Nielsen has just given us a great example of why one should always take publicly announced stats with a grain of salt.
The Bookseller caught up with a story from Book Expo America this morning. Sarah Shaffi reports that reps from Nielsen gave a presentation and shared bad news about the US ebook market:
E-book sales in the US declined by 6% in 2014 compared to the year before, statistics released by Nielsen show.
Information presented at BookExpo America and derived from PubTrack Digital, which collects e-book sales data from more than 30 of the largest publishers in the US, showed that in 2014 just under 223 million e-books were sold in America, down from almost 240 million units in 2013. E-books made up 26% of all book sales in 2014, down from 28% in 2013.
Update: The numbers are complete and utter nonsense.
While I wouldn't go so far as to say the numbers are bunk, I do wonder about their validity. I don't know whether The Bookseller made a mistake in their reporting or whether Nielsen's statistic model is off, but something looks wrong here.
To put it simply, Nielsen took revenue data from a few dozen publishers and derived market statistics for consumer ebook purchases in the US in 2014. While you can sometimes use this trick to find interesting data, in this case I think Nielsen used a survey group which was too small.
I have revenue stats from the American Association of Publishers in front of me, and they paint a different picture. The AAP collects revenue data from 1,209 publishers, and its data showed that publishers' ebook revenues were up 4.7% in 2014, to $1.6 billion.
I don't know about you but I have trouble accepting extrapolated market stats from such a small survey group when we have readily available data from a larger survey group which appears to contradict it.
Sure, it is entirely possible that both could be correct; the AAP's publishers could have gained market share in a down market. But the more likely possibility, in my mind, is that Nielsen's model was wrong.
Or did I miss something?