4 ms·

Scaling Data Science at Airbnb

75 points by uberneo 11y ago

JohnyLy 11y ago
Airbnb is a great example of how date science, metrics, users feedback and growth hacking have been totally integrated to the marketing development of the service. Their amazing growth proved how important it is to focus on people more than numbers.
jmduke 11y ago
I like AirBNB, but this article doesn't seem to espouse much besides "AirBNB has grown like crazy and that makes things tricky" and "We don't treat numbers like medieval doctors treated the humours". Data-driven decision making is a generally dope thing, but the number one pitfall of using data as your number one periscope is that for anything more complicated than Yes/No questions, you can pretty much skew things however you want. [^1] Along those lines -- the article talks about the importance of customer-driven decisions, which is something I 100% agree with, but doesn't really expand on it. It ends with the thesis "Data is the (aggregated) voice of our customers", but what steps do the folks at AirBnB take to make sure that this data is more than pale proxies for voice? [^1] I'm probably the umpteenth person to invoke this in the past 24 hours, but lies, damned lies, and statistics, you know?
- jitl 11y ago
  We marry quantitative data from our data teams with a gamut of user research from our marketing and customer experience teams. We use our qualitative data to check what our quantitative data is telling us, and vis versa. There's generally enough eyes on our statistics to keep them from turning into damn lies ;). Edit: my views not the company's, bla bla etc.
roel_v 11y ago
I couldn't stand the waffling after the first 3 paragraphs, so I have to ask to anyone who suffered thought this: does it say anything about how they use data to make things better for users, rather than for 'growth hacking' (ugh I loathe that phrase) or increasing click rates?
digitalzombie 11y ago
There are so many market buzz words in this article. I'm comparing this to https://medium.com/@samson_hu/building-analytics-at-500px-92e9a7005c83 https://medium.com/@samson_hu/building-analytics-at-500px-92... That was a much better article. Look at me I can throw buzz words! Democracy for data! The article for 500px is much more juicy and have great details on what they do. Perhaps I'm mistaking these data science vocabs for buzz words.
raybesiga 11y ago
Surely one black developer or data scientist good enough to join your team exists out there, AirBnB. Go find them!
- maest 11y ago
  Why?
mrits 11y ago
Can you write another blog about data science?

75 points by uberneo 11y ago

Airbnb is a great example of how date science, metrics, users feedback and growth hacking have been totally integrated to the marketing development of the service. Their amazing growth proved how important it is to focus on people more than numbers.

jmduke 11y ago

I like AirBNB, but this article doesn't seem to espouse much besides "AirBNB has grown like crazy and that makes things tricky" and "We don't treat numbers like medieval doctors treated the humours". Data-driven decision making is a generally dope thing, but the number one pitfall of using data as your number one periscope is that for anything more complicated than Yes/No questions, you can pretty much skew things however you want. [^1] Along those lines -- the article talks about the importance of customer-driven decisions, which is something I 100% agree with, but doesn't really expand on it. It ends with the thesis "Data is the (aggregated) voice of our customers", but what steps do the folks at AirBnB take to make sure that this data is more than pale proxies for voice? [^1] I'm probably the umpteenth person to invoke this in the past 24 hours, but lies, damned lies, and statistics, you know?

jitl 11y ago
We marry quantitative data from our data teams with a gamut of user research from our marketing and customer experience teams. We use our qualitative data to check what our quantitative data is telling us, and vis versa. There's generally enough eyes on our statistics to keep them from turning into damn lies ;). Edit: my views not the company's, bla bla etc.

roel_v 11y ago

I couldn't stand the waffling after the first 3 paragraphs, so I have to ask to anyone who suffered thought this: does it say anything about how they use data to make things better for users, rather than for 'growth hacking' (ugh I loathe that phrase) or increasing click rates?

digitalzombie 11y ago

There are so many market buzz words in this article. I'm comparing this to https://medium.com/@samson_hu/building-analytics-at-500px-92e9a7005c83 https://medium.com/@samson_hu/building-analytics-at-500px-92... That was a much better article. Look at me I can throw buzz words! Democracy for data! The article for 500px is much more juicy and have great details on what they do. Perhaps I'm mistaking these data science vocabs for buzz words.

raybesiga 11y ago

Surely one black developer or data scientist good enough to join your team exists out there, AirBnB. Go find them!

maest 11y ago
Why?

mrits 11y ago

Can you write another blog about data science?