AirBnB's Pricing Algorithm | Hacker News Reader

AirBnB's Pricing Algorithm

spectrum.ieee.org

106 pointstaigeair38 comments

Talk about burying the lede:

"That’s why we’ve released the machine-learning platform on which it’s based, Aerosolve, as an open-source tool."

And worth noting the link in the article doesn't work - it's meant (I believe) to send you here: http://airbnb.io/aerosolve/

> We selected a machine-learning model called a classifier.

Almost all machine learning tools are classifiers of some sort. It's like saying: we selected a car technology called "wheel".

I think you have an overly narrow view of machine learning or a very broad view of what classifications are.

Completely solving the problem of pricing in term of getting it right the first time is more difficult than predicting the weather. At least with the weather you can go outside and learn if it is raining. You can't just go outside look at an item and give it an optimal price that maximizes the amount and/or the amount of time you need to sell it within.

I work at a company that has pricing algorithms more complex than what retail stores use. I don't think they are useless. But, we all know they aren't "right", and they only got to be what they are through testing them out with actual customers. And some customers work to determine your algorithm and then use it to their advantage in a big way so you have to change it again, which affects other customers negatively, and the cycle continues. You can't just A/B test pricing and assume that will work in a hands-off fashion either, because customers will take advantage of that.

The complexity of pricing is that the best pricing only partly has to do with the thing being sold. You could argue it has nothing to do with the thing being sold. It has everything to do with the buyer: their needs, their resources, and their perception of the product, service, seller, the entire sales experience, market, economy, weather, etc. You name it and it probably has some affect on what the best price would be.

It would be a shame if AirBnb would price gauge based on profiling customers. I don't like, when I enter a shop, to be asked my nationality before they quote me a price.

This is already happening (depending of the country, your skin color might also make the price go higher), tourists typically pay more for a lot of stuff than what a local would pay. Though, sellers are usually a bit more subtle than "Hey, give me your passport, so I can tell you how much it costs". This is how a lot of people make a living in a lot of places. In the tourism industry, I think it's completely understandable and reasonable to base a price based on what the buyer is willing to pay, more than what it really costs.

Would this not get AirBNB into trouble?

Price fixing in a cartel is illegal anywhere in the western world for good reason.

The practice described in the article has nothing to do with price fixing.

This is not price fixing or anywhere even slightly related

AirBNB is colluding with its contractors to influence and set similar prices across competing room vendors.

How is that not exactly price fixing?

For one, because AirBnB has multiple competitors. HomeAway is the closest thing, but traditional BnBs and hotels are playing in largely the same space.

For two, they're not actually setting the prices, the property owners are doing that. They're recommending prices, which is very different.

Feels kinda weird to be defending AirBnB, since I work for Homeaway.

ETA: Probably the closest analogy would be the infamous Manufacturer Suggested Retail Price, which is quite legal as long as they don't actually force the retailers to stick to it.

Price fixing is when competitors agree to set prices. If AirBnb and Hilton and Marriott started agreeing on prices, then it would be price fixing.

This isn't price fixing because companies aren't considered their own competitors.

It might be a little fuzzy if the people posting rooms on AirBnb are considered competitors, but even then it's not price fixing because there's no requirement to use the algorithm's suggested price, and no single owner can influence the general price of rooms, because there's still hotels, motels, non-AirBnb rooms, etc.

I still don't understand how's staying at an airbnb rental is more compelling than staying at a hotel. From my experience, Airbnb has only a single upside: you get to stay at someone else's place, which has a quite different vibe from a hotel. One the flip-side, compared to a hotel, the prices are considerably higher, you have to negotiate with a host on checkin time (most of the times), and overall you don't have any guarantees on the quality of your rental the same way you have with a hotel chain. Correct me if I'm wrong, but Airbnb has not loyalty/rewards program available either, which could be quite beneficial for frequent travelers.

The whole "local experience" seems to be a mere marketing fluff. The Airbnb experience doesn't seem to imply much interaction with the host, the same way which it's predecessor Couchsurfing did. Once you checked into you Airbnb rental, you are on your own, the same way as if you stayed at a hotel, but with even less hand-holding. Hotels at least provide a (poor) local recommendation service.

I'm wondering what a typical customer of Airbnb looks like.

It's kind of fun to stay in somebody else's place. I do think there's something to be said for that. And, it IS frequently cheaper than a hotel. Many times, depending on where you're staying, it could also be more convenient. It's a B(nB) for the masses, and I don't really mind that.

There of course is still room for the hotel industry, and I don't really think this is making a dent in their business. It's just different, and that's ok.

Outside of US it seems that the main customers of Airbnb rentals are Americans, and therefore often you see a price tag a few times higher than a price of a hotel in the area. I've never seen an Airbnb cheaper than a hotel. If you want a local price, you got to use local websites, or simply ask around or walk in a hotel/bnb. Nothing beats an offline price negotiation with a hotel manager, especially for long-term stays.

Out of curiosity, where did you find those cheap and convenient airbnbs?

I went to Tokyo and found a room in Ginza for $35 a night on Airbnb right next to the subway station. Another one in Silverlake LA for ~$40 a night.

Just as an example, I booked an AirBNB in Brighton UK for around $100/night with hotels in the area running $200+/night.

I travel frequently and have tried replacing the usual hotel chain with an airbnb. I totally agree, the hassle of dealing with hosts and being left disappointed a few times, it's not worth it (for me).

Where I found airbnb more useful (and cheaper) is when my family is in town and we're traveling somewhere and we usually book a whole apartment/house with multiple bedrooms. That is usually cheaper than booking 2 or 3 hotel rooms. However, in that space there are lots of alternatives (Homeaway, VRBO, local bed & breakfasts with their owns websites).

My impression is that airbnb, in the end, is just like those other sites. More recently, I've found that more often than not, the place you book (especially in popular areas) is owned or sublet by someone who's making a business out of short term renting and with that, the whole local/experience/sharing charm is gone.

Short term vacation rentals are definitely is where the money are at.

Also, for my experience, the niche if affordable mid/long-term rentals (1-2 months) appears to be underserved. Sure I can negotiate a monthly rate with a hotel, but unless they are experiencing a lack of demand, it's not worth it for them - they can make more money on short stays.

The cheapest alternative for finding a sublet appears to be local sources (websites, asking around locals), but that could be too much hassle, especially if you don't have time for it.

Airbnb listings sometimes provide a monthly rate, but again in my experience it's not very competitive. Being not very price sensitive, I'd still go with a previous option - it has it perks of local immersion.

Homeaway and VRBO are the same site now.

Also, when AirBNB launched, their web interface was vastly better than VRBOs. VRBO is a lot better now, but it's still clearly an older site with some improvements.

I guess I fall into the "typical AirBnb customer".

I don't like hotels so much. It feels fake and uncomfortable to me. I don't like someone walking into "my" room and tidying it up every day. Or get some fancy magnetic key, and copy&paste furniture*

I feel much more at home at someone's, erm, home. I can cook something for myself. I am pretty much guaranteed a good internet connection. There's usually a washing machine, so I can wash my own clothes instead of paying silly prices per-item, or waiting at a laundrette. The hosts frequently leave some good tips for places nearby. It just feels so much better to me than most hotels.

I'd happily pay more for all of it, but in many cases prices are on par or even cheaper, compared to the size / value I feel I'm getting.

* sure, there are some cool hotels, and I like those too. But those are typically the exception.

I'll typically use Airbnb for business travel - it's always been cheaper than the average hotel (after some searching), and it fits well with my preference against neatly-packaged things.

That said: in my experience, the "local experience" schtick is pure BS. Airbnb is at an awkward midpoint between hotels and Couchsurfing: it lacks both the predictable experience of the former and the possibilities for awesome interpersonal/cultural exchange of the latter. Couchsurfing is not a real predecessor to Airbnb; the two are not at all the same thing, and confusing the two does a real disservice to what the "sharing economy" meant before it was conflated with the financial economy.

I used the word "predecessor" with a more of an ironic connotation. The awkward midpoint is spot on.

Like it or not, but all of the hospitality exchange networks seem to have failed so far (provided that a success implies creating value, with an indicator being profit and/or number of members). Even Couchsurfing, the most "successful" one of those was originally started as a service to find a free place to stay. It just happened that a great community formed around it. If we leave the community out, Airbnb in its essence is a "paid" hospitality exchange network.

In a way Airbnb derives from the ideals of sharing economy, being a more of a peer-to-peer marketplace which promotes itself as a hospitality exchange service.

I've just been burned by this. I'm staying in Eastern Europe for nearly 2 weeks. Booked a place on airbnb because I wanted an apartment instead of a hotel room. It cost nearly $1500.

Out of the more than 15, 5 star reviews, some noted there was noise, but almost every single one praised the sound insulation and said "the noise wasn't a problem for me" or words to that effect.

Well, it was a total problem for me so I contacted airbnb. It wasn't an emergency so I emailed them. Their website said I'd hear back within 2 days. That would have been fine. I just wanted to know what my options were.

After 4 days (4 sleepless nights to be precise) I emailed again asking what the hell was going on. Eventually, 5 days after emailing them the first time, I got my first response at all. They said I had to speak to the host, send them evidence, etc. I told them they could shove it and I'd be booking into a hotel because they'd taken so long.

If this had been Amazon, they'd have said "Sorry there was as problem, which of these other places would you like us to transfer you to?" I pointed out to them that a) the host wouldn't be able to make the bars downstairs any quieter, and b) they'd taken nearly half my holiday to even respond. Initially I'd only wanted advice on what my next steps could be, but after dragging their feet for so long, things moved on and I booked a hotel instead.

Their whole chain of responses to me consistently read like they were passing the buck, making excuses, and generally trying to blame me for not doing research (erm, I read those 15+ reviews), or contacting the host (despite the fact I initially asked for advice). It seemed they were just stalling. I pointed out that since I'm now in a hotel, I didn't have any inclination to go to the flat and record the noise, especially since their terms say they'll only refund you for unused nights, not for nights where they were too understaffed to reply on time (they blamed the slow response rate on "there are a lot of people travelling at this time of year". Yeah, and I'm one of them).

So all things considered, I'll be thinking twice before using them again. It's a shame because over the last 3 years I've spent over $6000 USD with them and had one particularly good experience. But their whole attitude smacks of "we're just the platform, just pay our service charge and piss off".

@nnd - I can't see how it's relevant, but since you ask, I'm told I look like a character from shrek...

This is anecdotal, but staying in Airbnbs made my cross country drive from Richmond, VA to Seattle much better than if I had stayed in hotels.

1) The prices were significantly cheaper. I stayed in a room in a beautiful home in Spokane for $35 a night. Sure, you can find rundown 2 star "hotels" for $50 - $60 a night, but trying to find something of comparable quality in a hotel would increase that substantially. Where Airbnb was great was in little towns and areas in the midwest. I stayed in the entire top floor of the most amazing, one hundred year old creamery turned bar turned home in Fertile, IA for $65. Hotels in any nearby city / town were much more and couldn't come close to matching the value.

2) Being able to stay in interesting places really is a benefit. I stayed with an amateur chef in Cincinnati who cooked a fantastic breakfast that otherwise would have cost me $15 at a hotel. I stayed with Yellowstone guides in Belgrade, MT that told me exactly where I need to go the park next time I visit. When I was tired some days and wanted the hotel experience, hosts were more than happy to let me have that.

3) If you want to be really sure you're staying somewhere great, look for superhosts. I never had an experience that didn't match up with the reviews that a host received. Reviews on airbnb actually make hosts and guests accountable for the quality of the stay.

4) The local experience is what you make of it. Hosts typically will let you control the amount of interaction you want to have. It's impossible for them to predict what everybody wants, so you have to communicate.

All of your points don't match up at all with what my experience has been. I love the experience and quality Airbnb helps provide and hotels are always my second choice now.

You're forgetting families, folks who prefer to cook their own food, and the fact that in many places, AirBNBs can let you sleep more people than a hotel room for less money. Also, depending on your destination, there might be AirBNBs where there aren't hotels (in city neighborhoods, near your relatives, etc.).

Hotels still work for some folks, some of the time. Likewise AirBNBs/VRBOs, etc. Different benefits, but they're definitely there.

Some Airbnb listings offer discounts for 7+ day or 30+ day stays, and the savings can then be substantial. Even for shorter stays, it can often still be cheaper than comparably clean+handy hotels, especially if you factor in that things like Internet and parking are almost always included for free.

I've found evaluating a listing basically means reading every review and assuming any negative element mentioned will be arbitrarily bad, even if the reviews are positive and say it's "no big deal." I remember seeing one listing where multiple reviewers complained there was structural damage to the building through which feral cats entered and urinated on guests' property but said it wasn't a major problem.

If the reviews mention street noise, assume there will be drunken revelers outside until dawn all night. If the reviews mention shedding pets, assume every surface will be covered in cat or dog hair. If they say the host is overly talkative, picture a humorless version of Robin Williams trying to be your personal genie.

If the negatives don't bother you--you love animals, you don't mind noise, you're not bringing a car so nonexistent parking isn't an issue--then go ahead and rent the place.

I use AirBNB on nearly every trip my family takes. AirBNB has the best interface for vacation property rentals. VRBO and HomeAway have improved a lot in the past few years, so it's less of a win now, but for an important window, the dynamic AirBNB interface blew the old static VRBO interface out of the water.

As for why I prefer an apartment rental, that's entirely down to family stuff. With two adults and two kids, we can share two queen beds in a hotel room, but that's tight. When the kids go to bed at 8pm, I have to take my computer down to the lobby to work or play. We can get two adjoining rooms, but that's expensive. In an AirBNB, I close the door to their bedroom, and sit in the living room right next door with my wife and all the amenities of the listing. When they get up at 5am due to jetlag, I can sleep in in the master bedroom while they play in the living room and watch TV. AirBNBs also usually have full kitchens. That's nice both for making full breakfasts in the morning (my wife likes eggs, and it's a lot cheaper to make a quick scramble on the kitchen stove than go out), and for re-heating the kids leftovers from the restaurant meal we had the night before.

Hotels also generally provide full daily housekeeping services. That adds to the cost. I don't need someone to make my bed for me every day. I can do it myself in 2 minutes.

It sounds like you are ignorant of airbnb's services and have likely never used them.

> I still don't understand how's staying at an airbnb rental is more compelling than staying at a hotel.

> the prices are considerably higher

I am in the middle of the valley and I rent bedrooms for $60-$120 depending on season and demand. I've seen entire apartments for less than $100 a night in my area as well. When I first moved here I stayed in a shared space for ~$40 / night for a few weeks.

I challenge you to show me a nice hotel for less than twice that price. The Sheraton near my office is usually > $350 / night.

> Once you checked into you Airbnb rental, you are on your own

Any good host will be more than willing to provide recommendations on local services. Many will also already have their guidebook on airbnb filled out which has a map of local recommendations. Airbnb is built on two way feedback between hosts and guests. As a host would you really risk a poor review over not spending 5-20 minutes to help orient a guest?

We've been an airbnb host with multiple properties for a few years now in Austin. We've done our own version of dynamic pricing (manually), and the graph highlighting Austin in the article is spot on. The "Pricing Tips" tool has a long way to go, but its relatively close (within 20-30 percent). It still hasn't caught some of the big events, mostly because it only looks forward 3 months, and for popular events like SXSW, we usually book 4-5 months out at very high prices. That said, big props to ABNB for trying this and putting it out there. Hopefully their algorithm takes into account what humans are actually doing with their own pricing, and using that in it's calculations.

I found that the Airbnb Price Tips feature works pretty well in mature markets but not in new locations. In some apartments we manage in Kenya for instance, there are two apartments which are exactly the same. One has no reviews and the suggested price is around 48 €. The other one has 3 reviews with 4.5/5 ratings and the suggestion is 30 €. So, it's good to look at but you always need to see if it makes sense to you. Anyway, amazing job, they left the competition 10 years behind.

This is the difference between airbnb and google. Google would have Hal Varian look at this. Airbnb has a music major re-invent the wheel (not sure if the managed to re-invent the whole thing).

There is a whole field of study called pricing mechanisms. For example, second price auctions work well because of a reason.

Why attack Airbnb for trying out a new approach?

Also, Google is not that great at some things they do. For example, Google is somehow much worse at navigation than Waze even though they were at it for much longer and had tons of data. Now they even have Waze data and are still worse at it than Waze.

Is it just me, or is google maps even worse than Apple maps at this point?

I think Google maps has some top-notch turn-by-turn navigations. One thing I particularly like is how detailed the voiceover is, i.e. recommends you a lane to stay in, etc This is one thing i wish Waze would be better at, because overall it seems like it's able to find faster routes than Google, although sometimes I wonder if it's only perceived, due to the way they gasify it.

Apple maps are absolutely horrible in turn-by-turn. Either I'm driving too fast, or the turn announcement is way too close.

so what you are saying is that someone with a creative background also managed to do non-trivial data science work that affects more than a million decisions made?

I admire that, a lot, actually.

I can't stand AirBNB. Prices are ludicrously high. I feel like prices are higher just because it's a nicer cuter interface than Craigslist. Obviously I can't prove this. Also wish I could filter by monthly for everything not map a certain daily range to a monthly range in my head.

I feel Airbnb offers a value-added service and, in my personal experience, that has justified the extra price. Also keep in mind that the spirit of Airbnb was short-term vacation rentals; not long-term tenancies.

With Airbnb I can evaluate the host's social "credit" to some degree (have they connected with LinkedIn? What references have they provided?) and transparent straight-up reviews from past guests. Having Airbnb act as an arbitrator when things go wrong is priceless.

Reply on news.ycombinator.com