The share of the variation in a measure that a regression line explains, from 0 to 1.
R-squared, the coefficient of determination, says how much of the variation in the measure being predicted a regression accounts for. An R-squared of 0.8 means the inputs explain 80% of the spread around the average and 20% is left unexplained. It describes fit on the data used; it does not say the relationship is causal or that it will hold next quarter.
R-squared = 1 − (sum of squared residuals ÷ total sum of squares), where the total sum of squares is each value's squared distance from the mean. In Excel: =RSQ(known_y, known_x) for one input, or the ToolPak's regression output for several.
Monthly sales have a total sum of squares of 500 and, after fitting against marketing spend, a sum of squared residuals of 100. R-squared = 1 − 100 ÷ 500 = 0.80.
Read as proof of cause. Raised by adding inputs, since it never falls when one is added; adjusted R-squared corrects for that. Two series that both trend upward can show a high value with no real link. The full guide is regression in Excel.