micro-payments model.

During the first lunch on February 3, a newsroom staffer pointed out some interesting math: the New York Times website averaged 20 million unique visitors a month. Of those 20 million, three million generated 60 percent of the traffic. A dollar per month from each unique visitor would generate $20 million, but $10 per month from the more dedicated group of users would generate $30 million.

Intense debate. The internal debate about reinstating a paywall was intense. Many print editors, afraid that the website was cannibalizing print readership and given the global economic crisis, thought the Times should try a paywall again. The world had changed, says Geddes. We needed the money. Keller wanted to find a pay model that worked, for two reasons. What we do is distinctive and important, and readers value it. Therefore, just as a philosophical matter, they should pay for it and we should charge for it to send the message that it is different from commodity news, he says.

We were something pretty extraordinary and our readers knew it, which is why they were so loyal. A corollary to that was the Times had always had two revenue sources, circulation and advertising. And that had seen the company through some hard times in the past. When one pedal was not performing, you pressed on the other pedal. So the idea that we at least carry that principle over to the digital side seemed to make great sense.

But many on the digital side argued that a paywall would decrease the reach of the Times , retard website growth and divert resources from future products like apps that could provide readers with new services and generate revenue. Building infrastructure for the paywall meant integrating existing print and digital systems and building e-commerce system software. This would commandeer most of the papers IT resources for a year, meaning that the digital side would have to pull back on exploring other avenues for new revenue.

TimesSelect didnt provide a clear lesson, either. For those who favored a pay model, it was proof that people would pay. For those who did not, it was proof that a pay model would fail. It provided great ammunition for the people who wanted to argue that a pay model was a mistake. And there were lots of those people insidesmart, reasonable, thoughtful people, not reactionaries says Keller. Landman was the articulator of the misgivings from the web people, he adds. Nisenholtz was the main skeptic on the corporate side.

They had good company outside the building. Many newspapers had misgivings about charging for content, notably the Guardian and the Washington Post . CEO Don Graham refused to create a paywall, noting that 90 percent of Post readers were outside the newspapers geographic distribution area so it could not link Web access to paper subscriptions as many publications did. He also felt a fee would restrict the Posts influence, which he wanted as wide as possible. The Posts website general manager, Goli Sheikholeslami, told executives at a October 2009 newspaper conference that too few readers would be willing to pay for online newspaper content to make selling it viable. ^[28]

Models . At the Times , the steering committee looked at a variety of pay models, says Keller, including some that were completely alien to the company. These included ones used by outfits like Weight Watchers, National Geographic , and Consumer Reports as well as the few successful newspaper paywalls. The idea was to get a sense of who was able to charge for content and why, and to see what the Times could learn from them, says Denise Warren , general manager of NYTimes.com. ^[29]

The newspaper models included the Wall Street Journal , and the Financial Times , a UK business paper that had launched a metered paywall in October 2007. The metered paywall allowed free access to five stories per month, or 30 per month to those who registered, charging only heavier users. ^[31] Looking at all the models was like investigative journalismyoure just exploring and you dont know what you are going to find out, says Warren. The committee considered dividing the paper into component parts and selling the more popular sections, similar to TimesSelect; a donations model like National Public Radio; and at least in passing, the question of whether we should approach philanthropic organizations and whether you could actually have a for-profit business that in some way capitalized on the goodwill of foundations, Keller says.

NYT business executives took a detailed look at micropayments, but concluded that they would not generate sufficient revenue and would be difficult to administer. The company also investigated the membership modelproviding extras for paying subscribers. But it didnt test well, says Nisenholtz. If people felt the core product of the New York Times wasnt worth paying for, apparently they wouldnt pay for extras, either.

It became clear that if the Times was going to charge, it had to be some kind of paywall that required people to pay for regular Times content. The paywall used by the Wall Street Journal , Consumer Reports , and some newspapersa hard wall that prevented all nonpaying customers from going to all or part of a websitewas quickly ruled out because of the TimesSelect experience. A hard wall would have completely destroyed the ad revenue machine, says Warren.

The FTs metered paywall was different because it allowed anyone to read a set number of stories per month before shutting them out, and it allowed access through a search engine. This had worked for the FT, which garnered 101,000 subscribers in 2007 and 117,000 in 2008 at a relatively high subscription price of around $300 per year. ^[32] But many remained skeptical that the FTmodel could translate for the Times audience because, like the WSJ, the FT had a specialized business audience. Keller explains:

A lot of us felt the same way about the FT that we felt about the Journal that its a specialized publication. It goes to people who expense their subscription. Its bankers and business executives and CEOs and corner office guys. It is not the newspaper for the masses or even for the educated masses. Its a very specialized publication. So you cant emulate them.

The steering committee started to talk about what should remain free. It made a distinction between a hard paywall and one that both allowed a certain number of stories per month, and was flexible enough to allow readers linking in via blogs or search engines to read the Times for free, says Keller.

It dawned on me, and I think it dawned on other people at the same time, that we werent talking about a paywall. We were talking about a kind of semi-permeable membrane... We could make it as open as we wanted and we could calibrate. If we werent getting the subscriptions we wanted, we could [close it] a little bit more. And if we were losing traffic, we could open up the spigot a little bit. So that, I think, was the intellectual breakthrough.

Keller on a flexible paywall.

Once the steering committee realized that a paywall didnt have to be like an iron curtain, it started to look more closely at the Financial Times model.