A Q-Q plot draws empirical quantiles against standard-normal quantiles with a mean ± SD reference line, so points hugging the line indicate normality.
A Q-Q plot is the most useful single visualisation for normality assessment. Empirical quantiles (sorted observations) go on the y-axis, theoretical normal quantiles (z) on the x-axis. If the data are normal the points fall on a straight line passing through the mean with slope SD.
The shape of the departure tells you exactly what's wrong. S-shape ⇒ heavy tails (Cauchy, t-distribution). Inverted S ⇒ light tails. Convex curve ⇒ right skew. Concave curve ⇒ left skew. A normality p-value can tell you something is off, but only the Q-Q plot tells you what.
One numeric column.
Scatter of (theoretical z, observed value) + reference line.
Slight wiggles at the extremes are normal even for normal data because the order statistics there are noisy — focus on the body of the plot.
If only a handful of points at the extremes lift off the line and the centre is straight, the issue is outliers, not the distribution shape. Re-check with the box plot + ROUT.