YC Root AccessLecture 4 - Building Product, Talking to Users, and Growing (Adora Cheung)
CHAPTERS
- 0:04 – 1:05
From zero users to many: avoiding the “build in stealth, launch big” trap
Adora frames the talk as hard-earned guidance from her YC and Homejoy experience, emphasizing common early-stage mistakes. She contrasts the novice pattern—secretive building, big launch, then churn—with a feedback-driven approach that prevents wasted effort.
- •Advice is directional; every startup is different
- •The novice cycle: build privately → splashy launch → low retention → desperation tactics
- •Why lack of early feedback leads to short-lived user spikes
- •Learning from pivots and missteps at Homejoy/earlier projects
- 1:05 – 3:37
Make compressed, focused time for startup progress
She argues that startups require sustained, uninterrupted blocks of time to maintain context and make real progress. Fragmented effort (a couple hours a day) often leads to excessive context switching and slow iteration.
- •Prioritize contiguous work blocks over scattered sessions
- •Context switching kills productivity, especially for engineering/building
- •Treat startup work like deep technical work that needs immersion
- 3:37 – 5:07
Start with a one-sentence problem—and ensure it’s a problem you truly have
Before building, founders should articulate the problem clearly and validate it with real people. Adora shares an example where she and her cofounder pursued a “make people happier” platform for coaches/therapists—only to realize they wouldn’t use it themselves.
- •Define the problem in one sentence
- •Check founder-problem fit: are you passionate and personally invested?
- •Validate that others share the problem via conversations
- •Pathjoy lesson: spending a year on something you wouldn’t use is costly
- 5:07 – 8:40
Immerse in the industry by doing the work (becoming a ‘cog’ first)
Adora encourages founders to deeply learn the industry they’re entering—even by temporarily working inside it. In Homejoy’s case, doing cleaning themselves and then joining a cleaning company revealed operational inefficiencies and constraints incumbents accepted.
- •Spend 1–2 months understanding the industry’s real workflows
- •Hands-on experience reveals inefficiencies you can exploit
- •Homejoy example: cleaning firsthand → training/job at a cleaning company
- •Service businesses benefit from founders doing the service themselves
- 8:40 – 9:41
Become obsessively knowledgeable about the competitive landscape
She describes an “obsessive” research process: list competitors, read extensively, and mine public filings for insights. Trust and strong decision-making come from unmistakable expertise in the space.
- •Catalog all competitors and adjacent companies
- •Read deeply (articles, search results, filings, earnings calls) to find “golden nuggets”
- •Most information is noise—but occasional insights are pivotal
- •Industry expertise builds credibility with users and partners
- 9:41 – 11:43
Choose an initial customer segment and storyboard the full user journey
Rather than targeting everyone immediately, she advises picking a focused customer segment to optimize for early. Before writing code, founders should storyboard the entire experience—discovery, onboarding, usage, and post-use evaluation—to clarify what to build.
- •Focus early on a narrow segment to avoid diluted product decisions
- •Storyboard end-to-end: acquisition → site messaging → purchase/usage → reviews
- •Think beyond the website: how users find you and what happens after use
- •Turn the ideal flow into a plan, then code from that
- 11:43 – 13:45
Build an MVP that’s actually viable—and nail the one-line positioning
She highlights that MVP means minimal but still “viable,” not a thin feature with a broken experience. Clear positioning matters early: Homejoy improved conversion when it simplified messaging to a direct functional promise (price + outcome).
- •MVP = smallest set that solves the problem while still feeling complete
- •Use user conversations and the storyboard to define the minimum feature set
- •Create a one-sentence value proposition users immediately understand
- •Homejoy positioning shift: complex platform pitch → “Get your place cleaned for $20/hr”
- 13:45 – 17:46
Getting the first users: start with your network, then do unscalable hustles
Adora explains how early users often come from founders’ immediate circle, but that may not be enough. She shares Homejoy’s scrappy tactic of recruiting at a street fair by handing out cold water to start conversations—illustrating how low conversion is normal at zero-to-one.
- •First users: founders, family, friends, coworkers—use it yourselves
- •Leverage online communities (#showhn) and local mailing lists
- •Expect very low conversion; you’re proving demand, not scaling yet
- •Homejoy story: street fair outreach + cold water to earn attention and bookings
- 17:46 – 23:41
Customer feedback loops: go talk to users, measure retention, beware biased honesty
She recommends both inbound support channels and proactive user conversations, preferring informal, trust-building discussions over interrogations. Quantitatively, retention is the core metric, with reviews/NPS as faster leading indicators; she also explains why feedback honesty varies by relationship and payment.
- •Provide easy contact paths (email/phone) but prioritize outbound user talks
- •Make feedback conversations comfortable (coffee/drinks) to increase honesty
- •Track retention; use reviews/NPS as faster signals before cohorts mature
- •Honesty curve: friends/family bias; paid users and strangers can be more candid
- 23:41 – 27:45
Iterate fast: manual before automation, ignore edge cases, don’t build a Frankenstein
Adora urges founders to optimize for the next stage (10→100), not the million-user future. She advocates manual processes to learn what to automate, warns against perfectionism, and cautions that blindly implementing every feature request creates bloated products that hide the real problem.
- •Build for the next growth stage, not the end-state scale
- •Manual first reveals which steps matter before you automate them
- •Temporary brokenness beats paralysis; perfection is irrelevant early
- •Avoid “Frankenstein” features—ask why users want something, not just what
- 27:45 – 29:38
Launch early: stealth and fear of copycats slow learning
She argues that waiting to launch for fear of imitation is usually counterproductive. Unless the business requires massive upfront capital, shipping sooner accelerates learning and helps you out-execute inevitable fast followers.
- •Assume good ideas will be copied regardless of launch timing
- •The cost of delaying is lost user feedback and slower iteration
- •Launch early unless huge capital requirements make it impractical
- 29:38 – 32:39
Choosing and running growth channels: focus one at a time, keep iterating, revisit later
Adora outlines an early-stage operating cadence for growth: pick one channel, execute intensely for a week, then decide whether to continue or move on. She stresses continuous iteration because external platforms change, and notes that failed channels can become viable later as your economics improve.
- •Run one channel at a time to get clear signal quickly
- •Write playbooks and iterate; channels change constantly (ads/platforms)
- •Revisit previously failed channels when your unit economics improve
- •Homejoy example: early Google Ads were too expensive versus incumbents; later might work
- 32:39 – 46:26
Three sustainable growth modes: sticky, viral, and paid (and how to measure them)
She introduces sticky growth (retention/CLV), viral growth (referrals), and paid growth (ads and acquisition spending), tying all to sustainability. The chapter includes a cohort retention walkthrough, referral program components, and the CLV vs CAC framework plus payback-time risk management.
- •Sticky growth: deliver a great experience; measure via retention and CLV
- •Cohort analysis: watch retention curves flatten and improve over time
- •Viral growth: great product + strong referral mechanics (touchpoints, incentives, conversion flow)
- •Paid growth: only works when CLV > CAC; analyze by segment and manage payback time (e.g., 3–12 months)
- 46:26 – 49:57
When to pivot: diagnose growth, retention, and unit economics—then act fast
Adora explains pivoting through Homejoy’s many prior ideas and offers criteria for moving on: inability to grow, poor retention despite iteration, or broken economics. She recommends setting an optimistic-but-realistic weekly growth plan and treating multiple weeks of no/negative growth as a signal to re-evaluate and adjust.
- •Pivot triggers: no growth, low retention, or unsound economics
- •Set a week-by-week growth plan (early growth should be achievable manually)
- •If fully executing and still flat/negative for ~3–4 weeks, something is fundamentally wrong
- •Expect lulls; keep pushing through to restore an upward trend
- 49:57 – 52:21
Q&A: Getting users to switch from an existing solution via moments of clear differentiation
In response to a question about switching costs, Adora explains that users rarely change for many small improvements. Instead, win them during moments when your product is clearly better (e.g., urgent next-day need), then let accumulated conveniences convert them over time.
- •Switching costs are real; trust in incumbents can be hard to break
- •Find high-need moments where your advantage is obvious (e.g., next-day availability)
- •After first use, small benefits add up (online payment, easy rescheduling)
- •Lead with 1–2 clear differentiators, not a list of 50 minor improvements
