How accurate is AirDNA/Everbooked/Mashvisor?

Question

Hey BP. How accurate are short term vacation rental market research tools such as AirDNA, Everbooked, Mashvisor?

I am looking at Destin, FL and these sites are putting out numbers for a 2/2 condo as generating 40-50k in revenue a year on AirBNB. It will then show similar listings nearby with revenues they have made in the similar range (40-50k). Is this legit? Besides the variables of how well an Airbnb listing can do based on reviews, am i missing anything here?

Martha Jardine Wilson · Answer

@Carl Mccrory Hey Carl, Here's a quick summary of how we gather and process our data here at AirDNA Every day we ping five million rental listings around the world and look for any changes to the availability of every calendar day over the next yearWe record any changes to the calendar and record some key characteristics such as length of stay, booking lead time, property booking history, reviews, host response rates, and much more.Our proprietary algorithm analyzes every new set of unavailable calendar days to determine whether the dates are a reservation or blocked by the ownerWe then record each unique reservation and determine the value of each reservation using the rates that were available the day before the booking was made.How the algorithm works: In 2015, we were able to secure a dataset of over 30 million actual Airbnb reservations. We were able to dissect each of the reservations to create a list of signals, that in combination, could predict what is a reservation vs. a blocked set of dates with 92% accuracy.Over the course of the last few years, we've acquired data on over 100,000 short-term rentals from an assortment of software provides and large vacation rental managers which allow us to continually adapt our model to the evolving marketplace. So.. how accurate is all of this algorithmically generated data?Is our data 100% fail proof? Absolutely not. Is it light years ahead of anything else on the market? There's no doubt in our minds. CBRE has compared our algorithm output with Airbnb's official announcements and has found it extremely close. Our estimate for the total revenue generated by every listing in the DC market for an entire year was 96.2% accurate. Feel free to reach out to us and we can discuss Destin, FL in more detail.

Paul Sandhu · Answer

AirDNA was spot on in my market. Then I did some research and realized that only my properties were in their data set.My market is not your typical market. Google "Worst place to live in Kansas", that's my home town. Our homicide rate is right behind Baltimore and East St. Louis. My renters are contractors, those guys you see on big erections of steel at a petrochemical refinery.

Jon Crosby · Answer

@Carl Mccrory, just my 2 cents, but the accuracy of these sites is really dependent on how entrenched you get into the data. I believe the data to be accurate in so much as the amount of data that can be scraped from Airbnb, which is actually a lot of very valuable data if you know what to do with it. All of these products have their own proprietary algorithms (including Artificial Intelligence and Machine Learning) to come up with an expected ADR or Occupancy rate (and many other valuable data points) so there is real data analytics going on with these products. In the markets where they all 'agree', it's my opinion that those markets are likely fairly predictable to your expected rental activity. But in markets where the returns a little more diverse, I think there is more value placed on good old fashioned customer service, marketing and what amenities the property provides that will dictate your actual occupancy and nightly rates. Like any tool (or person), trust but verify and once you find a property that looks good, then dive into the numbers head first and reach out to local property managers, find similar properties manually on the sites and go over reviews and calendars and get as much info as you can to fine tune your variables to figure out a conservative revenue number. Also, yearly revenue means little in the grand scheme of things, it's really your Op Ex ratio that is the silent killer with STRs so be sure once you have a yearly revenue you are comfortable with, truly dive into the operating expenses involved. Cheers!

Julia Bykhovskaia · Answer

It might be - but take it with a grain of salt. Some data is better than no data but there are numerous flaws and even if it’s correct - could be deceptive. AirDNA in my experience is more accurate than Mashvisor. Keep in mind:- the system is not great at telling whether the blocked day is a paid day or blocked by the owner - some people only rent their properties during a part of the year and not full time- AirDNA and Mashvisor rely on Airbnb data. For HomeAway and Booking.com they have to guess prices Still, directionally, it can give you some sense how well the properties are doing on a high level. The best way to find out - find someone who is doing it and see if they just tell you their numbers :)A trick you could use: call property management firms, play damn, say you want to buy a condo and give it to them to manage and ask how much a property like that could generate in their opinion. They have a lot of data.

Luke Carl · Answer

Airdna is way off in my brief experience with it. But I also study my market and can look at any given property and guesstimate how much it will gross and get pretty darn close. Of course it all varies greatly depending on who’s managing. Regarding Destin it’s hard to make a buck there unless you pay cash, then can make a bit more than a mutual fund. If you finance you’ll be lucky to not lose money.

Mike V. · Answer

@Martha Jardine Wilson, thanks for chiming in. Very interesting.Unfortunatly my experience hasn’t been as accurate as you articulated above. I have properties that AirDNA claims is a top 10 performer (based on the Airbnb info you posted above I assume) in my market, yet 80% of my reservations come from VRBO. How does the system come to this conclusion if it’s only accounting for 20% of my reservations?I do see a ton of value in seasonality trends and number of units by bedrooms but aside from that in my market (Hawaii) you guys are way off.

Travis Rasmussen · Answer

@Carl MccroryIt was great for me and I recommend it to pretty much everyone, but as @Jon Crosby pointed out, you need to get entrenched in the data.You also want to price out "booked up" properties with lots of reviews. You do this by pricing out a 7 day stay for different months and seasons. This helps you compare data.Also, a point for us it we made sure we wanted to be the "best" Airbnb. Make it look like a model home, not just a "Short term rental...." We wanted people to be "WoW'd" with ours, I think this helped us get right on the data points correctly by providing this quality of product.

Tyler Work · Answer

Hey @Carl Mccrory Those numbers sound about right, but there is a catch. They include cleaning fees and are gross revenue figures. If a 2 BR condo is $120 to clean you'll spend about $1,000/month on cleaning and another $500/mo on PM fees (I've assumed 20%, I have no idea what rates are down there). When you back those out you're at $1800/mo (assuming $40k revenue) and that doesn't include HOA fees, expenses, etc. Is that enough to cover the mortgage on a $400k beachfront condo? Probably not. I use AirDNA data all the time as a starting point but, in general, they can be very misleading. I've reverted to doing my own projections based on the properties I actually manage. @Julia Bykhovskaia has some great points!

Steve Stotter · Answer

@Tyler Work Cleaning fees should be used as a money maker, not an expense. The short-term renter pays the cleaning fee and you charge a markup on that, whether it's 10%, 20%, 30%, etc.

Peter Jetson · Answer

There was a previous thread on the same question a while ago and owners said their actuals were far away from the AirDNA numbers, which were ridiculously overestimated. This cooled me down from purchasing any STVR. I also take PM "predictions" with a pinch of salt as they have skin in the game, and would rather encourage you to become their customers than not... Not all of them are like that of course... In fact numbers varied by 40% according to whom I asked for the same unit!