Your data
Paste a table straight from Excel or Google Sheets, or drop a CSV file. Nothing is uploaded anywhere — the whole calculation runs inside this browser tab.
CSV, TSV or plain text. Files saved in windows-1251 are detected and re-read automatically, so Cyrillic headers do not turn into garbage.
Parsing options
Result
Save and share
The drawing library is downloaded only when you press the button, and only on this page.
This page has no server side at all. The table you paste is parsed by JavaScript inside your own browser, the chart is drawn on a canvas element on your machine, and the export file is assembled locally. Nothing is uploaded, stored or written to any log. You can disconnect from the network after the page has loaded and everything will still work — which is the simplest way to verify the claim yourself.
Facts and limits of this method
| Coefficient | Pearson r for every pair of numeric columns |
|---|---|
| Diagonal | Always exactly 1 — every column correlates perfectly with itself |
| Symmetry | The matrix is symmetric; the halves above and below the diagonal are identical |
| Missing values | Handled pairwise: each cell uses the rows where both of its columns are present |
| Strongest pair | Reported separately, excluding the diagonal |
| Column limit | Practical readability ends around 10 columns — that is already 45 distinct pairs |
When it misleads you
- The multiple comparisons problem is severe here and almost never mentioned. Ten columns produce 45 pairs; at the usual 5% threshold you should expect two or three to look significant by pure chance. A correlation matrix is a generator of hypotheses, never a source of conclusions.
- Every cell is a Pearson coefficient, so every Pearson limitation applies forty-five times over: linear relationships only, and one outlier can manufacture a strong-looking pair.
- Pairwise handling of missing values means different cells can be computed from different subsets of rows. Two coefficients in the same matrix may not be based on the same underlying sample.
- Correlation between two columns says nothing about what happens when a third is held constant. Columns that both grow with time correlate with each other regardless of any real connection.
How it is calculated
Every numeric column is paired with every other, and Pearson r is computed for each pair. The diagonal is 1 by construction and the matrix is symmetric, so the information content is the triangle above the diagonal — the rest is there for reading convenience.
Missing values are handled pairwise rather than by dropping entire rows. A row missing one column still contributes to every pair that does not involve that column, which preserves data but means the cells are not all computed on the same rows.
The strongest pair excluding the diagonal is reported at the top, because that is the first thing anyone looks for and scanning a grid for the largest absolute value is exactly the kind of task a person does badly.
In the sample data, ad spend, sales and visits all rise together across the period. Their mutual correlations are near 1, which illustrates the trap: the picture is real, the arithmetic is correct, and the causal story it suggests may be entirely wrong.
Questions and answers
How do I read the matrix?
Find the row of one variable and the column of the other; the cell where they meet is their correlation. Values near ±1 mean a strong linear relationship, near 0 means none. The diagonal is always 1 and carries no information.
Several pairs look significant. Can I trust them?
Be careful. With ten columns there are 45 pairs, and a few will look strong by chance alone. Treat anything you find here as a hypothesis to test on fresh data, not as a result.
Why are three of my columns all correlated with each other?
Very often because all three are growing over time. Shared trend is the most common cause of a cluster of high correlations, and it says nothing about any direct connection between them.
Can I get Spearman correlations instead?
Not in the matrix — it uses Pearson throughout. For a specific pair where outliers or curvature are a concern, use the dedicated Spearman page.
Is my data uploaded?
No. Every coefficient is computed in your browser.