Everybody knows the old business school joke about selling at a loss and making it up on volume. A casual analysis of Amazon’s balance sheet and press releases suggests that they sell many things at a loss and make it up with MarketPlace revenue. Another way to look at it would be to say that Amazon runs a highly successful MarketPlace for third party sellers with a struggling retail store attached. I’ll run through the rough numbers, and you can make up your own mind. Feel free to call me an idiot if you think my estimate of MarketPlace impact is too high, I tried to be very conservative. Just do me the favor to point out the flaws. Estimates are necessary because Amazon doesn’t break these numbers out in their financial reporting for competitive reasons. They don’t give any data about Kindle sales or eBook purchases in their reports. In fact, it’s not even clear to me whether Kindle device sales are counted as “Media” or “Electronics and other Merchandise”.
Amazon reported in their 2009 Annual Statement that 30% of their unit sales came from MarketPlace vendors. MarketPlace vendors vary from large retailers taking advantage of Amazon’s storefront to sell their items, to college kids reselling textbooks and grandparents reselling gifts from their progeny. The only difference between the large sellers and the small sellers, in terms of revenues earned by Amazon, is that small sellers pay $0.99 per listing before other fees, while serious sellers pay $39.99 a month to list as many items as they desire. Both pay closing fees on certain items (it’s $1.35 for books) and a percentage of the selling price to Amazon, varying from a low of 6% on PCs and 8% on electronics and cameras, to 15% on books and 20% on Jewlery.
Now let’s estimate how many items they sell. Amazon used to report the total number of items shipped during the holiday season in a press release each year, but they stopped after 2005. That year, the Delight-O-Meter reported over 108 million items ordered from Amazon worldwide from November 1st through Christmas. I’m leaving October out of the estimate for the time being, we’ll stick it back in a couple inches down. Amazon’s 2009 sales were more than double 2005 sales (2.8X), so could lead us to estimate that Amazon’s 2009 holiday sales were 302 million items ordered. This extrapolation is a weak point in the analysis as the averaging selling price in 2009 might be appreciably higher than 2005, especially given growth in Electronics. So instead of 2.8X, I simply doubled the 108 million items to 216 million for the 209 holiday season. From Amazon’s 10K, we know that 4th quarter 2009 sales accounted for 39% of the year’s sales (it was 35% back in 2005). If we assume that holiday items sell at the same average price as year round items, it yields a first order approximation of over 338 million items shipped during the first three quarters of the 2009.
If we now estimate that October was an average month, we can include another 38 million unit orders (338 million / nine months). This should make 216 + 338 + 38 = 592 million a first order approximation of the number of orders Amazon took in 2009. For a second iteration, we now add the 38 million estimate for October to the 216 million holiday season estimate for 254 million units ordered in the 4th quarter, and recalculate the first nine months as 397 million units. That puts the total year (minus the week after Christmas) at 651 million units worldwide. I’m not going to try for a third iteration or we’ll be here all day.
Since Amazon sold about $24.5 billion in total merchandise in 2009, that would put the average item a little under $38 dollars. I expect that MarketPlace orders would average much less, unless there are lots of people buying computers and diamond rings through third party vendors, but it’s important to keep in mind that both Amazon and Marketplace sell some big ticket items, like Kindles and big screen TVs. This leaves us with the estimate that Amazon processed orders for 651 million units in 2009, with MarketPlace vendors share at 30% amounting to 195 million units. This calculation assumes that the average MarketPlace item sold for the same amount as the average new-from-Amazon item, which I suspect is extraordinarily conservative.
So, the remaining question is how much does Amazon net from the average marketplace unit sale? For a book sold by granny, we know that the minimum is $0.99 + $1.35 = $2.34, and that assumes that granny sells the book for a penny. Serious sellers save the $0.99 listing fee, but at least one professional seller I spoke to won’t bother listing books for less than $10, which means Amazon makes a minimum of $1.50 (15% of $10) plus $1.35 = $2.85 on each of his sales. And I suspect these book examples are very conservative for Amazon marketplace, where I’ve paid around $100 (around $10 net to Amazon) for a replacement laptop battery, amongst other transactions. So let’s round down the average of the two book examples and say the average unit sold through MarketPlace nets Amazon $2.50 even.
We’re left with 195 million units at $2.50 profit each, for $487 million dollars profit. Yes, I know that it costs Amazon something to run the MarketPlace, but we’ve been so conservative at every turn that I’m going to ignore that as part of the noise and consider some other numbers. Amazon’s reported net income for 2009 was $902 million. That profit remains after they lost $849 million dollars on shipping. Under what I think is a very conservative estimate, more than half of Amazon’s net profit came from MarketPlace vendors. Another bit of conservatism was interpeting Amazon's holiday season as Nov 1st to December 25th (and not bothering with the week after Christmas), when in other places, they imply that the holiday season is Thanksgiving to Christmas.
Now let’s go back and look at a few of the variables. $38 dollars for the average unit selling price does seem awfully high, no matter what the Delight-O-Meter extrapolation suggests. If you had been asked to guess the average selling price of an item on Amazon, I wouldn’t be surprised if you’d chosen $20, which would almost double the number of units sold, and almost double the MarketPlace earnings so they accounted for all of Amazon’s profit. But let’s leave the average selling price alone and consider the average Amazon net per MarketPlace item. At $5.00 net per marketplace transaction, all of Amazon’s reported profit would come from MarketPlace. It true that some advanced MarketPlace vendors have been moving to “Fulfillment By Amazon” and saving a little on fees, but it really only makes sense for higher average ticket items, and my understanding is that it hasn’t really caught on yet. And if the average MarketPlace transaction was really $38, a number we stuck with to hold down the total number of items sold, it would almost certainly put Amazon’s net per MarketPlace transaction over $5.00. And don't forget our assumption that average item prices had risen. If the average selling price in 2009 was the same as 2005, what with all the inexpensive eBooks on Kindle, falling electronics prices, etc, the total items sold would have been a couple hundred million higher.
All of this suggests one conclusion. That it’s very difficult for any other bookstores selling new books, online or off, to compete with Amazon without losing their shirts. Amazon’s secret may be that they’re losing their shirt as well, but making it up in MarketPlace.