Project Wordsworth Recap: Findings and Statistics

By James G. Robinson

Project Wordsworth was conceived by 17 Columbia Journalism students as an experiment to find out what their longform stories were worth. Was it possible to make money from writing creative nonfiction without depending on a major publication? Here is a brief recap of what they found out in the first week and a half of their experiment, through the eyes of one of their professors.


Core Principles

Over the course of the semester, the students came to a consensus on three core principles that guided their decision-making as Project Wordsworth was developed:

  1. “Our stories are worth something.” These students saw themselves as working writers, not charity cases, and felt they should be paid for their work. (They just weren’t just not sure how much.)
  2. “Without sharing, our stories won’t find an audience.” Their target audiences (those most likely to read, share and pay for their work) are communities of people interested in the topics they were writing about … not necessarily people they had direct contact with.
  3. “The design of our site should equally encourage reading, sharing and paying.” Paying should be encouraged but not required; a hard paywall wouldn’t work, because it would discourage sharing and block casual reading.

The class split into three groups: one handling e-commerce, another tasked with designing the site, and the third serving as a social media team.


What To Charge?

The biggest question facing the e-commerce group was how much to charge for each story. They settled on an approach similar to Radiohead’s pay-what-you-want model for their 2008 album “In Rainbows” – at the end of each article, readers were strongly encouraged to pay (via PayPal) what they thought the story was worth. By leaving it up to the reader to put a value on each story, the students would collect a rich data set on willingness to pay at each price point.


Design + Site Architecture

Keith Collins (one of the students on the site design team) has posted a “behind the scenes” description of the site’s design and architecture here.


Social Media: Getting The Word Out

The social team launched an email, Twitter, Reddit and Facebook campaign the moment the site went live. More on that will come in a later blog post by the team.

Caroline Chen’s story, “The Paradox of the Proof,” was quickly picked up by Hacker News and Reddit, and spread throughout the Internet from there. Websites that linked to her story included:

Additionally, the former NFL player Clinton Portis tweeted a link to Vidur Malik’s story, “Hit Man: How Sean Taylor Left His Mark.”


The Results: Traffic

By any measure, what unfolded exceeded all of our expectations. Through Sunday May 20 (just over ten days) over 140,000 people had visited the site, paying over $3,700 to the authors. Collectively, the stories had drawn over 11,000 Facebook likes/shares and been tweeted nearly 2,000 times.

The chart below shows traffic and purchase data to each of the 17 stories over this time period. Caroline Chen’s “Paradox of the Proof” resulted in the majority of readers and purchases. But all stories performed well, given that the students were starting from scratch. Note the high time spent across the board, indicating that many of those who visited the site took time to read them.

Wordsworth Recap

Bear in mind that this represents traffic for just over a week. Moving forward, it will be interesting to see if some stories simply take longer to find their natural audience.


The Results: Who Paid, and How Much

Here’s a frequency distribution of payments to Project Wordsworth over the first ten+ days of the experiment, showing the number of people who chose to pay at each price point.


  • Not surprisingly, $1 was the most popular choice, followed by $2, $5 and $10.
  • 14 people gave $0.25 or less, which went entirely to PayPal as commission.
  • A few dozen people (including some family members) gave more freely than would be expected of an average reader. Some of these seem to have been intended as charity for the story subjects, particularly for Katie Campo’s story on Hawa Salih. One generous soul (not known to the author) gave $500.
  • We have no clue why someone would give $5.77.


The Results: Pricing and Revenue

The next chart shows the impact of variable pricing on the experiment. We used the following methodology to estimate how much money we would have made if we had chosen one of the price points above:

  • We assumed that the total number of paying customers at each price point would be the number of people who actually paid that amount or greater.
  • We then multiplied that number by the price point to get an estimated revenue amount.

For example, 390 people paid $2 or more. So we estimate our revenue would have been $780 (390 x $2) if the price had been set at that amount.


Of course, there are many assumptions in this method, not least of which being that people would be just as willing to pay given a set amount as they were given an empty box. (The fact that this was a student experiment, with participation from friends and family, probably also has its own biases.) But given those caveats, our data suggests two interesting outcomes:

  • Had they chosen a single price for each story, $5.00 would probably have been the optimal amount. That number was discussed as a possible price early in the project, although it seemed high for a single story. But it seems that enough people were willing to pay that amount to make it worth considering. (And 60 additional people were comfortable at a $3-$4 price point.)
  • None of the individual price points came close to matching the revenue from variable pricing. This data indicates that had we chosen a single price, the best we could have hoped for would have been just under $1,000. By giving our readers the option to choose what they thought each story was worth, we more than tripled that amount.

These results were somewhat surprising given the students’ assumptions at the start of the project (and the conventional wisdom among the journalism community). Project Wordsworth suggests that even without an established brand, you can get people to pay for a great longform story. And the proper price for that story may be closer to five dollars than one.


What’s Next?

The Project Wordsworth team will soon be writing another blog post to explain what’s next for the website. Stay tuned!

James G. Robinson, Director of News Analytics at The New York Times (@JamesGRobinson) was an adjunct professor in the course that produced Project Wordsworth.