Lecture 4 - Building Product, Talking to Users, and Growing (Adora Cheung)
Executive Summary
In this historical Stanford startup lecture, Adora Cheung describes learning the cleaning business firsthand, finding initial customers, gathering candid feedback and measuring whether customers return. Her Homejoy examples connect manual work with product design, channel testing and retention. Notes cover selected business passages of the public auto-captions for the 52-minute lecture, checked against the official Stanford CS183B archive. The date is the October 2, 2014 video upload. Homejoy figures and growth claims describe 2014, not the company today. An early therapy-platform example is not summarized; growth and payback guidance is Cheung's view, not a universal business rule.
Key Takeaways
Immerse yourself
Based on three years of pivots after YC in 2010, she says to take compressed time to focus on the idea. Her advice is directional, not a rule, since every business is different.
Learn the service firsthand
Cheung says she took a job at a cleaning company to learn the work and its daily operations. This helped her understand inefficient booking and scheduling rather than assume that reading about the industry would be enough.
Be obsessive about the space
List every competitor, read every search result, and read public competitors' S-1s, quarterly financials and earnings calls. Most give little, but you find golden nuggets, so become an expert.
Recruit initial customers manually
Cheung describes approaching people at local fairs and offering cold water before pitching a cleaning. Her broader point is to meet likely users where they already gather and learn from small numbers before expecting a scalable channel.
Ask for candid feedback
Provide a clear way to contact the company, then talk directly with customers. Cheung warns that surveys tend to capture extremes and that friends may be less candid than strangers who have paid for the product.
Temporary brokenness
Homejoy accepted only 3 to 5 percent of cleaning applicants and learned which interview questions predicted performance before moving them online. Automating too early blocks iteration, and temporary brokenness beats permanent paralysis.
Cohorts and pivoting
Track the share of each monthly cohort that returns and aim for the retention curve to flatten. Pivot when you cannot grow, nothing sticks or the economics fail, and set a growth plan up front.
The Conversation Breakdown
Immerse yourself
⏱ Untimed topicBased on three years of pivots after YC in 2010, she says to take compressed time to focus on the idea. Her advice is directional, not a rule, since every business is different.
Learn the service firsthand
⏱ Untimed topicCheung says she took a job at a cleaning company to learn the work and its daily operations. This helped her understand inefficient booking and scheduling rather than assume that reading about the industry would be enough.
Be obsessive about the space
⏱ Untimed topicList every competitor, read every search result, and read public competitors' S-1s, quarterly financials and earnings calls. Most give little, but you find golden nuggets, so become an expert.
Recruit initial customers manually
⏱ Untimed topicCheung describes approaching people at local fairs and offering cold water before pitching a cleaning. Her broader point is to meet likely users where they already gather and learn from small numbers before expecting a scalable channel.
Ask for candid feedback
⏱ Untimed topicProvide a clear way to contact the company, then talk directly with customers. Cheung warns that surveys tend to capture extremes and that friends may be less candid than strangers who have paid for the product.
Temporary brokenness
⏱ Untimed topicHomejoy accepted only 3 to 5 percent of cleaning applicants and learned which interview questions predicted performance before moving them online. Automating too early blocks iteration, and temporary brokenness beats permanent paralysis.
Cohorts and pivoting
⏱ Untimed topicTrack the share of each monthly cohort that returns and aim for the retention curve to flatten. Pivot when you cannot grow, nothing sticks or the economics fail, and set a growth plan up front.
Core Frameworks & Ideas
Storyboard the whole journey
Map discovery, signup, the service itself and the moment after use. Build a small feature set that solves the actual problem across that journey rather than merely shipping one isolated feature.
Sticky, viral and paid growth
Cheung separates retention, customer referrals and acquisition spending. Each needs a good experience, but referrals also need suitable touchpoints and paid acquisition needs sustainable economics.
Manual before automation
Run a process yourself until you understand what predicts a good result. Cheung describes learning useful screening questions before turning them into an online application.
Actionable Next Steps
- ✓Spend a week doing the customer's job yourself.
- ✓Chart monthly user cohorts and watch whether the curve flattens.
- ✓Write a growth plan with a date at which you will decide to pivot.
- ✓Test one acquisition channel with focused effort before splitting time across several.
Notable Quotes
“you should go and do that service yourself”
“you should become an expert in the industry”
“temporary brokenness is much better than permanent paralysis”
The One Thing to Remember
you should go and do that service yourself