Ask AI to improve chart labels without changing the numbers
Changing a chart title from revenue to profit doesn't change the data. It changes what your audience believes the data says.
Example output
A chart with the right measure and unit
Example chart (fictional): monthly revenue in GBP, January–March 2026 — £120,000, £150,000 and £100,000.
Read the prompt
Improve this chart's labels and annotations for [audience].
Use the attached table and measure definition as the authority.
Keep every source value, period and denominator unchanged.
State the measure, currency/unit and date range in the chart.
If you scale displayed values, label the scale and show the conversion.
Do not infer profit, causation, a forecast or an explanation of change
from a revenue series alone. Return the chart and a label-change record.What you'll make
a chart whose title, units, period and notes describe the numbers accurately, plus a short record of what changed.
What you'll need
the chart’s source table (a screenshot alone can hide the data), what the measure means, its unit and period, and who will read it. A work AI chat — ChatGPT, Claude, Copilot or Gemini — and the chart in Excel, Google Sheets or PowerPoint. About ten minutes.
1. Tell it what must not change
Attach the chart and its table and ask:
Suggest clearer labels and notes for this chart for [audience]. The attached table and this definition are the source: [measure definition].
Keep every value, period and unit unchanged. The title must name the measure, the unit and the date range. If you scale the values, label the scale and show the conversion.
Do not suggest profit, causes, forecasts or reasons for change that the table doesn't show. List each label you would change, old and new.[audience] is who reads it, such as “the leadership team”. [measure definition] is what is counted, such as “invoiced revenue, GBP, excluding VAT”.
Why this works: naming the measure up front stops a friendlier title quietly turning revenue into profit, and the old-and-new list makes every change reviewable.
What you should see: a short list of label changes, each with the old wording and the new.
2. Trace one point from the table to the chart
Pick one point. Read its row in the table, then the axis label and the value on the chart. Do they agree, unit and all? Then check the highest and lowest points the same way. “Revenue (GBP)” and “Revenue (£000)” can both be right if the values match the scale. A unit change without a conversion is a numerical error.
Why this works: three points take a minute and catch most scale and unit mistakes, which are the ones that look most convincing.
What you should see: the same value, unit and period in the table, on the axis and in any note.
3. Make the change yourself where you can
Most label fixes are quicker by hand. In Excel or PowerPoint, click the chart title or axis title and type. In Google Sheets, double-click the chart, choose Customize › Chart & axis titles and edit the text. Ask the AI to regenerate the chart only when the scale or a conversion is wrong.
Why this works: retyping a title can’t change a value. Regenerating a chart can.
What you should see: the new labels on the chart, and the values exactly as before.
Check the result
- The title names the measure in the table, not a different one.
- The unit and scale match the values shown.
- The period covers exactly the rows in the table.
- No note gives a reason the table doesn’t contain.
The usual failure: the new title changes the measure or the scale. “Profit (£m)” over a revenue series in pounds is two errors in one tidy label: the wrong measure and a scale a million times too big. Spot it by reading the title aloud against the table’s column heading.
The source measure is [measure and unit], not [wrong label]. Correct the title and axis unit, keep every value and period unchanged, and remove any note that gives a reason the table doesn't show. List exactly which labels changed.[measure and unit] is what the table says, such as “Revenue in GBP”. [wrong label] is what the chart now says.
Worked example: the revenue that became profit
Fictional example. A finance assistant asked for a clearer title on this chart:
month,revenue_gbp
January 2026,120000
February 2026,150000
March 2026,100000The suggestion came back as: title “Profit (£m)”, note “Spring campaign drove February’s growth”.
- Two errors in the title. The table has no costs, so it can’t show profit, and the values are in pounds, not millions. The repair prompt, with “Revenue in GBP” and “Profit (£m)” filled in, gave “Monthly revenue (GBP), January–March 2026”, the label on the example chart at the top of this page.
- The note had to go. Nothing in the table mentions a campaign. A correct note says February is the highest of these three months, at £150,000.
- The fix was typed, not regenerated. The values were already right, so the new title went in by hand. The chart still reads £120,000, £150,000 and £100,000.
Task kit for this job: Download (ZIP) · Reusable instructions
Go deeper
Fit the notes to the decision
A leader comparing months needs the period and the values. A finance reviewer also needs the measure’s definition, such as whether it excludes VAT. Ask for notes aimed at the reader’s decision, from the same table.
Handle missing months separately
A missing month shouldn’t become zero on a line chart. The missing-months guide shows how that changes the line and the interpretation, and explain a chart without inventing covers the difference between what a chart shows and why.