Stellar makes it easy to set up, run, and analyze A/B tests with built-in Bayesian statistics.
Bayesian statistics directly answers the question "What is the probability that B is better than A?" This is much more intuitive than interpreting p-values or confidence intervals.
Unlike frequentist methods, Bayesian analysis allows you to evaluate results at any time, without requiring predetermined sample sizes or stopping rules.
Bayesian methods provide probabilities that directly inform business decisions, helping you understand the risk associated with choosing one variant over another.
Bayesian methods work well even with smaller sample sizes, giving you useful information earlier in your testing process.
Stellar's script is just 5.4KB - at least 10 times smaller than other A/B testing solutions. It loads instantly without affecting your site performance, SEO ranking, or Core Web Vitals.
At just 5.4KB and served on a CDN, Stellar's pure JS script is 25x smaller than competitors like VWO or AB Tasty. With minimal dependencies and a fixed size regardless of test count, it preserves your website's speed, Core Web Vitals, user experience, and SEO performance.
Bayesian A/B testing is an approach to statistical analysis that uses Bayes' theorem to update the probability of a hypothesis as more evidence becomes available. In A/B testing, it directly calculates the probability that one variant is better than another, rather than relying on p-values and confidence intervals.
Traditional frequentist methods calculate the probability of observing your data given a hypothesis (p-value), while Bayesian methods calculate the probability that your hypothesis is true given the observed data. This makes Bayesian results more intuitive and directly applicable to business decisions.
Common thresholds are 95% for high confidence or 90% for moderate confidence. However, the appropriate threshold depends on your specific business context and the cost of making a wrong decision. For low-risk changes, you might accept a lower probability like 80%.
Yes, one advantage of Bayesian testing is that you can evaluate results at any time without penalty. Unlike frequentist methods, there's no need for predetermined sample sizes or stopping rules. You can check results continuously and make decisions when the probability reaches your desired threshold.
Up to 25k MTU
Built-in analytics
Visual web editor
Custom CSS & JS editor
AI editor assistant
Basic A/B testing
Limited variants & goals
Limited concurrent experiments
50k MTU included
Unlimited variants
Unlimited concurrent experiments
Advanced targeting rules
Dynamic keyword insertion
Priority support
Custom MTU volume
Custom solutions
Integrations support
Unlike competitors who charge in fixed large blocks, we dynamically bill in small increments of 1k MTU, ensuring you never overpay
Compared to average competitors like VWO's $4,700/year for 50k MTU, our equivalent annual cost is just $1,118
We only count users who actually see experiments. Multiple experiment views from the same user count as just 1 MTU