How to Start Your Airbnb Business?
Suggestions for Airbnb newbies to start their own business — a Seattle case study
Overview
The purpose of this project is to provide the suggestions for Airbnb newbies to start their business.
To do so, I extracted the insights from the datasets. The finding includes:
- The pattern between location and price
- The pattern between time period and price
- The attributes that customers care about
After doing the analysis, I turned data insights into suggestions. Also, I provided a reference table for Airbnb hosts to set prices based on their listing locations.
These help Airbnb newbies to start their business by knowing how to set a good price and how to be a customer-focused host.
In this project, I choose Seattle as our case study, and I used Seattle Airbnb dataset in 2016 from Kaggle as my research dataset. This dataset is originally from Inside Airbnb, which including the price, availability, review score and related information of each Airbnb listing in Seattle in 2016.
Introduction
If there are empty rooms in your house and you want to start a small business by renting a house for short-term needs, it is quite easy and convenient nowadays. Thanks to Airbnb, everyone could start their rental business by becoming a Airbnb host.
But how? How to start your Airbnb business as a beginner?
How to start an Airbnb hosting as a beginner?
There are two key actions for an Airbnb newbie to start with, they are:
- Setting a good price
- Be a customer-focused host
Setting a good price
Price setting is the first problem every beginner encounters. The goal of a business is to make profit, Airbnb hosts have no exception. To do so, Airbnb hosts need to track the cost and revenue of their listings in order to maximize the profit.
Cost of a listing includes the opportunity cost of the house, decoration cost, cleaning fee, administrative expenses and so on. It is case by case, and differs from every listing. As a result, it is hard to extract the general insight from our dataset.
However, revenue is where we can put effort on. Revenue of a listing mainly comes from the rental price which Airbnb hosts could set up by themselves. We can extract some general patterns from our dataset to help Airbnb hosts set a reasonable price.
Some common factors that influence the price setting, such as “Location”, “Time period”, would be a great start.
Be a customer-focused host
Being a customer-focused host is crucial since it will make a positive loop on your business. If you know your customers well, you could put more efforts on what they care about. This will attract more customers to book you listings, finally, brings you more benefits.
To know what customers care about, we reference the Airbnb super host program.
Airbnb has a special program called Super host. According to Airbnb website, super hosts are those who provide excellent experience for their guests, and meet the requirements formulated by Airbnb.
Super host is like an official certification, which conveys the information: “This listing is really great!” to the customers.
We could use our dataset to compare super hosts to general hosts, and find out what kinds of attributes customers might concern the most about the listings. This will help Airbnb hosts increase their quality of the listing, and attract more guests.
Research Question
To know how to set a good price, I would like to know whether location and time period influence the price setting. If there is a pattern between location and price, or between time period and price, we can adjust listings price based on these patterns.
To be a customer-focused host, I would like to know what kinds of attributes distinct super hosts and general hosts. This would help Airbnb hosts to learn from the role model, and focus on what customers care about.
So, here’s the three main research questions:
- Does the location influence price setting?
(1) Do different neighborhoods in Seattle have different prices?
(2) Can we have a reference table that helps Airbnb hosts to decide their price based on neighborhoods - Does the price change periodically?
(1) Does the price change monthly?
(2) Does the price change between weekdays and weekends ?
(3) Is the price higher on holidays than other days? - What are the main attributes that super hosts are different from general hosts?
Let’s start to answer the above questions!
1. Does the location influence price setting?
According to our dataset, there are 87 different neighborhoods in Seattle. To compare the price on a fair basis, I calculated the mean price per one night per one bed as our indicator for further analysis.
1.1 Do different neighborhoods in Seattle have different prices?
After calculating the mean price in each neighborhood, I put the result on the choropleth map as below. We can easily see that the mean price is in the range from $50 to $144. The highest price is about three times more than the lowest price, which indicates that the mean price significantly differs among different neighborhoods.
Compare our choropleth map to the Seattle census area map, which is provided by Seattle.gov. We found that the Downtown/ Queen Anne/ Magnolia area has a higher price than other areas; south of Beacon Hill has relatively low price than other areas.
The reason why the mean price varies in different locations might be the distance to the city center, the house price differ in different areas and the demand of the customers preference.
1.2 Reference table that helps Airbnb hosts to decide their price based on neighborhoods
According to our map, we found that location matters with the Airbnb price, as a result, I created a table which contains the statistical information of price in each neighborhood.
There are nine columns in the table as below:
- neighborhood: 87 neighborhoods in Seattle
- mean_price: average price without outliers
- min_price(outlier): the lowest price which including outliers
- min_price: the lowest price without outliers
- 25th percentile: 25% of prices is lower than this value
- 50th percentile: also known as median. 50% of prices is lower than this value; and 50% of prices is higher than this value
- 75th percentile: 75% of prices is lower than this value
- max_price: the highest price without outliers
- max_price(outlier): the highest price which including outliers
Notes:
1. An outlier is a data point that significantly differs from other observations. As outlier might have large influence on the mean price, I decided to remove outliers.2. I still remained the maximum price and minimum price which including the outliers as these outliers provide values for those who has special listings which differs from general listings.
This table could give Airbnb hosts the price reference based on their listing locations.
Take “Adams” as an example, reference to the table, we know that the price is from $25 to $150 without outlier, and the mean price is $80.16. Since the mean price ($80.16) is higher than 50th percentile ($77), this indicates that the price in Adams has a right skewed distribution, which means there are higher dispersion in mid to high price.
To consider how to set price in Adams, firstly, hosts need to determine which class their listing is.
Let’s say I am setting my listing as a high class listing, then I would like to look up the price range between 75 percentile to max price since it contain 25% of high price listings. As a result, I might consider setting a price between $105 to $150.
Notes:
Some of neighborhoods have NaN value in “mean_price”, “min_price(outlier)” and “max_price(outlier)” columns, this is because these neighborhoods only has one listing. It is not reasonable to use only one price to represent the mean price of whole area, also, it has no outlier since it is the only one listing.
Let’s take a quick look at the highest and lowest neighborhoods.
According to our table, the top 3 areas which has the highest mean price are Central Business District ($143.44), Pioneer Square ($137.92) and Pike-Market ($137.45); the top 3 areas which has the lowest mean price are Brighton ($51.33), Holly Park ($51.67) and South Beacon Hill ($52.67).
2. Does the price change periodically?
As I did above, to figure out whether the price changes periodically, I used the mean price per night per one bed as the indicator.
2.1 Does the price change monthly?
After calculating the mean price in each month, we found that the price is fluctuating by month.
As the table and figure shows, July has the highest mean price, which is $101.47; January has the lowest mean price, which is $90.53.
Mean price is fluctuating seasonally. Between June to August, the mean prices are all over $100, which might indicate that the demand in these three month is higher than other months. This is reasonable since these three month is between summer vacations, as a result, It might bring more visitors/travelers to Seattle.
On the other side, between January to March, the mean prices are all under $95, which indicates that the demand is low in these three months.
2.2 Does the price change between weekdays and weekends ?
After calculating the mean price by week days, we can easily see that the mean price on Saturday is the highest, which is $100.61, followed by Friday ($100.5). It indicates that the demand on the weekend is higher than weekdays. The mean price on other days have no significant difference, they are around $95.
2.3 Is the price higher on holidays than other days?
In my intuition, prices might be higher on holidays, but I am not sure whether this statement is true, as a result, let’s check about it!
I selected “Federal Holidays” as our target. We can find that Independence day has the highest mean price among all the holidays, which is $100.11; and Martin Luther King, Jr. Day has the lowest mean price, which is $88.97.
Since we know that the mean price fluctuates by the month, in order to compare the price between holidays and other days, we need to calculate the difference between the holidays and its monthly mean price.
As the table shown below, surprisingly, most of the holidays have a lower mean price than monthly mean price. Only on Veterans Day and Memorial day, the mean price is higher than the monthly mean price. These results indicate that holidays did not make the price higher than other days.
3. What are the main attributes that super hosts are different from general hosts?
The best way to know what distinct super hosts and general hosts is from customer’s feedback.
Our dataset includes seven columns related to customers’ review score, they are:
- review_scores_rating
- review_scores_accuracy
- review_scores_cleanliness
- review_scores_checkin
- review_scores_communication
- review_scores_location
- review_scores_value.
review_scores_rating is the overall review score; others are individual items score.
First, let’s take a look at how many super hosts are in our dataset.
According to our dataset, we have 778 super host and 3038 general host. Super host is account for about 20.39% among all the hosts.
The number of review_score_rating for super host and general host are 751 and 2420, respectively. Dividing these numbers with the number of hosts, we get the response rate for super hosts is 96.53% ; and the response rate for general hosts is 79.66%.
We found that super hosts receive higher response rates compared to general hosts. This indicates that customers are more willing to give the review score to the good listings.
Next, let’s look at the review score. the overall score of super hosts is 97.4; and the overall score of general hosts is 93.65. We found that the difference is less than 4, but it can not explain anything since most guests tend to give high review ratings (higher than 90) as the figure shows below.
Although it is hard to describe the meaning of the score difference between super hosts and general hosts, we still can compare each item and find out which items that customers care the most.
We found that the highest score difference is in Cleanliness, followed by Value and Accuracy. This indicates that these kinds of attributes might be the reason that distinguish super hosts and general hosts.
To be a customer-focused host, we found that Cleanliness is the most important thing the hosts should care about.
Conclusion
For a Airbnb newbie, it is crucial to know how to set a good price and be a customer-focused host.
To set a good price, we found that price changes geographically and periodically. As a result, it is recommended to adjust the price based on the listings locations and the time patterns.
Location finding
- I have created a table based on different neighborhoods in Seattle. It contains statistical summary of price as below.
Please follow below steps to use this table:
Step 1: Lookup the neighborhood columns that match to your listings location.
Step 2: Based on your listings setting (i.e. cost, types, services), determine which class your listing is belong to.
For example,
If I set up my listing as a simple room for backpackers, I would determine my listing is belong to economical class.If I spent lots of money in decorating my listing, and intend to service VIP customers, I would determine my listing is belong to high-end class.
Step 3: Based on your listing class, lookup statistical columns which related to your class.
For example,
If my listing is belong to economical class, I would lookup “min_price” and “25th percentile” columns for reference.If my listing is belong to high-end class, I would lookup “75th percentile” and “max_price” for reference.
Step 4: According to the reference price, decide your listing price
Time Pattern finding
- June to August is high demand period, with mean prices all over $100; January to March is low demand period, with mean price all under $95. It is suggest to raise your price from June to August and lower your price from January to March.
- Saturday and Friday is the two high demand days in a whole weeks. It is suggest to raise your listing price during these two days.
- Holidays have no specific influence on the price. As a result, there’s no need to adjust price during holidays.
To be a customer-focus host
- Cleanliness is the most important items that distinguish super hosts and general hosts. Followed by Value and Accuracy. It is suggest to focus more on these items.
If you are interested in the code used for this analysis, please visit my GitHub repository.
