Return on Token: Who Owns the Hours AI Gives Back [66]

The episode opens with a CFO's question that sparked widespread, uncoordinated AI tool purchasing: a company bought 11 AI subscriptions across four regions without demonstrating measurable time savings. The hosts introduce "Return on Token" as a framework for addressing this problem.

The core issue is that when AI tools save time, leaders often don't capture the value because the freed hours get consumed by unmeasured activities — internal calls, Slack threads, and other low-value work. The speakers present a case study where demo prep time dropped from two days to half a day, but the recaptured time wasn't intentionally directed anywhere.

The solution involves three key practices: First, naming one specific output per role that the time should be spent on (e.g., more customer conversations for SEs, CRM follow-ups for AEs) before purchasing the tool. Second, measuring through monthly manager reviews asking what teams did with the time and what pipeline impact resulted, rather than auditing individual hours. Third, pricing the entire solution holistically — including learning curves, process rework, and tool overlap costs — not just license fees.

The speakers estimate that at scale, a central "go-to-market engineer" role (approximately 100k salary) can justify itself by giving four AEs, one SDR, and one SE each 20-30% of their administrative time back. They also discuss the tension between standardization and autonomy at different company scales, suggesting a named 20% "tinkerer" role for smaller teams.

A critical guardrail emerges: speed without quality checks leads to automating bad processes. The episode concludes with a practical framework: pick one admin task per role, target 20-30% time savings, document expected outcomes before purchasing, and cancel within one quarter if the freed time isn't spent on intended activities.