Stop Flaunting ARR — This Metric Is Deceiving Everyone
Deep thoughts on AI and aspirations —— ByteDance Deep Think Circle
In recent years, founders’ social profiles have practically become ARR display walls. Zero to a hundred million in a few months. Breaking a million today, ten million next quarter. Each number steeper than the last, each caption more confident than the previous. Even the investment world has adopted this screening standard: below $100 million before Series A, not even worth a look.
But after seeing countless such displays, I’ve become increasingly skeptical. Not skeptical of the growth itself, but skeptical that we’re measuring growth with a distorted yardstick.
Flaunting Numbers Is Flaunting Status Anxiety
Let’s start with a psychological layer few dare to call out: flaunting ARR is fundamentally a competition for identity currency.
When “moving fast” becomes the dominant startup narrative, numbers take on two functions—externally they’re signals, internally they’re placebos. What founders are really flaunting isn’t revenue, it’s the declaration “I’m still in the game.” What audiences see isn’t information, it’s another round of comparison. The side effect of this atmosphere is very specific: it pushes people who should be slowing down to build products to chase a presentable number out of anxiety. So growth becomes about building reports, and product becomes about crafting narratives.
Numbers themselves aren’t wrong. What’s wrong is that they’ve shifted from being “operational results” to “social currency,” and currency gets diluted by inflation.
Three Traps in the Numbers
Using ARR as a yardstick means stepping into at least three pits.
The first is methodology. Many proclaimed ARR figures annualize very loosely: extrapolating a year from one explosive month, counting one-time revenue as recurring revenue, treating customer prepayments as subscriptions. The strict definition should be subscription annual value that’s contractually bound, predictable, and sustainable—but in public displays, these constraints are often quietly relaxed. The most valuable thing about a number isn’t how big it is, but how it was calculated.
The second is survivorship bias. What you see is the zero-to-hundred-million curve. What you don’t see is that most companies on the same path have already disappeared. Public numbers are inherently “stories of survivors.” Using them as benchmarks is like treating lottery jackpots as average salary levels.
The third is ecosystem distortion. When inexperienced people treat others’ numbers as their own passing grade, they make decisions that betray long-term interests—cutting quality to inflate numbers, doing one-off deals to pad client counts. These moves look good on reports but are mortgaging real business.
What Are the Real Signals
Strip away the numerical packaging, and what’s actually worth watching is surprisingly simple: retention and repeat purchase.
Throw three questions at any growing product, and the answers are far more honest than ARR—Who’s paying? Why do they keep paying? If all growth activities stopped tomorrow, would this paying cohort stay? The first two questions test payment quality, the third tests demand authenticity. A business with beautiful numbers might not be able to answer any of these questions. Conversely, a business that answers all three cleanly—the numbers are just a matter of time.
Behind this is an attribution problem: ARR is an outcome variable, not a process variable. Optimizing for outcomes as if they were processes inevitably leads to fabricated growth. Optimizing for processes (retention, repeat purchase, customer value) as processes—outcomes will follow automatically. The difference is whether you’re watching the dashboard or the gas pedal.
The Other Side of Fast Growth Has Never Been Free
There’s another often-overlooked symmetry: the price of hypergrowth.
Once growth takes off, a company must build capabilities that normally belong to large companies in the shortest time—hire people who can keep pace, establish culture without budget reports, handle public scrutiny that never misses. A single pricing adjustment can trigger a media tsunami. Three years ago, product decisions at this level had no audience at all. With a team not yet at a hundred, you’re dealing with problems companies of several hundred encounter.
This isn’t saying fast growth is bad. It’s saying fast growth is a high-leverage survival mode: returns are amplified, and so is fragility. Companies that collapse under huge numbers or get devoured by a single misstep aren’t fewer than slow companies.
My Additional Take: AI Era Valuation Metrics Are Shifting
If you only focus on ARR itself, you’ll easily miss a structural change underway: AI company valuation logic is shifting from “revenue scale” to “per-customer value.”
The reason is that AI product cost structures have changed. High inference costs, marginal costs no longer approach zero—the old logic of “big revenue, thin margins, rely on renewals” is broken. What can truly support valuations is how much real value a customer can extract from the product, and whether that value is repeatable. In other words, beyond revenue numbers, the market is now asking: “Is this actually a business with unit economics that work and customers can’t leave?”
For founders, this is actually liberating: you don’t need to chase an eye-popping ARR to fundraise. What you need is to prove per-customer value works, and show retention to those who want to see it.
Don’t Mythologize “Healthy Growth” Either
One last contrarian note: the healthy growth doctrine has its own survivorship bias. Companies that simultaneously achieve high retention, high repeat purchase, and high renewal rates were always lucky few. Treating this as the only correct template simply creates new anxiety.
The most solid stance might be: be honest about your own numbers, be skeptical of others’ numbers, stay sensitive to real customer needs. Numbers are results, quality is cause—this has been said many times, but the reverse is equally important: quality can only be validated by time, and time gives no discount to anyone good at telling stories.
Key points: Flaunting ARR reflects status anxiety; three traps are metrics manipulation, survivorship bias, and ecosystem distortion; retention and repeat purchase are the real signals; fast growth has a price; AI era valuations look at per-customer value; don’t mythologize healthy growth doctrine either.