Dave "CAC" Kellogg and Ray "Growth" break down one of the oldest productivity metrics in business and explain why, in the age of AI-native software, it has never mattered more. This episode covers the full arc from Frederick Taylor's factory floors to Cursor's $3.3M per employee, with the rigorous definitional discipline the Metrics Brothers are known for.

What We Cover:

The metric's 100-year history. Revenue per employee traces its roots to scientific management in the late 1800s, gained traction as a Wall Street efficiency screen in the 80s and 90s, and became a standard signal of business model quality in M&A diligence. The core math is simple: annual revenue divided by headcount. What is not simple is how you define the denominator.

FTE vs. employee: why the definition matters more than the formula. The E in FTE stands for full-time equivalent, not full-time employee, and that distinction drives real measurement decisions. How do you count a part-time contractor? What about 200 offshore developers on a third-party vendor's payroll? Ray and Dave walk through the practical choices, including why offshore headcount is almost never counted on a 1:1 basis and why that decision can dramatically change your benchmark comparison.

Public SaaS companies in 2025: the benchmark is $395K. Using the Benchmarkit SaaS 100 index (134 public SaaS companies), the median revenue per employee in 2025 is $395K, up from $327K in 2022, a 21% improvement in three years. ARR per FTE runs about 5-7% higher at $413K. The shift reflects the industry's move from growth-at-all-costs to efficient revenue growth.

Private SaaS companies: size matters. ARR per employee scales materially with company size. At the $5-20M ARR stage, the median is $144K. By $100M+ ARR, the median reaches $300K. The recurring-revenue tailwind from a large renewal base is a significant driver as companies scale.

AI-native companies have reset the benchmark entirely. Where the historical range for enterprise software was $200-400K per employee, AI-native companies operate at a fundamentally different level. Cursor reached $1.67M per employee at 60 people, and now runs at $3.3M per employee at 300 people. Midjourney is at $4.7M. Anthropic is in the $3-5M range on a run-rate basis. This is not a modest improvement over traditional SaaS. It is a 10x shift.

One important caution on the AI numbers. Many of the figures being cited by AI-native companies are monthly run-rate revenue annualized (last month times 12), not trailing 12-month GAAP revenue. When growth is compounding fast, that distinction can dramatically inflate the productivity figure. The Metrics Brothers flag this as a meaningful source of confusion in how the benchmark is being discussed today.

The AI tailwind may be temporary, at least in part. Current customer acquisition friction for AI software is unusually low, given experimentation budgets and departmental purchasing. As enterprise procurement tightens (74% of enterprise AI purchases now involve IT), GTM investment will likely increase, and revenue per employee for AI-native companies may stabilize or compress. Ray and Dave estimate that steady-state productivity is more likely to be in the 3-5x range over traditional SaaS, not 10x.

Revenue will replace ARR as the standard numerator. The rise of usage-based and hybrid pricing is rendering ARR less meaningful for a growing share of companies. Snowflake, Datadog, and MongoDB do not report ARR. As AI-native pricing models proliferate, Ray and Dave expect the industry to converge on revenue as the standard numerator across productivity benchmarks.

What about revenue per agent? Ray raises the forward-looking question: as AI agents take on SDR, sales, and other GTM functions, how do we measure agent productivity? Dave's take is that "revenue per agent" is likely a dead end, partly because agent instances are nearly impossible to count and partly because the right way to price and measure agents is to decompose their capabilities, not to anthropomorphize them as headcount equivalents.

The Bottom Line:

Revenue per employee is a deceptively simple metric with genuinely complex definitional choices underneath it. For B2B SaaS executives, the 2025 benchmarks are $395K (public) and $144-300K (private, depending on scale). For AI-native companies, the numbers are in a different category entirely, though some of that gap reflects accounting choices as much as true productivity gains. The metric is worth tracking closely, both as a board-level efficiency signal and as a leading indicator of business model quality.

See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.