When I’m faced with a complex dataset, I don’t start by picking a chart type but start by asking what decision this visualization should help drive. Sometimes that means simplifying layers of data into multiple small visuals rather than one “big” chart.

Other times it’s about showing contrast: e.g., a side-by-side heatmap and scatter plot to show both macro patterns and micro relationships.

Also found that the best visualizations often come from iteration. The first draft is rarely the final one , it takes a few rounds of refining, testing with the audience, and even discarding flashy visuals if they distract from clarity. Tools help (Tableau, Plotly, even pen-and-paper sketches), but the real process is balancing insight with comprehension. After all, a chart only succeeds if the right person understands it at the right time.