Does That Stat Graphic Clarify the Number or Just Decorate It
You are a few posts into an agency’s portfolio and you hit a stat card. Big number, soft gradient behind it, a smaller line underneath, an icon in the corner. It looks competent. Whether it communicates is a different question, answerable in about two seconds.
What a stat graphic is actually supposed to do
A stat graphic has one job: make a number, or a comparison between numbers, faster to grasp than the same information in a sentence. If a reader gets the point quicker from plain text, the graphic failed at the only thing it was built for.
Notice what the test excludes: whether the number is real, or means what the caption implies. This piece asks one thing, did the design make the number easier to read.
The signs a graphic is decorating, not explaining
All of this is visible on the asset itself. No caption or backstory required.
- An axis with no label or no scale. A line climbs, a bar rises, and nothing says what is counted or over what period. The shape does the persuading, unanchored.
- A chart shape with no reference point. A pie, donut, or bar hands you a proportion with nothing to weigh it against. You cannot tell whether the slice is large, small, or entirely ordinary.
- Effects layered on the number itself. Gradients, bevels, drop shadows, an outline in a second colour. Ask what any of it adds to reading the digits. Usually it lowers contrast, the opposite of the job. This is chart junk, ink carrying no information.
- Subtext that competes with the headline. The supporting label sits close to the main number in weight, size, or colour, and your eye cannot tell which to read first. Two elements at the same volume means neither is the headline.
- Colour that changes meaning mid-graphic. The accent marks the emphasised figure in one panel and the comparison figure in the next. Once colour stops being a consistent code it is decoration wearing the costume of information.
What a well-built stat card looks like instead
Picture a card built by someone who cared. The number is unmistakably the largest, highest-contrast element, so your eye lands there first. Beneath it sits one supporting line, clearly secondary in weight, saying what the number counts. Not three lines. One.
If there is a comparison, its baseline is consistent and stated once, legibly: same unit, same time frame, same population on both sides. You can see you are comparing like with like rather than assuming it. One accent colour does real work, and it means the same thing everywhere.
Restraint is the tell. The interesting question about a good card is what is missing: no second chart, no third colour, no icon added because a corner felt empty. Resisting those additions is a decision, and across an account it becomes a visual system, not just one good post.
Judge it at the size it’s actually seen
Most portfolio pages show stat graphics at a size no audience ever saw, full bleed on a desktop. Pull the asset up on a phone at feed thumbnail width instead, because that is the size it competed at. Plenty of graphics survive full-screen scrutiny and collapse there. Supporting type drops below legibility, subtle contrast becomes grey on grey, and the hierarchy turns out to have rested on detail that vanishes: a thin rule between figures, a superscript unit, a faint tint difference between bars. If the point only arrives once you zoom in, the design handed its job to the viewer.
Chart type is a separate decision from chart style
The mismatches are obvious once named. A pie or donut showing change over time is the common one: a pie has no direction, so it cannot show a rise or a fall, where a line or a bar pair shows both instantly. The other is a bar chart comparing exactly two points, dragging in an axis, gridlines, and a legend to deliver what a single labelled comparison delivers faster. Styling rescues neither. A mismatched chart type can be beautifully executed and still be the wrong tool, and a well-chosen chart can be executed badly. Both judgments count, but only style announces itself; fit has to be asked about on purpose.
Where this stops being a design question
A graphic can be ruthlessly clear about a number that means very little, and a clumsy one can present something entirely sound. Legibility and honesty are separate judgments, neither predicts the other, and this piece makes only the first. Portfolio material looks more finished than it is, which is also what a portfolio leaves out.
A five-point check for the next stat graphic you scroll past
- Is there a labelled axis or scale wherever one is needed, or does a shape rise without saying what it counts?
- Is the headline number unambiguously the visually dominant element?
- Does any comparison use a consistent baseline, stated on the graphic rather than assumed?
- Does colour mean the same thing in every part of the graphic it appears in?
- Does it still read in under two seconds at actual feed size on a phone?
Five yeses is rarer than it sounds, and rarest on a template-heavy account, where judging craft on an account with almost no room to move is harder still.
Work worth looking at closely
See this year’s selections and read the work with this checklist in hand. If your agency’s data graphics hold up at thumbnail size, you can nominate an agency.
FAQ
Does a clean, well-designed stat graphic mean the number behind it is trustworthy?
No. Legibility and honesty are two separate judgments. A graphic can make a meaningless number easy to read, and an ugly one can present something sound. This piece covers only the first.
What is the fastest way to test whether a stat graphic actually works?
View it at feed thumbnail width on a phone, not zoomed into a portfolio page. If the headline point does not land in roughly two seconds there, the design has not done its job, however polished it looks larger.
Is adding more chart detail, more colour, or more elements ever the safer choice?
No. Extra elements add noise more often than clarity, and each takes attention from whatever you wanted read first. One dominant number, one accent colour, one clear comparison is a deliberate decision, not a shortcut or a sign of low effort.
Sources
- Chartjunk, Wikipedia, on the term for decorative graphic elements that convey no information, coined by Edward Tufte in 1983. Fetched 2026-09-06.
