Modeling this is a bit tricky. To help guide you along, I$$ve provided some thoughts that should help

you structure your model. In particular, it will be useful to consider the impacts of no paywall vs$.$ leaky

paywall vs$.$ bulletproof paywall on advertising revenue $($driven by reach and frequency$)$ and

subscription revenue for three behavioral segments:

$$ Light users. They read $<10$ articles per month in the current situation with no paywall,

mostly just clicking through to read articles others have posted on Facebook, a colleague

emails them a link at work, an occasional article that catches their eye in a Google search, etc.

$$ Medium users. They average $30$ articles per month with no paywall, sometimes clicking

through on someone else$$s post and sometimes going directly to the site to read the news.

$$ Heavy users. The New York Times online is a primary news source for these avid news

consumers. They visit almost every day and on average read $>100$ articles per month.