Goodbye Traditional Statistics

Statistics for the Internet Age is Here

Data has changed, Statistics should too

Traditional statistics were developed over 100 years ago, for an offline world. These methods are no longer up to the task of helping businesses make decisions with real-time data. If you look at results in real time, as most optimization platforms allow you to do, you'll see winning results where no winner actually exists. In fact, your error rate can skyrocket to over 30% if you don't take precautions.

Without taking precautions, like setting a sample size and committing to a minimum detectable effect in advance, you could make changes to your website or product that at best, have no impact, and at worst, could be hurting your bottom line.

Faster decisions you can trust for digital experimentation

At the core of Optimizely’s experimentation platform is our Stats Engine: a unique approach that was developed in conjunction with leaders in statistical research at Stanford University. Stats Engine embeds several innovations to deliver what every digital experience optimization effort needs—accelerated speed and greater scale of experimentation while also improving the accuracy of results and outcomes.

Introducing always-valid results

Results are valid whenever you check

With traditional statistics, peeking at your results any time other than your set sample size increases the chance you’ll find a winner when there isn’t one. With Stats Engine, check whenever you want. With a method called sequential testing, you’ll see your statistical significance increase with more time and data—no more waiting.

Experiment with as many variations and add as many metrics as you want

Adding many metrics or variations to an experiment also increases your chance of finding a winning result that isn’t real. Optimizely’s Stats Engine corrects for multiple goals and variations with false discovery rate control. Now you can add as many as you want without sacrificing accuracy.

Optimizely's Stats Engine increases confidence

With Stats Engine

setting a 95% significance
level results in a


chance you'll make the right decision

With traditional statistics

setting a 95% significance
level results in a


chance you'll make the right decision if you don't avoid pitfalls*

* We simulated millions of A/A tests under both frameworks and calculated error rate by determining the percentage of experiments that declared a winning or losing variation given a continuous monitoring policy. (1000+ visitors)

Interested in how Stats Engine affects your results?

Statistics has never been easier

Always-valid Statistics

Make a statistically sound decision whenever you look at your results, as soon as data is available.

Significance on Your Schedule

No need to commit to a sample size or duration in advance. You can also see conclusive results up to twice as fast.

Leave the Stats to Us

No need to change how you experiment. Stats Engine works exactly the way you expect it to, without any extra work.

Start making better decisions today

Lorem ipsum dolor sit amet, consectetur adipiscing elit.

Suspendisse enim nulla, hendrerit sed rutrum quis, sagittis finibus purus. Suspendisse tristique neque sed neque mattis, eu commodo ex sodales. Etiam massa quam, scelerisque scelerisque magna sed, congue imperdiet mauris. Praesent vel egestas tortor.