Oyster, the eBook subscription service whose high-profile launch we covered on this blog last year, has been in the press this week talking about how user analytics are driving its development as a business. In an interview with FastCo, Oyster's co- founder and the chief product officer Willem Van Lancker, has given some broad hints towards what its data is telling the company about how users consume written content.

For example, Oyster claims to have discovered for definite that reading is a night-time business, seeing most activity and most enquiries from users between 8pm and 2am (US Eastern Time). This insight, Lancker added, has led Oyster to remodel the way it delivers customer service so that it has more people available to answer questions and resolve issues out of standard office hours . Similarly, Oyster also exclaimed surprise at the number of queries it got from customers who wanted to gift an Oyster subscription.

There are two ways of looking at this particular insight. Either than an eBook subscription service like Oyster is irresistible to gift-givers looking for a present for friends and family who are heavy readers, or Oyster is merely benefiting from launching a service into an environment where virtual gift-giving is now well-understood. Numerous eBook retailers and audiobook subscribers launched their own gifting services in the latter part of 2013. And although none of these retailers has released sales numbers for gifted content (more on this point later), the fact Oyster entered the market at the same time as businesses such as Audible and Sainsbury's suggest it is less of a leader than it would have us believe. There are also lasting questions as to whether the heavy service user (who hasn't paid for their own subscription if it's a gift) is a blessing or a curse for a business like Oyster. This more general New York Times article about eBook subscription published at the end of 2013, warned that the all-you-can-eat model put forward by Oyster and its competitor Scribd could prove unsustainable.

Throughout the Fast Co article, Lancker alludes to data points that support Oyster's business decisions to date, but apparently refuses to be drawn on the data itself. We learn, for example, that users in Atlanta are 30 times more likely than the average to to read a romance, but there is no indication of how this average is defined. whether this average is US or international nor (most importantly) how big is the pool of data from which these conclusions are drawn. Without this context, Oyster's insights start to feel less driven by large data sets and more anecdotal, which is in itself underlined by the quote that ends the piece.

"The most exciting thing for us is taking someone who isn't a reader and reintroducing them to books," he says, citing subscriber feedback. "They weren't reading books at all, and now Oyster has brought them back into it, and now they're reading a book a month."

Reading this one can't help but be reminded by the economist Roger Brinner's now legendary warning against drawing authoritative conclusions from partial information: "the plural of anecdote is not data". Oyster may well want us to think it is sitting on a goldmine of user insights, but the selective way in which it uses data points, coupled with the youth of the service and Lancker's care to stress that curation also plays an important role in its business model suggests otherwise. There are also two ways of interpreting this. The first is that Oyster does not yet have sufficient user data to make statistically valid claims about reader behaviour and is falling back on anecdote as a way to bridge that gap. The second is that for all its claims about wanting to help the book business to develop by providing data-driven insight, it still wants to keep the data itself in an Amazon style 'black box'.