http://happyantipodean.blogspot.com.au/2013/05/google-wallet-promises-to-help-media-to.html

Lots of people said that subscriptions wouldn’t work, and the numbers show that they were wrong. The websites that went to a subscription model early – and were criticised for doing it – now look like visionaries, as returns from ads have plummeted. Many people use a few websites for news – say, The New York Times and The Guardian – and so they can benefit from the fact that those companies share the same micropayment platform; other publishers can piggyback on that investment and use the same micropayment engine too. There is inherent scale in the model once you get a few big publishers on-board.

Has anyone explored doing a micro-payment where, if you’ve got an account, you’ll be charged for content consumed until you reach the subscription level, after which you can keep consuming, but without additional charge? The flip side would be that on months when you don’t consume much, you save a few bucks.

I just wanted to comment on the results we presented a year ago in our “white paper,” because I receive many questions about how representative they are.

First, the sample was not small at all: over 40,000 uniques in a month. As expected, the numbers of buyers — 1,281 — and paid transactions — 3,767 — were smaller, of course, but still statistically significant.

Second, these numbers represent real users and real actions — they are not opinions or declarations, which often are the case in other studies. Here, user visits a site, sees desired content, clicks on it, and — most importantly — completes a pay-for transaction to open/consume that content. Comparing to more traditional CTR or “leads,” our conversion rates are quite “real” and reliable.

Finally, we do not compare our rates to the paywall conversion rates. Please note that paywall conversion is often calculated as the ratio of the total number of digital subscribers, as they accumulate over time, to the current number of uniques generated by a given site. A more appropriate measure would be to compare the number of new subscribers in each month or quarter to the number of uniques the site had when the paywall was implemented. If you do that, then the conversion rate of the NYT meter — to use that popular example — is not 1.5% or even 1%, as it was often said, but a fraction of one percent. More importantly, it is decreasing, not increasing; in the last quarter, the NYT reported a net gain of only about 40,000 new digital subs.

Open, on-demand systems do not have this “inherit math problem.”

Greg is right. Publishers however will only start using them kicking and screaming. Their business model is advertisement revenue fueled by subscriptions. Offline, however, they also sell per issue so why not online?