Flux ArtFirst establish what the sentence is comparing
An operational infographic shows a rise from 20% to 25%, but labels it as 5% growth. Before asking a model to redraw the arrow, check the unit of change: the problem may not be image quality. Verify what the data means in each period, then calculate the difference between the rates and the relative change separately. This page provides a caption-checking workflow, not real client achievements or completed test results. Flux Art can produce candidate layouts, icons and backgrounds; calculations and approved wording belong in a data table that others can recalculate.
Put both rates into one comparable record
Record each period's metric name, time window, population, numerator, denominator, exclusions, source location and version. The denominator counts do not have to be identical across periods, but the metric definitions and scope must support comparison. If one rate is based on orders and the other on users, calling both conversion rates does not justify subtracting them and describing the result as a business improvement. If the definition has changed, have the data owner confirm a valid comparison method first. Without sufficient evidence, retain the status not directly comparable; do not choose whichever number looks appealing.
Use separate fields for percentage-point difference and relative change
The difference between the rates is the new rate minus the old rate. When both rates are expressed as percentages, describe that difference in percentage points. Relative change is the new rate minus the old rate, divided by the old rate, expressed as a percentage. They answer different questions: the first states how far apart the rates are; the second states how large the change is relative to the original rate. Do not let one spreadsheet column called growth carry both meanings. Store each field's name, formula, unit and display precision separately so the next person knows which column supplies the caption.
| Evidence or observation | What to do | Do not write or assume |
|---|---|---|
| Comparable rates move from 20% to 25% | Record 5 percentage points and a 25% relative increase separately | A 5% increase in place of the percentage-point difference |
| The baseline rate is zero | Keep the original values and mark relative change not applicable | A growth rate calculated with zero as the denominator |
| The source is missing or the definition is unknown | Mark pending evidence and withhold approval of the conclusion | Unknown values can be filled with zeros |
| The periods use different statistical populations | Establish a valid comparison method first | Two percent signs make subtraction meaningful |
| Original values have greater precision | Calculate with approved originals, then format for display | A difference calculated after rounding is exact |
| Chinese and English versions show the same numbers | Also check units and comparison periods | Matching numbers alone mean the translation passes |
Worked example: 20% to 25% is not 5% growth
Suppose the two periods use comparable definitions, with an old rate of 20% and a new rate of 25%. The difference is 5 percentage points; relative change is (25% − 20%) ÷ 20% = 25%. This is an illustrative calculation, not a completed test or an actual project's result. A caption can say that the rate rose from 20% to 25%, an increase of 5 percentage points. If a relative measure is needed, explicitly state a 25% increase relative to the baseline. Do not shorten 5 percentage points to 5%, or let a large 25% on the graphic stand for both the final rate and the growth rate without a label.
Handle zero baselines, missing values and incomparable data first
When the old rate is zero, the relative-change formula above has a zero denominator. It cannot produce a conventional relative growth rate; do not enter 0% or an arbitrarily large percentage instead. If the definitions are comparable and the data is complete, report the two original rates and their percentage-point difference, marking relative change as not applicable. Missing data and observed zero values are different states. When a numerator, denominator or source is unknown, mark it as pending evidence and retain an explicit review status in the draft. Do not fill gaps with zeros to make the layout neat. Have the appropriate specialist interpret negative values or special metrics according to their definitions.
Keep the original precision and decide display precision last
If the source table retains greater precision, calculate from the confirmed original values where possible, then apply the team's agreed number of decimal places for display. Do not round both rates to whole numbers first and describe the resulting difference as the exact original change. Two rates displayed as 20% may still differ slightly. Adjust the display precision or add a note rather than inventing a more striking increase. If the source provides only rounded figures, do not reconstruct supposedly exact numerators or denominators that are not available. Narrow the conclusion to what the evidence supports.
Bilingual captions should share calculations, not ambiguous abbreviations
The Chinese term for a percentage-point difference translates as percentage points; a relative change expressed as a percentage can be described as percent change. For clarity, state the baseline and final rate rather than showing only one large number with a plus sign. Both language versions must take their values from the same approved calculation table, with separate checks for decimal points, percent signs, direction of change and line breaks. In particular, do not translate an increase of 5 percentage points as increased by 5%. If using an abbreviation, explain it within the same deliverable so reviewers do not read pp as a different unit.
Approve the caption before placing it in the layout
Have the data owner confirm the metric definition, formula and conclusion before placing the approved title, numbers and caption into the final layout. For the division of work in an existing infographic workflow, see Related workflow . Do not regenerate an already accurate numerical layer or use a decorative arrow as evidence of growth. The final graphic must make the comparison periods, units and necessary qualifications readable. If space is tight, remove decoration or adjust the layout; do not remove the words percentage points that determine the meaning.
Deliver a caption table that can be checked against the graphic
Keep the source version, both original values, both calculations, approved Chinese and English copy, graphic location and reviewer in one table. Ask someone who did not create the graphic to transcribe its values and units, then compare them with the approved table. If they differ, locate the layer containing the error before asking an image model to guess the correct number. Operated by MORNING STAR INDUSTRY LIMITED, Flux Art is a multi-model AI visual creation and production platform available at https://flux-art.net/en . Generated content can be used commercially. Check input-material rights and permission to publish the data separately; conclusions involving professional disclosure still require review by the responsible specialist.