Upload your raw experiment export, map the variant and outcome columns, and get the right statistical test, the uplift with a 95% confidence interval, and a plain-English verdict. Free.
or click to browse · max 3 MB
Running your significance test...
Sent to — per-variant rates or means with confidence intervals, the statistical test results, uplift with 95% CI, a plain-English verdict, R code, and AI insights.
Analyze another fileThe tool detects your outcome type automatically. A conversion-style outcome (0/1, yes/no, TRUE/FALSE) gets per-variant conversion rates with Wilson confidence intervals and a two-proportion test (or chi-square plus pairwise tests vs control when there are more than two variants). A numeric outcome gets per-variant means with t-based intervals and a Welch t-test with Cohen's d (or ANOVA plus Holm-adjusted pairwise Welch tests vs control). Either way you get the absolute and relative uplift, a 95% confidence interval on the difference, and a plain-English verdict.
Use it the moment an experiment ends — or before ending it — when you have row-level data and need a defensible significant / not-significant call.
Not for repeated measures of the same user, sequential (peeking) analysis, or outcomes that are categories with more than two levels.
Built for: Growth, product, and marketing teams reading their own experiment exports
Typical data source: A CSV export with one row per user/session, the variant assignment, and a conversion flag or numeric metric
One row per user: which variant they saw and what happened. The outcome can be a 0/1 flag, yes/no, or a numeric value:
Minimum 20 rows · Best with 500-100,000 rows and 2-4 variants
Standard-library analysis: upload row-level experiment data (one row per user or session), map the variant column and the outcome column, and get a definitive read on your test. The tool detects whether your outcome is a conversion flag or a numeric value, picks the right statistical test (two-proportion, chi-square, Welch t-test, or ANOVA), and returns the uplift with a confidence interval and a plain-English verdict on significance.
Each variant's sample size and conversion rate (or mean) with a 95% confidence interval — the raw material behind the verdict.
The gap at a glance: bars with error bars per variant. Overlapping intervals usually mean the test hasn't decided yet.
Every comparison vs the control: absolute difference, relative uplift, 95% CI, and p-value, with a yes/no significance call.
Plain-English interpretation — what the numbers mean, what's significant, and what to do next.
Did my split test actually win?
Upload one row per user with the variant they saw and whether they converted. The tool runs the right significance test, reports the uplift with a confidence interval, and tells you plainly whether the result is real — and if not, roughly how much more data you'd need.
See our FAQ for details on pricing, data privacy, and how the analysis works. Every report includes a Methodology section showing the statistical test, assumptions checked, and diagnostics run.
Run any analysis on your own data — validated R analyses, interactive reports, AI insights, and PDF export.
Try Free — No Credit Card