The Metered Model and the New York Times

The Metered Model and the New York Times

The New York Times reported their second-quarter results on July 22nd.  One topic of conversation on the analyst call was the metered model under development.  Another topic was Internet revenues — specifically advertising.  It sparked a question for me, about what the conference call might sound like in July 2011 with a metered model in place.  How would the ad and subscription revenue compare?

I decided to do some of my own analysis.  The goal here is not to replace Alexia at J.P. Morgan, rather to examine the impact of a combined subscription and ad revenue model.  To model the New York Times metering, I needed to develop some assumptions: number of readers, revenue per reader, and subscription revenue per reader.  With that data in hand, I could look at 5%, 10%, and 15% subscription conversions and get a high-level assessment of impact. 

Between the press release and the call transcripts, the News Media Group did $50.4M in advertising revenue during Q210.  This is split between (20M uniques/month – 80% of visitors) and (5M uniques/month).  Assigning 80% of the revenue to, the site did $40.32M in ad revenue during Q210 (no subscription revenue currently).  This puts the revenue per reader at $2.02/quarter or $8.08/year (NOTE: erroneously, I tweeted a higher number earlier today).

From recent news reports about New York Times reader surveys, my analysis assumes an incremental $15/month in revenue per reader for subscribers.  This assumption is the lowest figure presented in the survey.  The final could obviously be a higher or lower average increment per subscriber, but my choice was to use the lower bound of any publically stated number from the New York Times.

Properly dialled in, the metered model does not materially impact a number of unique readers.  It should on the other hand, monetize the most loyal readers.  So what would happen at 5%, 10%, and 15% subscription rates?  Here is a summary of the current baseline:


Here is the incremental impact of a metered model. Note in particular the impact on average revenue per reader per year to increase the topline.

Conversion RateReadersIncremental $/Quarter$/Reader/Quarter$/Reader/Year$/Year
5% 1M $45M $4.27/r/q $17.06/r/y $341M
10% 2M $90M $6.52/r/q $26.06/r/y $521M
15% 3M $135M $8.77/r/q $35.06/r/y $701M

In the scenarios painted here, the average revenue per reader increase by 2-4X.  The above scenarios do not take into account any possible changes in CPM which could come about either due to better targeting or better ad units.  Nor does the analysis include any affiliate network revenue associated with bloggers to drive traffic.  So in addition to the subscription revenue, other revenue sources could continue to boost the topline.

Caveat Emptor
This was a quick and dirty analysis.  For example, The 20M used as the reader count is the average per month is not precise.  It is certainly the case that some readers are only drive-bys and therefore the actual readers for the quarter is higher than 20M by some percentage.  That being said multiple drive-by readers equal the effect of one normal reader so the 20M average seemed reasonable. 

Another example is that the revenue of and are probably not split 80/20, but that is the best guess I cold make.  It should be within the ballpark.

And finally, another example is that bloggers maybe some of the most loyal readers but may be unlikely to pay.  The New York Times may run a special “blogger” program as part of an affiliate arrangement and not charge them.  This would lower the number of conversions but likely by a small percentage.

The point of the exercise was not to create a perfect model but to show the merits of the metered model can be substantial if appropriately leveraged.

Press Releases

Press Releases

Scout Analytics™ Quantifies the Inaccuracy of Cookies as a Measure of Unique UsersEven a 100% Reliable Cookie Overstates User Counts 2 to 4 Times

Seattle, Washington — Feb. 17, 2010 — Scout Analytics™, the leader in behavioural analytics for maximizing the value of customer engagement, today released new research findings detailing further limitations of cookies, a commonly used method to track user behaviour on websites. During the past 6 months, Scout Analytics tracked hundreds of thousands of subscribers through a combination of patent-pending tracking technologies of the device and biometric signatures. Scout Analytics discovered cookies have an inherent weakness that causes them to overstate the user counts on an average of two to four times.

“Virtually all measurement techniques have some rate of error, but online marketers who have a heavy reliance on cookies need to know this method has astonishingly low accuracy,” said John Lovett, senior partner for Web Analytics Demystified. “Because of this, we expect new innovations in measurement technology in the near future that will no doubt minimize marketers’ reliance on cookies and dramatically improve measurement accuracy.”

For paid content providers and marketing professionals, the traditional approach to usage tracking is through cookies, a small piece of data that is stored on a device by the web browser. Each new browser that accesses an application is given a cookie, and the application relies on this to track user behaviour. However, user tracking based on cookies doesn’t track actual individuals – it tracks cookies browsers.  Several types of errors are introduced as a consequence. First, users access applications through a variety of browsers and machines—from the office, at home, mobile devices, etc. Second, cookies are often reset by the user or even automatically by the browser. The assumption that an individual user will utilize a single cookie browser is therefore flawed.

To research the extent of these consequences, Scout Analytics used tracking techniques of the device and biometric signatures to follow the behaviors of hundreds of thousands of named users accessing paid content products. The biometric signature identified unique users through an individual’s typing pattern to eliminate errors in user counting such as account-sharing. The device signature identified unique devices through data elements collected from the browser to eliminate errors in device counting such as cleared cookies. By correlating the named user account, biometric signature, and device signature, an accurate mapping of individual to devices could be produced.

The scope of the research included analyzing more than twenty million visits to various paid content products. From the device signatures of each visit, Scout Analytics was able to identify nearly 600,000 unique devices. Further correlating the biometric signatures from the visits, Scout Analytics identified more than 175,000 unique individuals approximately 45,000 of whom were unlicensed. When each distinct paid content product was analyzed, Scout Analytics found the number of users to be two to four times overstated by a theoretical 100% reliable cookie (i.e., even if cookies were never deleted).

“With the contraction of ad and subscription revenues in the information industry today, publishers need better audience data to optimize revenue,” said Matthew Shanahan, senior vice president of strategy for Scout Analytics. “There are a number of alternatives to cookies that can improve insights such as those used in our own study.  New devices such as phones, readers, and tablets will only make the consequences of tracking and accuracy more pronounced.”

Scout Analytics is the leading provider of behavioural analytics that maximize the value of customer engagement in recurring-revenue businesses. Scout Analytics’ unique SaaS offering delivers revenue optimization across multiple industries, including real estate, financial information services, SaaS and digital media.

Scout Analytics is a venture-backed company headquartered in Issaquah, Washington. To learn more about Scout Analytics, visit or call 425.649.1100. Follow the Scout Analytics blog at