In a world where new AI companies are emerging at breakneck speed—and valuations are being set on vision more than revenue—it’s worth asking: are we still solving puzzles, or are we navigating mysteries?
Here’s a great excerpt from the Farnham Street blog that really puts this into perspective:
Conventional SaaS Metrics Are Built for Puzzles
In early-stage tech, we’ve long relied on a set of trusted metrics: ARR, ACV, YoY growth rate, Rule of 40, CAC to LTV ratios. They help founders to track their progress and help investors benchmark traction, efficiency, and scalability. But these metrics assume a relatively linear path to growth, or at least a path that can be modeled.
What we’re seeing today is that this assumption breaks down for many of the newest AI-native companies.
When growth becomes asymmetric—when the right product, at the right moment, can leapfrog years of expected traction overnight—those conventional signals become lagging indicators at best, and red herrings at worst. Look no further than OpenAI. One day, an API company. The next, the most important software layer in the world.
So What Should We Measure?
In the age of building “AI companies”, the most important “metric” may no longer be revenue. It may be speed of execution. How quickly can you build, ship, iterate on, and react to feedback? How fast can you test a hypothesis, learn, and move to the next one?
Speed now compounds. It’s not only about doing more with less—it’s about learning faster than the market. In that sense, metrics that used to feel secondary are starting to rise to the top.
Metrics That Matter Now
- Cycle Time: How long does it take from idea to prototype? Prototype to launch?
- Experiments Run Per Quarter: A proxy for how aggressively a team is testing assumptions.
- Customer Feedback Velocity: Not just what users are saying—but how fast you’re getting that input and acting on it.
- Burn Efficiency: How many learnings or validated experiments per dollar spent?
These aren’t metrics for a pitch deck—they’re operational tools. They tell you whether you’re moving fast enough to win in an environment where the first to learn often beats the first to market.
From Puzzles to Mysteries
There’s comfort in puzzle-solving. But if you’re building (or backing) an AI company, you’re not working toward a single, knowable outcome. You’re navigating a mystery. That means building for speed, learning in public, and getting comfortable with ambiguity.
So what about the “old” metrics? While previous SaaS benchmarks are no longer the scoreboard, there is still real value in product, user and (our favorite) Go-To-Market metrics. One thing is for certain: In this new era, progress looks a lot more like momentum than milestones.
