Startup Class Lecture 4 Notes – Adora Cheung

These are my notes from Lecture 4 of the Startup Class taught my Sam Altman. Adora Cheung, the founder of Homejoy gave a guest lecture on building product, talking to users and growing, based on her learnings building Homejoy.

Also, my notes from the previous lectures are here.

Prerequisites for Starting a Startup

  • You should have a lot of time which you can spend to concentrate on your startup.
  • If you’re in school, it’s better to have a couple of days a week entirely, than a few hours every day. The reason for that is that it’s a lot easier to focus on it when you have those large time blocks, and being able to immerse yourself in it is very important.

Starting a Startup

1. Decide what problem to work on

  • You should be able to describe the problem your idea is solving in one sentence.
  • Two considerations
    1. Are you passionate about the problem?
    2. Do other people face this problem?
      • You answer this one by going out and talking to users.

2. Come up with a solution

Once you have a problem and are able to state it, the next step is to think of solutions, and the best way is by immersing yourself in the industry.

How to Immerse Yourself in the Industy

  • Become a cog in the industry.
  • It sounds counterintuitive, since most people say that you shouldn’t be a player in an industry if you really want to disrupt it.
  • That might be true if you spend 20-30 years in the industry, but you should look to spend only a couple of months. That should be enough time to learn the details.
  • You should become obsessed about the space, so much so that you know what everyone is doing and everything that is going on. Reading about the space helps a lot.
  • If there’s a service element to it, you should go and do that service.
  • Adora learned to clean when she started Homejoy, and in doing so learned about how a local cleaning company works, which gave her valuable insights which she leveraged in building Homejoy.

In short, you should become an expert about the industry.

3. Before you start building

A. Identify customer segments

Eventually, hopefully everyone will want to use your product/service. But when you start, you want to target a certain segment so you can optimise for them.

B. Storyboard the user experience

This involves thinking through the entire experience which starts from how the customer will find you all the way to what happens after they use the product/service.

4. Build

So after all of this, you build. Build an MVP, which is the smallest feature set to solve the problem you’re trying to solve. Talking to users, storyboarding helps you figure out what the feature set it.

5. Getting your first few users

You should have your product positioning down. You should be able to explain to them in one line what the functional benefit of using your product/service.

Who are the early users

  • You
  • Family, Friends and Coworkers
  • Online communities (Reddit etc)
  • Local communities (Mailing lists etc)
  • Go to places where there will likely be people facing the problem you’re trying to solve.

6. What to do after you have your first few users

  • Make sure they can contact you (email, phone number)
  • Reach out to them and do interviews/surveys
    • Have a conversation with them and get them to feel comfortable and share their honest feedback1 with you
  • You should be also seeing how the product is doing at a macro-level: track customer retention, track net promoter score2.

Over time, reviews and retention should go up as you incorporate feedback and add new features. This means you are doing a good job.

7. Growing the Userbase

Things to keep in mind

  • Optimise for the next stage of growth: build the features that are needed now
  • Manually do things yourself before figuring out whether it’s worth automating it early on.
  • Temporary brokenness is much better than permanent paralysis: Don’t focus on perfection early on, focus on the generic case of the core user and add edge cases later.
  • Use the Frankenstein approach: Listed to feedback, but don’t build every feature users request. Understand why they’re asking for it, and see if the problem they want solved is best solved by making the feature they suggested.
  • Ship the product. Don’t be stealth. Just launch.
  • Once you ship and working on growth. Take one channel and focus on that, until it caps out or you realize it doesn’t work, and then move on. Revisit the ones that didn’t work later on.

Types of growth

  • Sticky: get your existing users to come back and use more/pay more.
  • Viral: people talk about your product/service and tell their friends which leads to more users.
  • Paid: paying to take out ads and the lke
Sticky
  • Need to deliver a good experience
  • Customer lifetime value3 and retention cohort analysis4 become important
Viral
  • Need to wow users
  • Need to have good mechanics for referrals/letting users talk about you
  • Good referral programs involve:
    • Well placed customer touch points: where do they know they can refer other people
    • Good program mechanics: benefits to refer people (money/credits to both users)
    • Optimised conversion flow for referred users
  • Examples: Search engine marketing, Facebook ads, local deals, groupon
  • Simple way to think about it: CLV > CAC5 (in aggregate)
  • Advanced: CLV for each segment > CAC for that segment
  • Need to consider payback time and sustainability
  • A good payback time is <3 months

8. Pivot if needed

In general, a good rule of thumb is that it is time to move on if one of the following is true:

  • bad growth (after you’ve tried a number of growth approaches and built out a few requested features)
  • bad economics (at some kind of scale)
  • bad retention (users aren’t sticking to your service)

  1. Remember the honesty curve when getting feedback, the further away people are from you, the more likely they are to give you honest feedback, though after a certain degree of separation, they don’t care about giving you feedback. Further, you get the best feedback if you make someone pay for something, because then the random people will really let you know if they don’t like it, and the so the level of honesty goes up the more random someone is. 
  2. Net promoter score is calculated by asking users how likely they are to recommend a product/service to a friend on a scale of 1 to 10, and taking the average of the responses. 
  3. Customer lifetime Value(CLV) is the net revenue from a customer over a certain amount of time (12 months, say). 
  4. Segment the users by the month they joined and compare the retention curve across months. If things are going to plan, the curve should be dipping less i.e. retention for new users is improving as time goes by. 
  5. CAC refers to Customer Acquisition Cost i.e. the cost of acquiring a customer.