Lessons from George Bonaci

9/27/2025, 12:00:00 AM

George Bonaci is the VP of Growth & Demand at Ramp.

Philosophy: Growth as Science

  • Growth = Science: Playbooks rarely transfer 1:1. Start with a blank slate — form a hypothesis → run experiments. Most marketers apply what they already know; real growth is discovering what works for your company.
  • Mindset: Think like a scientist — experimental design, measurement, and rigor.
  • Velocity > Perfection: Bias toward running more experiments faster vs. perfecting one.
    • The tradeoff: A scrappy experiment that 3x's a conversion rate to hit the quarter is fine. Later, tighten rigor to understand causality. Velocity and rigor are dials based on time horizon.

Portfolio Approach

  • Treat growth like a VC portfolio with intentional allocation.
    • Long‑term (big swings): High‑risk, high‑reward — step‑changes. Most will fail.
    • Short‑term (incremental): Higher‑confidence 2–5% gains that help you hit this quarter.
  • When a bet works → push to saturation: Don't just 2–3x spend. Scale the channel quickly to its asymptote and find where the response curve decays. Most teams scale winners too slowly.

Diminishing-returns response curve for media investment → incremental sales.

  • Finding Alpha: Iterate + Experiment

    1. Do what's new: Be early where the channel is under‑priced or novel (e.g., first B2B brand on TikTok).
    2. Do what's contrarian: Try what "shouldn't" work (e.g., Direct Mail becoming a major channel).

Mental Models

  • Pre‑mortems & Post‑mortems
    • The value of a post‑mortem depends on the quality of the pre‑mortem.
    • Pre‑mortem: Part of initial design — enumerate not just 3–4 big risks but the many small failure modes; assign probabilities.
    • The real learning = unpredicted failures: When something fails for a reason you didn't anticipate, it reveals a flawed assumption in your model of the world.
      • Predicted failures mostly confirm existing beliefs; limited learning.
      • Investigate the surprise, update the model, and generalize the lesson across future experiments.
    • Key post‑mortem question: "Did we learn something we can generalize to other parts of the business?"