My Notes on Julie Zhuo's presentation - Data-Informed Vs Data-Driven Design

Facebook has to be one of the most interesting applied examples that can be presented because of the hug number of users it currently has.

I had the opportunity of listening to Mark Trammell talk about Twitter on last year's Web App Masters Tour at Philadelphia, and now I was able to listen to Julie Zhuo talk about Facebook.

Here are my notes on what she had to say:

• Data helps us understand how products are used and how they can be improved
• Photo uploading had been outsourced and the code was a mess. It wasn’t reliable and it was poorly designed.
• Facebook redesigned a very cool experience of photo uploading but it wasn’t that accessible for users because of the need to download a plugin – Only 34% of the users tested were successful
• Change to a more native feel, with a simple workflow, increases 11% of success rate.
  
  
• Data can be used to sanity check changes to a product
• The composer example – changes in the main status update process in an attempt to increase sharing.
  
  
• Qualitative data also has a value in these tasks, but it’s mainly for understanding how users feel.
  
  
• You can spot specific use cases that are common with the interaction between your users and the product
• New moms are a very intense use case amongst Facebook users.
• Enterprises that want social media presence are also very active users. Not only big companies with social media departments, but also small companies with lesser knowledge of it.
  
  
• Growth of Facebook users – Process fragmentation can improve the experience
• Sign-up process fragmentation brought a 3% increase in sign-ups (which is equivalent to 9 million more users a year, the size of the population of Sweden).
  
  
• Deactivation process – Giving it the feeling that your friends on Facebook will miss you reduced the deactivations by 7%, which is equivalent to 1 million users a year.
• Facebook created a team focused on user engagement.
• Reads and writes were the first statistics considered, but 85% of them are generated by the 20% of the users, which rendered them not as useful as defining data.
• They believe real innovation invariably involves disruption – The news feed addition was hated at first, but it was related to their aversion to change.
• With such a stratospherically big amount of users, changes to the product can’t just go live without previously recollected and solid data.
• The greatest mistake, for them and for everyone, is taking no risks.

She also talked about how they considered the Dislike button but decided it was a risk not worth taking since you could better express that 'dislike' through comments and it was more appropriate way to do it. I found that interesting, since a lot (and that's A LOT) of people have wondered about it in the past.