Startup Class Lecture 16 Notes – Emmett Shear

These are my notes from Lecture 16 of the Startup Class taught by Sam Altman. Emmett Shear, gave a talk on how to run a user interview.

My notes from the previous lectures are here.


  • Started Kiko Calendar with Justin Kan
  • Sold it on eBay
  • Didn't talk to users and didn't know much about calendars
  • Second start up was
  • Where there own users in that case
  • But they needed to broaden the context/audience of
  • That's when they actually went out to talk to users


  • Determined that broadcasters were the most important people
  • Realized that users would follow the content and the broadcasters
  • Its important to decide which users to talk to – need to find the target users
  • Talked to both the viewers and the broadcasters

Conducting a user interview

  • Lets work though an example of how to conduct a user interviews assuming the product is a ;ecture focused note taking app
  • Step 1: Think about who the best users to talk to before you decide to build this out are (Think about the 3-5 types of people you would talk to, and think about which the most important type is)
    • college students divided by major/are of study and studying style (take a lot of notes vs. not)
    • IT administrators
    • Parents of kids
  • Step 2: Find them and talk to them
  • Don't focus on talking about optimizing user flows in the early set of interviews
  • Users think they know what they want but they often don't
  • Step 3: Based on the insights you pull from an interview, think about the features you would want to build on top of/instead of the existing solutions
    • E.g, Google docs but with smaller "sticky" notes
  • Step 4: Once you have an idea, you want to actually go out and validate the idea. Again go out and talk to to users
    • Advice for these interviews:
      • Build out some version of hte product such as user flows/ a mockup and take it to users and see if they're excited about it
      • Never ask if the "Is this feature actually good or not?" because everyone just responds saying its great.
      • The money test works well: see if you can get them to give you their card.

You get the horseless carriage effect where you're asked for a faster horse instead of asked to design the actual solution to the problem.


  • Want to talk about how interviews helped us at Twitch
  • Spoke to a lot of broadcasters from collecting 20+ pages of feedback.
  • Saw some common themes, and detailed feature requests, but none of these features were so big that they stopped using the service.
  • Then did some competitor user interviews speaking to people who weren't using the service but were on competitor broadcasting services.
  • The features they listed were the "killer" features that resulted in them not using
  • An even more important thing we did is talk to non-broadcasters, and see why they weren't using a service like ours.
  • Talking to competitors users is normally only a short term win, but talking to non-users enables you to learn about how you can expand the size of the market.
  • Need to learn about what features would be needed to get users who have never used your product to try it out if you want to expand your market
  • Built broadcasting into Xbox and PS4 to make it easier to broadcast based on learnings from interviews with non-users.
  • Learn about the main goals and build features that help users that meet those goals.
  • Don't necessarily build features exactly the way users ask for it.

Tips for user interviews

  • Don't show users your product: It puts things in their head. You want unfiltered views/opinions.
  • Talk to who you need to talk to, not who is easily available.
  • Don't compensate people for interviews – find the people that care enough about the problem that they are willing to talk about it
  • Definitely try to do the interviews in person or using Skype rather than over email – enables you to get interactive feedback and learn a lot more.
  • In terms of groups to focus on first, users using competing products is a good start, since you just need them to switch rather than create a new behavior in new users.