How to Get Engineering Buy-In for A/B Testing

"We don't have time to add experiment flags."

"Just ship it and we'll see if metrics change."

"Testing slows us down."

Sound familiar? Getting engineering buy-in for A/B testing is one of the biggest challenges product managers face.

Here's how to change that.

Understand their objections

Before you can address resistance, you need to understand it. Most engineering objections fall into three categories:

1. Time concerns

Engineers worry that adding experiment infrastructure takes time away from building features.

The reality: Modern A/B testing platforms (Amplitude, Statsig, LaunchDarkly) make implementation straightforward. A feature flag is often just a few lines of code.

Your response: "The first experiment takes 2 hours to set up. After that, it's 15 minutes per experiment."

2. Complexity concerns

Engineers worry about maintaining multiple code paths.

The reality: Good experimentation practices include cleanup. Winning variants get shipped, losing variants get removed.

Your response: "We'll add experiment cleanup to our definition of done. No experiment runs longer than 4 weeks without a decision."

3. Ownership concerns

Engineers feel like experiments are the PM's thing, not theirs.

The reality: When engineers own experiment results, they build better products.

Your response: "I want you to be part of forming hypotheses, not just implementing them."

Make it personal

The key to buy-in is making engineers care about experiment outcomes.

Predictions help here. When engineers predict whether their implementation will win, they're invested in the outcome.

Try this: Before launching an experiment, ask the engineer who built it to predict the outcome. Make it public. Create friendly competition.

Show the ROI

Nothing convinces engineers like data. Track and share:

• How many experiments prevented shipping bad features
• Revenue impact of winning experiments
• Time saved by killing losing variants early

Give credit

When experiments win, make sure the engineer gets credit alongside the PM. Experiment wins should be career wins for everyone involved.

Start small

Don't try to experiment on everything at once. Pick one high-stakes feature and run a proper experiment. Use that success to build momentum.

The bottom line

Engineering buy-in comes from involvement, not mandates. Make engineers partners in experimentation, not implementers of it.