Michael Siebel — Building Product

Published in

TLDR: Christopher Lam

3 min readSep 11, 2018

Notes on Michael Siebel about Building Product in YC Startup School. I am participating in this years’ 2018 Startup School.

JustinTV make a crap load of mistakes but what saved it was:

Team was extremely technical — not intimidated by any technical problem.
Egos tied to the startup; the feeling that if *this* failed then you have failed at life
Really low cash burn due to a bunch of 23 year olds sharing low living costs

What problem are you solving?

You have to figure what you are doing and what problem you are solving.
JustinTV did broadcasts — it easy if anyone wants that: did anyone watch it?
Can you state the problem clearly? It only takes one sentence.
Have you experienced the problem yourself? Make something that you would use yourself.
Can you define your problem narrowly? What can we address immediately?
Is the problem solvable? Eg. Poppy is ‘Uber’ for babysitters. Why do people need it? People have various reasons to hire babysitters, ie for babies or toddlers. Poppy discovered that the trust/skill factor has to be extremely high for them to hire a babysitter for your baby. Babysitters aren’t as replaceable as cab drivers and really skilled babysitters are already working as nannies. That talent pool for Poppy may not exist, therefore, the problem may not be solvable.

Not everyone.
How often do they have the problem. Eg. car shopping websites’ customers don’t have a recurring problem (as buyers purchase cars every 7 years). Its the sellers/dealers that have the problem selling used cars.
How intense is the problem? Do I have to go to the doctor and buy something? How often do you move towards that solution such as buying a car in lieu of taking a taxi?
Are they willing to pay? If you have extremely intense problem, then a $100 product would be a great deal. A priced product is better than a free product.
How easy are they to find? B2B customers can be easily reached in America. What channels? Email/ads, etc?

Build the MVP fast, and make sure you solve the customer’s problem.
Give the MVP to customers as a test; if the customers don’t find it useful then its not useful. Products are not paintings.

Go after the easiest customers; not the hardest ones.
The MVP is pretty much a shite prototype, therefore, the customers that use it the most are the most desperate ones; the early adopters. 6 month talks with a enterprise company usually means they aren’t desperate.
Do not get product feedback from friends.
Investors / friends don’t usually have the problem, their feedback will need you astray.
Fire shite customers.

No. Unless there a structure behind the discount and pushes the user to purchasing the product.

Google Analytics + something else eg. Mixpanel
Google Analytics is not great for tracking actions; you need to track actions and events, not just page-views.
Pick 5–10 important stats.
Spec out the stats you are going to track and improve.

Write down the spec, as four devs have four different interpretations of the same idea.
Short dev cycles
Have a KPI goal; a number to track such as revenue and/or usage such as number of customers
Open brainstorm. A real brainstorm is one that any idea that’s stated is written on the whiteboard. Thoughts or ideas were backed by metrics.

Brainstorms are sorted in 3 categories:

1. New features & iterations, bug fixes and A/B split tests

2. Easy/Medium/Hard

Decide on an idea to do.
Written spec: go through and write down in detail ‘what do mean by adding video filters, etc?’
JustinTV had infrequent meetings, approximately two week cadences. Just get shit done.

Pivot = changing the customer and/or the problem (rare). This means you may need to start a new company.
Iterate = changing the solution (common). Maybe MVP didn’t solve the problem? Solve the problem better.
Identify the problem
Don’t be fake Steve Jobs; iterate and talk to customers
Ask specific what they want and make it.