Since I did a post about Airbnb’s business in New York, Airbnb has published its own study and so there has been some follow-up interest in my numbers, here, here, and here.
The data and the conclusions cannot be conclusive because they are a sample (though a big one) of limited publicly-available information. I’m happy to be corrected, but Airbnb would have to release some real numbers to do so, instead of silly 300 word “studies”.
I don’t have new data, and I think there is enough detail in the original post to understand the sources and the limits of the analysis, but here is a PDF presentation that tries to make the conclusions a bit more clear. Here is a version formatted for download and printing.
[gview file=”http://tomslee.net/wordpress/wp-content/uploads/2013/11/airbnbny.pdf” profile=”3″ save=”0″]
Again a great job, i really like your articles and criticism about the sharing economy with solid arguments. I have one question about this article.
You are using the reviews as proxy. In my experience I will not get always a review after a succesful reservation. I think 20 to 40 % of my guest didn’t leave a review for my room. Of course you have no other data to use, but i was wondering if you take in account that not every guest leaves a review plus I also think that entire apartments will get less reviews than private rooms as you will not get to know the host… what are the numbers in that case? Any estimation from your side?
Thanks in advance for your response and keep up the good work!
Thanks Marius – it is interesting to hear from someone who has been a host. Your suggestion about review rates for full apartments makes a lot of sense.
I haven’t been able to think of a way to identify stays with no reviews. The only data I can get at is listing pages (like https://www.airbnb.com/rooms/475831) and host pages (like https://www.airbnb.ca/users/show/2358846) and I haven’t used anything from the host pages yet.
There was some work done on eBay about the information in missing reviews (see here: http://www.aeaweb.org/annual_mtg_papers/2007/0106_1430_1601.pdf) which relies on timing information to guess at the missing numbers. Interestingly they guess at around 20% which fits with your own experience. But Airbnb don’t provide individual rating values or times, so I don’t see how to make even those kind of guesses. Any ideas appreciated!