It’s entirely possible that The New York Times will net a profit from their newly announced paywall, set to debut in a year’s time. But it’s by no means guaranteed. Even (momentarily) setting aside the journalistic or civic-minded concerns about shutting some readers out of the news, the whole idea makes little sense if the basic math doesn’t work out. Making money would seem to be the most basic marker of a paywall’s success.
Unfortunately, no one knows for sure whether it will. It’s all estimates, assumption, and guesswork — even if it’s relatively well informed, carefully researched guesswork. We just don’t know how readers and advertisers will react.
But now, with the debut of Paywall!, our revenue game, all that guesswork can be your guesswork. It allows you to explore the situation at the Times or at any other news site.
The calculator starts with numbers that attempt to model the Times’ online traffic distribution as best I can estimate it (see references below) and what seem like generous assumptions. If 60 percent of the site’s most loyal readers subscribe at $10 per month and no readers are lost, a paywall allowing 50 free page views per month should net about $500,000 per month in extra revenue. But this number is very fragile. It goes negative if the subscription rate drops just 7 percent, or the Times loses 6 percent of its readers, or if the monthly subscription is just $3 cheaper.
My estimates of the Times’ situation are certainly off by some amount; these figures are by no means definitive. But the Times must use estimates too, because no one can anticipate reader response. The lesson of Paywall! is not the final revenue number but its great sensitivity to many factors. Modest changes in site traffic or a few percentage points difference in subscription rate can quickly turn a profit into a loss — or vice versa. Whatever else it may be, putting up a paywall is not a move without risk.
How to estimate your paywall revenue
How do you play? First, hit “Turn Paywall On!” From there, “Views before paywall” is the most fun slider, and the number that many paywall discussions focus on. This sets the number of free pageviews (not the same as stories) that are allowed for each reader before requiring them to subscribe. As the number of free views decreases, the net revenue jumps as each audience segment hits the paywall, then falls from lost ad impressions. Somewhere, there’s a sweet spot.
The key to paywall revenue projections is to understand how different portions of the audience are affected differently. The model used in this calculator breaks the audience into five distinct segments. These can be given names such as “Fly-By” and “Daily,” but for accounting purposes each segment is completely described the number of unique visitors (readers), the number of pageviews per month, and the fraction of readers who will subscribe when they hit the paywall. (Of course, in the real world, people aren’t so neatly divisible into segments.)
The main graph shows these five segments as five bars. The height of each bar is the number of pageviews per month for that segment, and the width is the number of readers. Each pixel on this graph corresponds to a fixed number of pageviews times users, and therefore the same amount of advertising revenue. Ads shown to unsubscribed readers are in blue, ads shown to paid subscribers are in red, and ad sales lost due to non-subscribers stopping at the paywall are in gray.
The scroll bar at bottom of the graph zooms the display for better viewing. The calculator starts zoomed in for clarity, but by zooming all the way out you can see that only a very small fraction of readers will be affected by most paywalls. The crux of the paywall issue is that these are also the most valuable readers, the ones that a publisher can least afford to turn away. In terms of ad revenue, one Loyal may be worth a hundred Fly-Bys.
Ad revenue is captured in the “CPM per view” slider, measured in dollars per 1000 pageviews; it can be thought of as the per-ad CPM times the number of ads on each page. Some pages have higher CPM than others, so this value is an average across all pages actually served.
When a reader hits the paywall, several different things can happen. They may subscribe; they may come back next month when they have free views again; or they may never come back. The “Subscribed” and “Never came back” sliders model this.
“Subscribed” is the fraction of the most loyal readers who subscribe when they hit the paywall — that’s the width of those red “paid” bars on the graph. This figure is necessarily a guess, and the real world subscription rate will also vary by segment, with loyal readers far more likely to subscribe. That’s why there are segment-specific subscription rates in the boxes at bottom. The slider up top sets the maximum possible subscription rate, the rate for a segment with a relative subscription rate of 100 percent. Paywall revenue is very sensitive to subscription rate, because every non-subscriber also represents lost advertising impressions.
“Never came back” represents the fraction of the audience that simply disappears when the paywall goes up. Some regular readers will hit the paywall and switch to a source of free news — but even readers who wouldn’t hit the paywall may be lost, because the existence of a paywall can discourage linking. In the Times’ case, they’ve said that articles arrived at via links from other sites won’t count towards paywall metering — but that might just encourage people to browse Times content through an aggregator instead of the front page, which still amounts to a loss of casual readers. In any case, this slider subtracts readers from all segments in the same proportion.
Below the graph are the audience segment definitions. Each of five segments is described by the number of unique readers in that segment, the number of monthly pageviews of each of those readers, and the subscription conversion rate relative to the most loyal readers. The subscription rate slider and the relative subscription rate are multiplied to get the final subscription rate for each segment. A bit tricky, I know, but I wanted to make it possible to visualize global changes in subscription rate with one slider.
The number of pageviews for each segment is also calculated; note that the Loyal and Fly-By readers both represent a large fraction of pageviews. Again, this is the difficulty with a paywall.
Is a paywall worth it?
I’ve run through many different scenarios with this calculator, and the conclusion always seems to be the same: A tuned paywall can make money for a large free site, but the details matter greatly. Reader reaction is key; small variations in response have big effects on net revenue.
Perhaps a subscription system will increase revenue — but perhaps not, and probably not enough to transform the economics of news publishing. “This is not going to be something that is going to change the financial dynamics overnight,” as Times publisher Arthur Sulzberger himself said. This ambiguity means that paywalls will not “save” the classic newsroom model, at least not for general interest news production.
But the figures can also be read the other way around. Imagine you were a news site considering switching from subscriptions to a purely ad-supported model. In that interpretation, free-to-the-consumer news seems just as financially viable as paid news — or just as fragile. This may mean that free news is here to stay, which would be good news for those who think it’s important that the public consume journalism.
This is a tough time for journalistic organizations, and theorizing doesn’t pay the bills. But perhaps this tentative conclusion will help by reframing the discussion towards delivering a more valuable product more cheaply, rather than valiantly trying to return to what was.
I have used the best estimates I could find of The New York Times monthly business as the default values for knowable figures in the calculator. For unknowns, such as the subscription rate for each segment, I made wild guesses and tried to err strongly on the side of values that are optimistic in terms paywall revenue.
The distribution of users between segments is the key parameter. I didn’t have detailed numbers from the Times itself, so I used the audience profile from another daily paper, helpfully provided by Damon Kiesow of The Nashua Telegraph. This defined the percentage of readership and relative monthly page views for each segment. The New York Times and The Nashua Telegraph are obviously less than identical, so caveat emptor.
The best public audience distribution data for the Times are from Quantcast, who estimate that 25 percent of visits come from 1 percent of readers. The Quantcast data are not directly usable in paywall calculations because they measure visits, not pageviews, but the concentration of visits in the top 1 percent of readers closely matches Kiesow’s analytics when sliced on visits, and the basic pattern reported by Steve Yelvington for local news sites.
Based on an “internal memo,” Business Insider reported last week that NYTimes.com received 14,849,000 unique visitors in December 2009, and that visitors viewed an average of 18 pages each. I created the final segment profiles by scaling the segment distribution data to these aggregate numbers.
Next up was CPM per pageview. The Wall Street Journal reported $100 million in NYTimes.com ad revenue for 2008, or $8.3 million a month. When divided by total page views, gives a CPM per page of about 31. Since there are typically three ads of various sizes on an NYTimes.com page, this estimate seems in line with industry averages of $10-$15 CPM per ad.
The relative subscription rates for each segment, I just made up — trying to be very generous. Only 2.4 percent of all readers subscribed at existing U.S. news sites with paywalls, according to a recent survey. The default numbers in the model represent 10 percent total subscription, when the subscription slider is at 100 percent.
If you have better information, or are willing to contribute analytics data on any of these parameters from your own news site, please send it along and I’ll update the model for everyone’s benefit.