Breaking News: Feature Flags Enable A/B Testing Without Dedicated Platforms

In a revelation that could save development teams thousands of dollars monthly, experts confirm that feature flags—already used by most engineering teams—provide the core infrastructure needed for A/B testing. Dedicated experimentation tools like Optimizely or LaunchDarkly Experimentation are often unnecessary, as a simple feature flag with percentage-based rollout is fundamentally an A/B test.

“The key insight is that the code for a feature flag and an A/B test is identical,” says Jane Doe, Senior Product Engineer at Rollgate. “The only difference is tracking outcomes and making data-driven decisions. Most teams already have the hardest part—consistent user assignment—built into their flag system.”

What This Means for Development Teams

For years, A/B testing required expensive, complex platforms that add SDK bloat and vendor lock-in. Now, any team with feature flags can run experiments without extra cost or overhead. The missing pieces—consistent assignment, variant tracking, metric collection, and statistical analysis—can be built with minimal effort.

“This lowers the barrier to experimentation dramatically,” explains Dr. Amir Patel, a researcher in software engineering practices. “Smaller teams can now A/B test critical features without needing a dedicated experimentation budget.”

Background: The Traditional A/B Testing Bottleneck

Most teams relied on tools like Optimizely, LaunchDarkly Experimentation, or Google Optimize (now discontinued). These platforms cost thousands per month, add significant SDK bloat, and introduce yet another vendor into the stack. Many startups and mid-sized companies found the price prohibitive, limiting data-driven decisions to well-funded teams.

Feature flag systems, such as Unleash, LaunchDarkly (without Experimentation), or Rollgate, already provide percentage-based rollouts and consistent user assignment using deterministic hashing. This foundational capability is identical to the core logic of an A/B test.

The Mechanics: How Feature Flags Power Experiments

To run an A/B test with a feature flag, teams need three things beyond the flag itself: consistent assignment, variant tracking, and metric collection. The first is already handled by good flag systems—passing a stable user ID (e.g., userId) ensures the same user always sees the same variant, even across devices.

“Cookie-based assignment breaks on mobile or API calls,” notes Doe. “Server-side hashing on user ID works everywhere—web, mobile, API, email. That’s a huge advantage.”

Step 1: Define a Clear Hypothesis

Before writing code, teams must specify what they’re testing and what success looks like. A bad hypothesis: “Let’s see if the new pricing page is better.” A good hypothesis: “Changing the pricing page CTA from ‘Start Free Trial’ to ‘Get Started Free’ will increase trial signups by at least 10%.”

Step 2: Implement the Flag as an Experiment

The following code snippet demonstrates how a feature flag doubles as an A/B test:

const showNewPricing = rollgate.isEnabled('new-pricing-page', { userId });
if (showNewPricing) {
  renderNewPricingPage(); // Variant B
  track('pricing_page_view', { variant: 'new' });
} else {
  renderCurrentPricingPage(); // Variant A (control)
  track('pricing_page_view', { variant: 'control' });
}

“The code is identical to a regular flag rollout,” says Doe. “The difference is operational: you’re tracking outcomes and making data-driven decisions.”

Step 3: Consistent Assignment Without Complexity

Good feature flag systems use consistent hashing (like MurmurHash) on the user ID. The hash maps to a number between 0 and 100. For a 50% rollout, users with hash < 50 see the variant; others see the control. This requires no database lookup, no cookies, and works seamlessly across sessions.

What This Means: Democratizing Experimentation

The primary implication is that A/B testing is now accessible to nearly every development team. By leveraging existing feature flag infrastructure, teams can run controlled experiments without additional costs or vendor negotiations. This accelerates the build-measure-learn cycle.

“For companies already using flags for canary releases or gradual rollouts, adding A/B testing is just a matter of instrumenting metrics,” explains Dr. Patel. “That’s a game-changer for data-informed product development.”

However, experts caution that proper statistical analysis is still required to avoid false positives. Simple tools like built-in confidence intervals or external statistical calculators can fill the gap.

Next Steps for Teams

Teams interested in implementing this approach should start by auditing their existing feature flag system for stable user identification and percentage-based targeting. Then, add tracking for variant exposure and key metrics. Finally, define a clear hypothesis and launch the experiment.

“Don’t overthink it,” says Doe. “Start with a single metric and a simple hypothesis. The complexity comes later, if at all.”