How Did AirBnB’s Sales Process Change With Rapid Growth?
A case study of one of the world’s most successful startups
“Thou shalt not spend more than you make.”
This could easily be the 11th commandment. It’s applicable to anyone from the moment they receive their first bit of pocket money. This commandment is also a well-respected guideline in marketing and advertising — the general rule of thumb is to not spend more than the business you’re generating.
Sales growth, cost per thousand (impressions), click-through, conversion, and open rates are just a handful of the possible metrics closely observed by any self-aware marketer. Seeing an improvement in any of these is like seeing your national team shoot a winning goal. It makes you, your boss, and your neighbor happy, regardless of whether you agree on the current state of politics.
My absolute favorite is the conversion rate. This metric expresses the percentage of visitors, prospective clients, or leads that end up turning into paying customers. It’s such an important metric! It gives a number to the last step before your business can generate profit after miles and miles of planning, brand building, and advertising. The conversion rate going up or down gives you a clear message that something in your sales pipeline is working well (or not).
What Is a Good Conversion Rate for a Startup? What About a Mature Business?
As a data scientist working in marketing, I found myself asking these basic, yet hard to answer, questions. You can find tons of stories online about factors that improve or deteriorate the conversion rate, but they all talk about a vague increase or decrease — not a rock-solid number.
“1400% increase! 80% drop!” Great. But what does your new number mean? That almost always remains an industry secret. There are estimates on what a good conversion rate should be, but once you start digging into where these numbers come from, you quickly realize that they are more guesstimates than anything else.
Recently, I was very excited when I stumbled across digital marketing data from Airbnb during their period of rapid growth between 2010 and 2015. The data was perfect for a case study of the sales pipeline of an early-stage startup that later became a hospitality tycoon. Now, Airbnb has overtaken Mariott, Hilton, and Wyndham combined in the number of rooms listed.
- Airbnb’s conversion rate from registered to paying users was 41% during its rapid growth between 2010 and 2014.
- Customers that entered their gender or age were two-times more likely to complete a booking.
- Airbnb sales dropped by 30% on weekends.
- April, May, and June were the hottest months for domestic (but not international) bookings.
- 69% of customers complete a booking within two weeks of account registration.
- Google’s digital marketing services were more effective than those of Bing.
First Glance at the Data
First, let’s take a look at what 500,000 data points look in a time series.
The dataset contains data on 213,451 users that created an account on Airbnb between January 2010 and June 2014. It’s not a very sophisticated dataset, and it could be recreated by any business that set up basic Google Analytics. It covers the dates of the first activity on the website, the number of accounts created, and first bookings. It contains some demographics about the user, such as age, gender, and the type of device used to sign up. Lastly, it shows some data about the types and providers of digital marketing that attracted users who ended up signing up. All of the users were based in the United States.
When you look at the graph above, you can see exponential growth in the user count with seasonal variation. This shape makes me believe that even though we are almost definitely not dealing with complete data, it could be treated as a representative sample from that period.
In January 2010, Airbnb had been on the market for nearly a year. It had raised $620,000 in a seed round with a $2.4 million valuation after graduating from Y-Combinator. In November 2010, Airbnb reached 700,000 cumulative nights booked and raised $7.2 million in Series A. In 2019, on average two million people stay in Airbnbs every night.
It’s safe to say that Airbnb is an example of a well-functioning and well-funded early-stage startup of Y-combinator quality. It’s a good case for answering the questions about a healthy startup sales pipeline.
A pinch of salt before moving any further — some data is from almost 10 years ago. Back then, the internet itself and the ways we used the internet were quite different. It might be that some of the insights, especially those about mobile users, are no longer accurate.
For example, take a look at Airbnb’s website from 2009. It’s far from its modern, cool, pinkish look, and — as most of the Internet was at that time—it was kind of ugly.
Now, let’s jump right into the sales pipeline of the early stage Airbnb!
When Do the Sales Happen?
I was captivated by the seasonal variation in sales that was visible on the initial time series. And so, I started the analysis by looking at sales versus time. Initially, my intuition was telling me that the summer months would be the most active, with some increase in sales around Thanksgiving and Christmas. I was almost right. Almost, because even though I got the summer right, it seems that the months of November and December are pretty dead for Airbnb.
Average sales in December, the slowest month for Airbnb, are barely 46% of those in the ‘hottest’ month — June. Here, I assumed that the data for the first booking is representative of overall sales (second, third, and fourth bookings too).
Then, I wanted to see if the international destinations rise during the summer months. It would make sense that for anyone living in America, a trip to sunny old Spain would be more likely to happen during summer. Yet, the increase in international destinations was underwhelmingly small. Most of the lines are flat. It might be that between 2011 and 2014 Airbnb was still very small outside of the States, and international offerings by hosts were unattractive.
If there is a variation in the month-to-month sales, what about the weekdays versus weekends? The highest sales occur on Tuesdays and Wednesdays. Sundays place last, and they pose a 30% decrease in sales when compared to the highest scoring weekdays. Is it a reflection of a lower number of bookings by businesses, or is it a sign of the infamous weekend chill? It’s difficult to claim either from the Airbnb dataset, but it’s definitely something to consider.
There was no significant variability between destinations and days of the week. It would be interesting to see if bookings to France spike on Fridays, but no. The destinations booked throughout the week are almost uniformly boring.
One last thing about the timing — most of the users sign up and book right away. In total, 69% of users book within two weeks of registration. This means that most of the paying users create an account to complete a booking and not to look around. Or, they sign up to claim Airbnb’s friend referral coupons.
The Golden, Shining Conversion Rate
Revenue is one of the most important indicators in business. There are many vanity metrics of clicks and opens, but nothing impacts sales and revenue like the conversion rate.
The idea of conversion rate stems from the common depiction of sales as a multi-step process that ends with a sale. On each step of the process, a percentage of customers is lost.
For example, in the case of an online shop, the first step in the sales process would be clicking on an ad and being redirected to the shop itself. Some users would leave the website right away (maybe they clicked the ad by accident), some would browse for a minute or two, and only a fraction would proceed to the next step — placing an item in the shopping cart.
How many of these customers would proceed to checkout, fill in their address, enter their credit card details, and hit the ‘Place Order’ button? Usually a tiny percent of those who visited the page in the first place. The average for the top 500 merchants in the United States is a whopping 3.32% according to Millward Brown Digital.
The 3.32% in this case is the average conversion rate. It’s the fraction of users visiting the website that ended up paying.
For the sake of clarity, I am going to modify the definition of conversion rate in the rest of this article. From now on, whenever I’m talking about conversion rate, I will be referring to the percentage of registered users that end up completing a booking.
So what does the conversion rate look like for Airbnb? On average, only 41% of registered users completed a booking. In other words, the majority of sign-ups never contribute a single dollar to Airbnb’s revenue.
But it wasn’t always like that. As the company grew, the conversion rate steadily dropped from 57% in its early stage to 38% four years later. Is it possible that the Airbnb coupon codes for friend referrals drove this decline? Or is it just a natural phenomenon in any growing business?
Gender has no influence on conversion rate, but not entering your gender does. It looks like those who entered their gender are two times more likely to complete a booking. When it comes to the total number of sales, women, men, and unknown (those who didn’t enter gender) are split equally.
A very similar pattern is visible in the distribution of conversion rates among age groups. Age is not important for conversion rate as long as you enter one.
An interesting pattern occurs when looking at the devices that people use to make bookings. The vast majority of sales were from users who accessed the website via desktop computers. Phones came second and tablets third. Bear in mind that the data is from the period of 2010–2014, when the iPhone 4 was still a cool thing to have. Airbnb only released their iPhone app in November 2010, so it might be that the trend is quite different these days.
Desktop computers not only drive the highest number of paying users, but they also have the highest conversion rate. In other words, the users who registered on Airbnb using its mobile app were 30% less likely to complete a booking. Well, maybe these quick productivity bursts during bathroom breaks aren’t that productive after all.
Digital Marketing — Lessons From Airbnb
The last bit of information that we can learn from the Airbnb is about the tools that digital marketers use. Most of the bookings happen through a ‘direct’ visit — that is by people who manually enter airbnb-dot-com in their browser. Then comes those coming through Search Engine Marketing (SEM). These are the customers who clicked a link to Airbnb advertised on the top of their search engine results for “hotels in Paris.” SEM is a paid service offered by search engines that shows your website on top of the search result list. Even though it can be easily filtered by an extension like Adblock, it seems to be working pretty well for Airbnb.
Search Engine Optimization (SEO) comes third. In this marketing channel, the content of the website is adjusted to appear on top of the regular, non-paid search results. The algorithm of the search engine looks at the website's content and tries to show it to users that — in the case of Airbnb–looked for a short-term room rental.
This opens a new possibility of using the Airbnb dataset — we can compare the search engines in how good they are at showing Airbnb’s website to people who look for Airbnb. The higher the conversion rate among people coming from a given search engine, the better the engine should be, right? So who do you think is going to come first? Google, Yahoo, or Bing?
Google it is. Right after a mysterious category of “other,” which includes non-English-speaking search engines (like the Chinese Baidu). I’m very confused about why Bing still exists. Bing comes built into the Internet Explorer browser on Windows, and most people use it only once to search “Google Chrome download.” On top of being only slightly popular, it does a pretty bad job of bringing in paying clients. Take a look at this data comparing Google’s and Bing’s paid-for Search Engine Marketing.
Yes, Bing brings customers to Airbnb. But why is it 15% worse than Google? Is it 15% cheaper? That might remain a mystery forever.