Scatter Plot

A scatter plot maps each data point to a position using two numeric variables — one on the x-axis, one on the y-axis. The resulting cloud of points reveals the relationship between the variables: whether they move together (correlation), form clusters, or contain outliers. It's the go-to chart for exploring how two measures relate, such as ad spend versus conversions across campaigns.

Use one when you want to see the relationship between two continuous variables across many observations — does CPC rise with competition, does watch time relate to conversion rate. Scatter plots excel at exposing correlation, clusters, and outliers that summary statistics hide. Avoid them for categorical comparisons (use bars) or when you have only a few points, where the pattern isn't meaningful.

Look at the overall shape of the cloud. Points trending up-right suggest positive correlation (as x rises, y rises); down-right suggests negative. A tight band means strong correlation; a diffuse blob means weak or none. A trend line makes the direction explicit. Crucially, correlation isn't causation — a relationship in the points doesn't prove one variable drives the other.

Add one when you want to summarize the direction and strength of the relationship, or to make a subtle correlation visible in a noisy cloud. A linear fit is the common default, but match the line to the pattern (it shouldn't be linear if the relationship curves). Always keep the raw points visible — the line is a summary, and hiding the points hides outliers and the true spread.

A bubble chart is a scatter plot with a third variable encoded as the size of each point. So x and y give position as usual, and the bubble's area shows a third measure — for example x = spend, y = conversion rate, size = revenue. Use a bubble chart when a meaningful third numeric dimension would otherwise be lost; keep it a plain scatter plot when two variables tell the story, since over-sized bubbles can overlap and mislead.