The Lean Startup
Table of Contents
Notes from the book The Lean Startup [amazon.com]
Following
- Vision: Creating a thriving and world changing business. This is the true north, the destination of a startup.
- Strategy: This follows from vision. It is how we achieve the vision. It includes business model, product road map, information about partners and competitors, and idea about customers.
- Product: The end result of strategy.
Product changes frequently through optimization. Strategy change, i.e. pivot, is less frequent. And the vision rarely changes.
1. Validated Learning
Startup is a institution designed to create a new product or service under conditions of extreme uncertainty. So, the objective is to learn how to build a sustainable business.
Do frequent scientific experiements and learn what the customer needs and discover valuable truth about startup's business prospects. Validated learning is backed up by empirical data collected from real customers.
However learning is an excuse for a failure of execution. For an employer, or investor, "I learned" is not comforting. If a failure is explained after the fact, then it is not validated learning, it is story to hide failure.
Value vs waste:
We do an experiment and it fails. Was all the effort waste? The effort that is not absolutely necessary for learning what customers want is waste and can be eliminated.
The experiments we do are products too.
2. Hypothesis
There are thwo important assumptions that need to be validated:
- Value Hypothesis: Does what we are building delivers value to customers once they are using it?
- Growth hypothesis: How do the customers know about our product?
3. Build Measure Learn
The objective is to minimize the time through once cycle of Build-Measure-Learn feedback loop.
4. MVP
Too much planning leads to analysis paralysis. The fault in assumptions made during planning can't be detected at the whiteboard, it requires interaction between product and the customers.
On the other hand, doing no planning and "just-do-it" philosophy is also bad. So, where is the balance? Answer: Minimum Viable Product.
The goal of MVP is to begin the process of learning. Its goal is to test fundamental business hypotheses. Follow the simple rule when building MPV: remove any feature, process, or effort that does not contribute directly to the learning you seek.
Example:
You give a one-month free trial to your customers. They will use it and then decide to stick or not.
Assumption: What proportion of users who see the website will register to the free trial?
To test this assumption, do we need all the features in the product? Is just the website, and a wait list enough?
Dropbox: People didn't understand what file issues dropbox was solving. They wouldn't understand until they use the product.
Solution: Create a video that demonstrate what the product works like. This video was the MVP which validated that the customers would want to use his product.
- An example of grocerry list suggestion service [at page 104]. Instead of working on something that might work in the future, they started small that was working and scaled it up.
5. Early Adopters
Before we sell products to general customer, we sell them to early adopters. They don't need perfect solution to capture their interest.
6. Quality
If we do not know who the customer is, we do not know what quality is.
7. Measure - Innovation Accounting
- Get a track of the current status of the company.
Build a MVP: Or multiple MPVs testing different assumptions (e.g. smoke test with marketing materials even without prototype)
Test the riskiest assumptions first. [#123]
- Get read data on current status: conversion rates, sign-up, customer lifetime values, etc.
- This is the baseline.
Tune the product baseline to reach the ideal product.
E.g. If customer retention is lower than required, and the assumption is that product design will improve that metric. Then, change the design and track the retention. This gives us validated learning.
Pivot or Preserve: If progress is good, continue. If not, the current strategy is flawed and needs change i.e. pivot.
We need horrible metrics of the initial baseline MVP to rise up to the ideal one established in the business plan. Otherwise, we need to pivot the strategy.
Types of pivot: [#177]
- Zoom-in pivot: Focus on one feature
- Zoom-out pivot
- Customer Segement pivot
- Customer need pivot
- Platform pivot
- Business Architecture pivot: e.g. B2B to B2C
- Value Capture
- Engine of Growth pivot: viral model, sticky model, paid growth model
- Channel Pivot
- Technology pivot
7.1. Metrics
Use cohort analysis (analysis based on users that joined the product at paticular period). E.g. look at percentage of cohort that eventually bought the product instead of total cumulative sales. This helps separate if the positive change was due to tweak we just did, or because of previous decisions. (See Figure at #127 vs Figure at #135))
Metrics must be:
Actionable: When a metric goes up and down, we need to be able to know what caused it. So that we can take relevant action.
E.g. split-test are actionable, cohort analysis from new feature is actionable.
- Accessible: The reports should be understandable by the employees and managers. Use easy units. Keep it in the same system as the product. Have a system/website where employee can see current experiments and past experiments.
- Auditable: Repots should not be too complex, and should be derivable directly from master data. There must be a way to check validity of the report with respect to the reality.
8. Five Whys
At the root of every seemingly technical problem is a human problem. Ask why five times to get to the root of the issue.
Be careful that when answering five whys we do not lead to five blames. To prevent that, everyone who is involved in detecting, fixing the issue is in the room. So that people don't blame whoever is absent.
Most mistakes are caused by flawed systems, not flawed people.