Fireside Chat, Interview
Mike Hudack: How Facebook, Monzo and Deliveroo Build Great Products | E1201
- Product teams are expected to comprise six to eight people, primarily engineers with one data scientist, one designer, and an optional product manager, operating under outcome-based goals that could yield a 20% revenue increase.
- Future product strategies will prioritize shipping fewer, high-quality features after developing theories of the world that may be incorrect by 5% to 50%, with a 10-minute to one-year development timeline for this theory.
- Experimentation protocols require testing features on 20% to 50% of users to measure metric deltas, with a mandate to unship any feature lacking statistically significant impact or revealing paradoxical outcomes before full rollout.
- Advertising and distribution plans focus on verticalized marketing to increase virality and conversion, noting that broad markets require specific community targeting in overcrowded sectors to avoid vague product messaging.
- Organizational structure at Deliveroo prioritizes the core "Delivery" organization, including ML systems and dispatch, over consumer interfaces, utilizing regression models to estimate delivery times and messaging systems to reduce rider search time.
- Monzo's outlook involves expanding geographically and shifting user perception from a prepaid card to a full bank, utilizing specific metrics like salary to measure success while navigating banking regulations that restrict pricing strategy tests.
- New product launches at Monzo, including kid accounts and investment features, are expected to broaden the demographic reach over the next 12 months, while the organization avoids crypto products deemed unsuitable for the current user base.
- Founders and leaders are advised to trust their intuition for hiring, avoid underestimating competitors in "knife fight" markets, and prepare teams emotionally for failure to prevent abandoning products that are only 80% correct.
- Attitudinal changes in user bases are projected to take three to four months to manifest, with a warning that data interpretation is prone to self-deception and that being emotionally attached to a product can cloud judgment.
- Personal work-life integration is expected to involve working 20-hour days without obsessively checking metrics, which is identified as a dangerous behavior leading to burnout, as family responsibilities can positively shift work mindset.
- AI and LLM technologies are forecasted to progress through a hype cycle, including a trough and plateau, before becoming a significant utility, with current experimentation expected to be insightful.
- Bad behavior by founders is expected to continue as a normal landscape occurrence, though the "founder mode" term is not anticipated to cause a dramatic increase in such behavior.