Five failed ideas before the product that stuck
Key takeaways from the Y Combinator Startup School Paris fireside with PostHog co-founder and co-CEO James HawkinsDealroom has a profile for this one. Try Dealroom → (August 2026):
The real lesson was pivoting. PostHog was the sixth idea Hawkins and co-founder Tim Glaser worked on — they cycled through five failed ideas in roughly seven to eight months, about a month each, committing fully to each before moving on.
Sleep-on-it rule. Their only discipline for quitting an idea: if something felt broken, force themselves to sleep on it one night, and if it still felt wrong the next day, start fresh — avoiding weeks of existential debate.
How the product was found. Frustrated by the friction of buying and implementing existing analytics tools, the pair — wanting self-serve pricing and no sales calls — built open-source product analytics; it emerged from the process of repeatedly spinning up new products.
Open-source trap. Hawkins warned that going open-source nearly trapped the company in on-prem enterprise debugging hell, and that a hackathon side project by one engineer became a multi-product strategy neither founder had planned.
Radical focus on AI. To build PostHog's "self-driving" AI product, Hawkins freed up essentially all his time — stopping interviews, investor work and one-on-ones — while Glaser took over running the rest of the ~125-engineer company.
Product over sales. Hawkins insists founder focus belongs on product, not sales; he fears flattening growth if both founders drift toward sales, citing advice from Stripe's Patrick Collison that you have to be good at both the top and bottom of the funnel.
Change is hard. Shifting mature products (each doing tens of millions in run-rate) toward an agent-first experience proved ~10x harder than expected, requiring intense hand-holding of staff and existing users through the transition.
Read more: Y Combinator
