From failed voice app to $1.4B analytics company: lessons from Amplitude's founder

Executive overview

Most B2B founders build too long before charging money — Spencer Skates did this with Amplitude and lost over a year. The pivot from Sonalight to Amplitude came from a single insight: companies needed deeper product analytics, not another surface-level tool.

Spend half your time talking to customers. Charge earlier than feels comfortable. Find the 99th-percentile idea, not just a good one.

The biggest lever is idea quality: a 99th-percentile idea matched to the right founding team compounds everything else.

Picking the right idea

• Sonalight (voice recognition) was a 95th-percentile idea — technically hard, but founders weren't passionate users
• Sam Altman challenged whether they were truly passionate about voice recognition; they weren't
• Amplitude was a 99th-percentile idea: clear market shift (marketing web → product web), large existing pain, perfect fit for their algorithms engineering background
• Legacy tools (Google Analytics, Adobe) were surface-level — no one went deep on the problem
• Switching ideas felt like failure but was the right call

Customer development before building

• Biggest Sonalight mistake: engineers defaulted to building instead of talking to users
• Before writing Amplitude code, reached out to 30 companies to validate the pain
• Committed to spending 50% of founder time on customer conversations — non-negotiable
• Seven companies expressed interest; only one converted, but that was enough to proceed
• Skipping customer conversations mid-build caused them to attract free users with no intent to pay

Charging earlier than feels ready

• Spent a year building before asking for money; should have launched in six months
• First paying customer (Super Lucky Casino) only happened because a customer asked "how much does it cost?"
• Defaulted to $1,000/month on the spot — customer agreed immediately
• Lesson: a weak "no" from early users means you're talking to the wrong companies, not that the product is wrong
• Ask for commitment early; it filters for real pain and accelerates finding the right customer profile

Growing from zero to $100M ARR

• $0 → $1M: beg anyone to use it, commit to building custom features, pull off miracles manually
• $1M → $10M: a kernel is working; momentum builds but still needs active pushing
• $10M+: heroic individual efforts stop scaling — hire outside executives, build repeatable processes, create forecast discipline
• Reached $100M ARR in ~6 years (2014–2020); 2,700 customers including Fortune 100
• Direct listing on public markets in 2021

The personal cost of building

• Deliberately sculpted every aspect of life around building the company
• Lost touch with many friends; accepted this as a conscious trade-off
• References the Harvard Business School framework: 10,000 hours of practice, expert coaching, enthusiastic family support
• Anyone can start a company at any age — but only if they can honestly commit all three elements