How to calculate correlation in Excel
Correlation tells you whether two columns move together, and how strongly — from +1 to −1. Excel gives you CORREL for a pair and a full matrix for many. Here's how, and the one-click alternative.
When to reach for correlation
A single coefficient turns a hunch into a number — but only if you read it with care.
Find drivers
Which inputs actually track with sales or churn?
Trim a model
Spot variables that say the same thing (high correlation).
Sense-check
A surprising correlation is a prompt to investigate, not a conclusion.
How to do it in Excel
One pair takes a single formula; a full matrix needs the Analysis ToolPak switched on.
Line up the columns
Each variable in its own column, one row per record, equal lengths.
One pair with CORREL
=CORREL(A2:A100, B2:B100) returns the coefficient from −1 to +1.
Many at once with a matrix
Data › Data Analysis › Correlation, select the range, get a full matrix. (Enable the Analysis ToolPak first.)
Read it honestly
Near ±1 = strong, near 0 = weak. Correlation is not causation — a lurking third variable can drive both.
The number is easy; the interpretation is where people slip. A strong correlation can be a coincidence, can run the opposite way to what you assume, or can be driven by a third factor moving both columns. Always pair the coefficient with a scatter plot — the shape tells you whether the relationship is real, linear, and free of a single outlier doing all the work.
The faster way — skip the setup
Computing and reading a matrix by hand gets old fast. DataHub Pro builds it, ranks it and plots it from raw values.
Easy number, tricky reading
- CORREL per pair, ToolPak for a matrix
- Enable Data Analysis add-in first
- No visual unless you build one
- Re-run on every new dataset
Matrix, plotted and ranked
- Upload the file — full matrix computed
- Strongest relationships ranked for you
- Each pair shown as a scatter
- Straight into a dashboard or report
FAQ
How do I calculate correlation in Excel?
For two columns use =CORREL(range1, range2), which returns a coefficient from −1 to +1. For many variables at once, enable the Analysis ToolPak and use Data › Data Analysis › Correlation to build a full matrix.
What is a good correlation coefficient?
It depends on context, but as a rough guide values above about 0.7 or below −0.7 are strong, around 0.3–0.7 moderate, and near 0 weak. Always judge it against what's normal for your field.
Does correlation mean causation?
No. A high correlation only means two things move together in your data. The cause could run either way, be coincidental, or be a third variable driving both — so confirm with a scatter plot and domain knowledge before acting.
Keep exploring
More visual analysis you can run on your own data — no sign-up needed to read.
See what really moves together
Upload an Excel or CSV file and DataHub Pro builds the correlation matrix, ranks the strongest pairs and plots them for you. Free to try, no card.
Start free →