The business platform for digital media

Lessons in paid content III: Don’t just slap a meter on it

Feb 16, 2016

This is another in a series of articles from Piano’s Lead Data Scientist Roman Gavuliak that were written last year and are being re-posted here for the edification of our faithful readers.

Lessons in Paid Content I: Not all content is created equal
Lessons in Paid Content II: The size of your audience
To read all of Roman’s articles please click here.

Our last post looked at the size of a target audience for a typical publishing site with 2 million monthly unique users as reported by Google Analytics. After accounting for Google’s user overestimation and users that only skim headlines, there were between 0.5M – 0.7M monthly readers left. Now, assume the example publisher’s online team decides to implement a meter limit of 10 articles per month; one that mirrors the current New York Times limit. What follows is the monthly content consumption for users in our example (based on similar real newspapers):

Screen Shot 2016-02-16 at 10.51.20
The graph shows that the share of users reading more than 10 articles (10+ category) is below 10%. This is not uncommon for many titles and means a metered paywall with a setting of 10 free articles per month affects (building on the example) between 40k – 60k users, users who can be effectively monetized. This doesn’t mean a metered paywall makes no sense, it means there must be realistic expectations about its impact.

This curve allows for optimizing paywall impact, regrettably it lacks a sweet spot where there is an optimal trade-off between pageview and user impact. However, the sweet spot doesn’t exist because article pageviews follow roughly a 80/20 rule (80% of article pageviews are done by 20% of users) and, changes impacting pageviews are really much smaller when compared to changes in terms of affected audience. Therefore there is an inherent risk associated with implementing a metered paywall with any setting.

However, there is a big upside – a more aggressive approach brings more opportunity. To wit: numbers from metered paywall estimation are presented in the following table:

Screen Shot 2016-02-16 at 10.36.26
The conclusion here is: don’t be scared!

The percentage of affected pageviews is only on articles. A metered paywall leaves homepages and section fronts freely accessible. The resulting percentage loss of pageviews from all pageviews is lower and decreases further based on the conversion rate (since the pageviews of paying users are retained and might even grow as these users now attribute a monetary value to the content they read).

Back to the example; lowering the meter setting from 10 to 4 free articles per month means increasing the absolute percentage pageview loss risk by 5.6% while almost doubling the absolute share of users affected (on all users reading articles). Keep in mind too, that people reading over 10 articles per month are more likely to pay than those reading only five articles per month.

So what is the takeaway here? Ten free articles per month might sound like a nice round number and it works for New York Times; but while user distribution is based on their reading level and follows a similar mathematical function, it differs in parameters. These differences are illustrated in the graph below. One daily Estonian title has roughly 1M monthly unique visitors while a Slovak daily has 2M monthly visits and two German titles have 1.9M and 2.7M respectively.


Despite having the smallest audience, a limit of 10 free articles a month would impact the greatest share of users at the Estonian daily newspaper.

A metered paywall then, is not about the size of your site, but about understanding your audience and putting this knowledge into action.