Proxy Metrics in ML
The discussion highlights the use of lifetime value (LTV) as a valuable proxy metric for capturing business complexities, particularly in subscription-based models. While LTV provides insights into revenue and margins, it is acknowledged that it doesn't encompass every aspect of the business, emphasizing the need for a balance between computational models and real-world data. The conversation also touches on the relationship between LTV and survival models, showcasing the innovative approaches taken by the team to enhance understanding of expected gross profit.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Reinforcement Learning for Personalization at Spotify with Tony Jebara - 609
Related Questions
Explain LTV and how I can maximize it
We're running an AI consultancy for SMBs, and honestly, landing clients is tough. We think the model works, but our runway isn't infinite. Beyond just 'gut feel,' what specific metrics – like client LTV, churn rate, or pipeline conversion rate – should tell us, 'Pivot now,' versus just grinding harder on sales? Also, what's our North Star Metric?