Here are my personal best practices (subjective):
Generally important:
Settle for a main message and try to most easily and effectively communicate it
Name your plot axes appropriately, e.g., „hours“, „dollars“, „dollars per hour“
If the axes titles do not already make it fully clear, indicate what unit the axis is representing, e.g., „Expenditure (in percentage of total daily USD spending)“
If your axis scales have a zero position, include it, e.g., in a bar chart with the height representing the count of items or for monetary values
If you use transformed axes, clearly indicate this, e.g., for a log-scale transformation
Use colours to encode information (nowadays, even scientific journal usually print in colour), e.g., food price series in „orange“ and energy price series in „blue“, if you compare two time series of prices
Remember that colours invoke emotions, e.g., „red“ for losses and bad events, „green“ for positive things
People are naturally drawn to bright colours and bold things first in a graph, so encode the most important thing for your main message in the brightest colour, e.g., they will usually spot the red line first and the rest of the plot only later
Otherwise, we (usually) read from left to right, up to down, so place the most important things in a title or graph ideally on the left-upper side
Always include a legend if you use colours
Remember that some people are colour blind
Only use 3-dimensional plots if the third dimension is necessary to convey your message
For bar charts: if you have categorical values on the x-axis and (many) text labels describing these, simply flip your axes so the text labels are on the y-axis instead
Use annotations and text labels to emphasize important values
Most important: do not use pie charts (never ever ;)), but bar charts instead (percentage values in pie charts are hard to read for us humans)
Depending on your audience:
- Applied: Use action titles for your main message, e.g., „Prices increased five-fold due to supply chain issues“
- Applied: Tell a story with one or multiple plots
- Scientific: Neutrally describe what your graph is showing (descriptive language)
- Scientific: Leave it to the reader to assess the evidence in the graph and come to a conclusion