Skip to content
a16za16z

The $700 Billion AI Productivity Problem No One's Talking About

Russ Fradin sold his first company for $300M. He’s back in the arena with Larridin, helping companies measure just how successful their AI actually is. In this episode, Russ sits down with a16z General Partner Alex Rampell to reveal why the measurement infrastructure that unlocked internet advertising's trillion-dollar boom is exactly what's missing from AI, why your most productive employees are hiding their AI usage from management, and the uncomfortable truth that companies desperately buying AI tools have no idea whether anyone's actually using them. The same playbook that built comScore into a billion-dollar measurement empire now determines which AI companies survive the coming shakeout. Timecodes: 0:00 — Introduction 1:07 — Early Career, Ad Tech, and Web 1.0 2:09 — Attribution Problems in Ad Tech & AI 3:30 — Building Measurement Infrastructure 5:49 — Software Eating Labor: Productivity Shifts 7:51 — The Challenge of Measuring AI ROI 13:54 — The Productivity Baseline Problem 17:46 — Defining and Measuring Productivity 20:27 — Goodhart’s Law & the Pitfalls of Metrics 21:41 — The Harvey Example: Usage vs. Value 24:18 — Surveys vs. Behavioral Data 27:38 — Interdepartmental Responsiveness & Real-World Metrics 30:00 — Enterprise AI Adoption: What the Data Shows 32:59 — Employee Anxiety & Training Gaps 37:31 — The Nexus Product & Safe AI Usage 41:08 — The Future of Work: Job Loss or Job Creation? 43:40 — The Competitive Advantage of AI 52:45 — The Product Marketing Problem in AI 54:00 — The Importance of Specific Use Cases SOCIALS Follow Russ Fradin on X: https://x.com/rfradin Follow Alex Rampell on X: https://x.com/arampell Stay Updated: If you enjoyed this episode, be sure to like, subscribe, and share with your friends! Find a16z on X: https://x.com/a16z Find a16z on LinkedIn:https://www.linkedin.com/company/a16z Listen to the a16z Podcast on Spotify: https://open.spotify.com/show/5bC65RDvs3oxnLyqqvkUYX Listen to the a16z Podcast on Apple Podcasts: https://podcasts.apple.com/us/podcast/a16z-podcast/id842818711 Follow our host: https://x.com/eriktorenberg Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details, please see http://a16z.com/disclosures.

Russ FradinguestAlex Rampellhost
Nov 30, 202557mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

Enterprises are buying AI fast, but can’t prove productivity gains

  1. Enterprise AI budgets are growing rapidly, but leaders widely suspect a majority of projects are wasting money because they cannot measure outcomes beyond licenses purchased.
  2. Measuring AI value resembles early ad tech attribution: without shared infrastructure for measurement and governance, spend increases faster than confidence in results.
  3. “Productivity baseline” is the core problem—companies struggle to define outputs, avoid Goodhart’s Law metric-gaming, and separate usage from actual business value.
  4. Adoption is constrained by employee anxiety and training gaps, so enabling “safe” AI use and practical workflow integration is as important as model capability.
  5. AI’s labor impact is framed as augmentation and competitive advantage rather than mass unemployment, with winners using AI to grow faster rather than simply cut headcount.

IDEAS WORTH REMEMBERING

5 ideas

AI has an ROI visibility crisis, not just a capabilities problem.

Companies can easily buy tools, but many executives report they only track “how much we bought,” not whether AI improved outcomes—creating a repeat of early internet advertising’s attribution fog.

The productivity baseline is missing, so “more productive” is often meaningless.

If AI lets a lawyer finish in 4 hours what used to take 8, the employee benefits immediately, but the company only benefits if outputs increase, responsiveness improves, or staffing/work allocation changes—none of which is captured automatically today.

Usage data must be paired with outcome measures; surveys alone are unreliable.

Self-reported productivity is biased (definition ambiguity, desire to please, fear of looking bad), and it ignores the reality that some licensed users never log in; behavioral usage telemetry is the necessary first layer.

Goodhart’s Law is a real risk in AI performance metrics.

If companies promote a metric (e.g., messages sent, lines of code, dollars spent on coding assistants) into a target, employees can game it—so metrics should guide internal diagnosis more than direct employee scoring.

A practical measurement approach triangulates three signals: usage, output/impact, and time/“tonnage.”

The proposed path is to correlate heavy vs. light tool usage with department-specific productivity indicators and with aggregate time/throughput—evaluated at group levels to avoid privacy and noise at the individual level.

WORDS WORTH SAVING

5 quotes

85% of the companies we talked to said they really believe they only have the next 18 months to either become a leader or fall behind.

Russ Fradin

Every board meeting I go in for my other four metrics, I have some report of how are we doing against those reports, and on AI, all I have is the amount of stuff we bought.

Russ Fradin

When a measure becomes a target, it is no longer accurate as a measure.

Alex Rampell

I'm sure that was very cool for him, but that's absurd. That's an absurd way to hope people adopt world-changing technology.

Russ Fradin

Cursor has taken mediocre engineers and made them good, but it's taking amazing engineers and made them gods.

Russ Fradin

Parallels between ad tech attribution and AI ROIAI measurement infrastructure and third-party analyticsUsage vs. value (Harvey example)Productivity baselines and definitionsGoodhart’s Law and metric distortionEmployee enablement: safety, training, governanceEnterprise adoption data: spend, waste concerns, 18-month urgencyInterdepartmental responsiveness as a proxy metricProduct marketing challenge: specific use cases vs “AI does everything”

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

Add to Chrome