A correlation matrix answers one question fast: which of my numeric columns move together? Marketing spend and revenue. Temperature and sales. Discount rate and churn. Excel also supports correlation analysis through formulas and the Analysis ToolPak.

Here is a browser workflow and some checks to help you interpret the result.

The basic workflow

Open Vizmatiq's import page — free, no account needed
Drop in your CSV or Excel file (for local analysis, stay signed out and avoid online features; see when data is transferred)
Go to Correlations

That's it. Vizmatiq detects your numeric columns and renders the full matrix as a color-coded heatmap — teal for positive correlations, red for negative, with the coefficient in each cell. Click any cell to see the scatter plot behind the number, which matters more than you might think (more on that below).

What the numbers mean

Each cell holds a Pearson correlation coefficient between two columns, from −1 to +1:

+1 — perfect positive relationship: when one goes up, the other goes up in lockstep
0 — no linear relationship
−1 — perfect negative relationship: one up, the other down

Rough reading guide: |r| above 0.7 is a strong relationship, 0.4–0.7 moderate, below 0.3 usually noise. The diagonal is always 1.0 (every column correlates perfectly with itself) — ignore it.

Three ways a correlation matrix will lie to you

1. Correlation is not causation. Ice cream sales correlate with drowning deaths — because both follow summer. Before acting on a correlation, ask what third factor could drive both.

2. Outliers can manufacture (or hide) correlations. A single extreme row can push r from 0.1 to 0.8. This is why clicking through to the scatter plot matters: if the "correlation" is one point in the corner dragging the line, it's not a relationship, it's an outlier. Clean the data first — here's the fast way.

3. Pearson only sees straight lines. A perfect U-shaped relationship scores near zero. Again: look at the scatter plot, not just the number.

The no-Excel alternatives, compared

This guide is published by Vizmatiq and is not an independent benchmark. The following are different ways to approach the task; no affiliation or endorsement by other providers is implied.

Vizmatiq. Provides a correlation heatmap and click-through scatter plots. Local calculations run in the browser. Signing in enables cloud sync; AI and sharing also transfer data. See our privacy policy.

Google Sheets. The CORREL function calculates the correlation between two ranges. Evaluate the spreadsheet workflow and data handling against your needs.

Python and R. These offer programmable approaches to correlation analysis. Setup, data preparation, and plotting require familiarity with the chosen libraries. No speed comparison has been measured here.

What to do after you find a correlation

A strong correlation is a lead, not a conclusion. The usual next steps:

Plot it over time — does the relationship hold across the whole period, or only in one quarter?
Segment it — a correlation that exists overall but vanishes within every region is a red flag for a confounder
Ask why — if you can't tell a plausible causal story in either direction, treat it as coincidence until proven otherwise

In Vizmatiq you can do all three from the same screen — chart the pair, filter by segment, or ask the AI to investigate the relationship (free tier includes 20 AI calls a day).

Try it now

Try a CSV and check the Correlations tab. Stay signed out and avoid online features if you want to keep this local file workflow in your browser.

How to Make a Correlation Matrix Without Excel (Free, No Code)