Fail to Win Big

“Go Fail. That’s an Order.”

The first time my manager told me to run an experiment designed to fail, I thought he had finally lost it.

“We’ve been doing A/B tests for months, optimizing every little thing. Now, you want me to… sabotage them?” I asked, my voice laced with horror.

He grinned. “Not sabotage. Strategic failure.”

And that’s how I learned about reverse experimentation — where failing on purpose teaches you more than succeeding ever could. It initially felt counterintuitive, like stepping on a banana peel just to see how you fall. But the insights? Pure gold.

What is Reverse Experimentation?

Reverse experimentation is the art of intentionally testing “bad” ideas to see how users react. Instead of optimizing for conversion, you optimize for chaos — removing features, adding friction, or tweaking the experience in unexpected ways.

Why do this?

• Uncovers hidden dependencies — What do users actually rely on?
• Reveals “dealbreaker” features — What makes them rage-quit?
• Highlights compensatory behavior — How do they adjust when a feature is gone?
• Challenge assumptions — Maybe that “crucial” feature isn’t that crucial after all.

Why is Reverse Experimentation Important?

Normal experiments show what works, but reverse experiments show what users can’t live without.

This helps teams:

• Identify high-impact features — If disabling something tanks engagement, it’s clearly vital.
• Prevent silent user churn — Spotting subtle pain points before they drive users away.
• Find unexpected behaviors — Users hacking workarounds = hidden insights.
• Make smarter product decisions — Double down on what truly matters.

Think of it like unplugging a random wire in a machine to see what stops working.
Surprising? Yes. But incredibly useful.

Normal vs. Reverse Experimentation

At their core, both normal and reverse experimentation aim to improve a product, but they take dramatically different routes to get there.

Traditional A/B Testing: Optimizing for Success

• Focuses on incremental improvements by testing small variations (e.g., button colors, CTA wording, new feature rollouts).
• The goal is to enhance engagement, retention, or conversion rates by identifying what works best.
• Experiments typically add or refine features based on hypotheses of what might improve user experience.
• Asks: “Which version drives better results?”
• Example: Testing two different onboarding flows to see which one leads to higher activation rates.

Reverse Experimentation: Learning from Failure

• Instead of adding or tweaking, it removes, alters, or degrades a feature to observe user reactions.
• The goal is to identify non-negotiable features, dependencies, and hidden user behaviors.
• Instead of assuming a feature is valuable, reverse experiments challenge its necessity.
• Asks: “What happens when we take this away?”
• Example: Hiding Instagram ‘likes’ to see how user behavior changes in the absence of public validation.

While traditional A/B tests optimize existing pathways, reverse experimentation stresses the system to uncover weak points and must-have features.

It’s like testing a parachute by opening it in freefall vs. testing it by not opening it at all and seeing what happens. One is safer. The other is… well, more educational.

Who’s Nailed Reverse Experimentation?

Netflix’s Playback Speed Experiment

Netflix tested variable playback speeds (0.5x to 1.5x) and found that users either loved it or hated it. Filmmakers protested, arguing that it disrupted creative intent. This experiment exposed a previously unknown power user segment — those who binge-watch at higher speeds — leading Netflix to refine and selectively roll out the feature.

Instagram’s Hidden Likes Experiment

Instagram tested removing public ‘like’ counts to measure their impact on mental health and content creation. Some users felt relieved from social validation pressure, while influencers complained about engagement transparency. This test helped Instagram understand the psychological role likes play in content consumption.

Facebook’s News Feed Adjustment

Facebook reduced public content visibility in users’ feeds to prioritize personal connections. They found that engagement dipped, but meaningful interactions increased. This experiment reshaped Facebook’s approach to algorithm-driven content delivery.

Twitter’s Reverse Chronological Timeline

Twitter allowed users to switch from an algorithmic feed to a reverse chronological one. The experiment revealed power users preferred real-time content, influencing how Twitter balanced algorithmic recommendations.

Final Thoughts? Go Fail (Smartly).

Reverse experimentation isn’t about deliberately ruining your product — it’s about strategically removing pieces to see what truly matters. Sometimes, breaking things on purpose is the best way to build them better.

Now go forth and fail spectacularly.