CSV comparison.
Import a CSV. Discover its stats. Compare what matters.
Projects & datasets
Each dataset keeps its own columns, selections, and version baseline.
Column setup
Detected automatically. Adjust the label, comparison groups, or column roles.
Comparison session
Autosaved in this browser.
Comparison table
Select rows to compare their numeric values.
Stat comparison
Bars share the same scale for the selected stat.
◇ Balance signals
Data quality
Checks cover all rows, including hidden columns.
Compare CSV versions
The table above is the newer version. Import an older CSV as the baseline.
How analysis works & supported files
Use a CSV with one header row and one entry per row (up to 2 MB, 2,000 rows and 80 columns). Quoted commas and multiline fields are supported. Empty or duplicate headers are assigned unique names. Missing cells remain missing, not zero.
Automatic detection separates text, numbers, percentages, and likely identifiers. Adjust the column roles below if needed. Numeric values support decimals, negative values, scientific notation, and quoted thousands separators. Percentages remain in percentage points. Dates and currency text are not automatically converted.
Outliers use the interquartile range (IQR): values below Q1 − 1.5 × IQR or above Q3 + 1.5 × IQR are flagged within each group. At least four valid values are required. If IQR is zero, values different from the shared quartile are flagged. Signals are statistical differences, not balance errors. Filters do not change the comparison baseline.
Import and analysis happen in your browser. Your session automatically saves in this browser. Export a session backup to move it to another device. Export preserves original columns and values.