I Was Terrified to Pivot My Product. Turns Out I Was Confusing Risk with Volatility.
Last March, I had a problem.
My project had been running for 4 months with metrics that went up and down like a roller coaster. One day 50 active users, the next day 15. Inconsistent engagement. Unpredictable conversions.
All the "experts" said the same thing: "You need to stabilize. Pivot carefully. Change is risky."
So I froze. Because changing seemed "risky."
Until I read Howard Marks' memos on risk.
The Distinction Wall Street Ignores (And Builders Do Too)
Howard Marks, the investor behind Oaktree Capital, has been writing about this for decades:
Volatility is not risk.
Risk is permanent loss. Volatility is temporary fluctuation.
Wall Street uses volatility as a proxy for risk because it's easier to measure. You can calculate standard deviation. You can make pretty charts.
But they're not the same thing.
And this confusion kills products.
What's Actually Risky in Product Development
Let's apply this to building products:
What people call "risky" (but is just volatility):
→ Metrics that fluctuate month to month → Changing your tech stack → Pivoting based on feedback → Experimenting with pricing → Testing new acquisition channels
What's ACTUALLY risky (permanent loss):
→ Building for 6 months without validating with real users → Ignoring feedback because "you know better" → Marrying yourself to a technical solution that doesn't scale → Spending all your runway on marketing before having product-market fit → Not diversifying your acquisition channels
See the difference?
The first are fluctuations. Reversible. Part of the process.
The second are decisions that can kill your project permanently.
Marks' Framework for Product Decisions
Marks has a simple framework for evaluating risk:
1. Can I lose everything?
If I pivot my product now, do I lose all progress? Generally no. You keep: your codebase, your learnings, your early adopters.
If I keep building without validating, can I lose everything? Yes. 6 months and you have nothing the market wants.
2. What's the permanent downside?
Switching from Next.js to another framework: time and effort, but reversible. Not using analytics from day 1: you'll never know what worked or why.
One is volatility. The other is permanent information loss.
3. Is the "risk" really just discomfort?
This is brutal to admit.
Most of what I call "risky" is simply uncomfortable:
- Talking to users is tedious → I call it "risky"
- Changing the plan means admitting I was wrong → "very risky"
- Pivoting means starting over → "too risky"
Marks would say: you're confusing emotional volatility with real risk.
My Mistake with the March Project
Back to March.
My metrics were volatile. That scared me.
But the REAL RISK was different:
I was building features nobody asked for because I was afraid to ask them what they really needed.
The volatility was information. The metrics were telling me: "There's something here, but it's not quite this."
The risk was ignoring that information.
How I Apply This Today
Now, when I evaluate a product decision, I ask myself:
Permanent Loss Test:
``` Decision: [What I'm considering]
If it goes wrong:
- What do I lose that I CANNOT recover?
- What do I learn even if it fails?
- Can I reverse it in 1 week? 1 month?
If it goes right:
- What's the upside?
- What information do I gain?
```
Real example from last week:
Decision: Completely change the onboarding for one of my products.
Initial fear: "Current users will get confused. Metrics will drop. It's risky."
Actual analysis:
- Permanent loss if I change: None. I can revert in 2 hours.
- Permanent loss if I DON'T change: 3 months of data saying nobody completes current onboarding.
I changed it. Metrics DID drop the first week (volatility). Then stabilized 40% higher.
The volatility was scary. The real risk was NOT changing.
Volatility as Information, Not Enemy
Marks has another key insight:
Volatility is expensive for funds because investors panic-sell. But for the individual investor with long horizons, volatility is your friend.
Translated to product:
If you're building to sell in 3 months, volatility kills you. If you're building for the next 3 years, volatility gives you valuable information.
Stable metrics from day 1 can mean:
- You've found product-market fit (great)
- Or nobody uses your product enough to generate interesting data (problem)
Volatile metrics mean:
- People ARE using your product
- You're learning what works and what doesn't
- You have data to iterate
The Danger of Optimizing for Zero Volatility
Here's the real problem:
Many builders optimize for "stability" because they confuse stability with safety.
Result:
- They don't experiment ("might drop the metrics")
- They don't pivot ("need to be consistent")
- They don't ask for brutal feedback ("might demoralize the team")
And they end up with a stable product nobody wants.
No volatility. But all the risk.
The Question That Changed Everything
Marks summarizes it like this: "The question isn't 'what's the probability this goes down?' but 'what do I lose if it goes down and I can't recover it?'"
For builders:
Don't ask: "How much can my metrics drop if I change this?"
Ask: "If I don't change this, what opportunity do I lose permanently?"
That's the difference between managing volatility and managing risk.
Your Framework for This Week
Take a decision you've been postponing because "it's risky."
Ask yourself:
1. Is it real risk (permanent loss) or just volatility (temporary fluctuation)? 2. What do I lose permanently if I DON'T act? 3. Can I reverse it if it goes wrong? 4. Am I confusing emotional discomfort with real risk?
You'll probably discover that the "risky" thing is staying where you are.
And the "volatile" thing is exactly what you need to learn.
---
*PS: Howard Marks has been writing memos since 1990 and they're public. If this resonated with you, look them up. They're the best education in thinking about risk you'll find. And free.*
