What FairCharts actually does
FairCharts reviews a chart, points out what is misleading or unclear, and hands back a corrected, presentation-ready version with the reasoning behind every change. It is built on decades of visualization research, so each fix is tied to a named principle, not a matter of taste.
It is critique first. You bring a chart that is almost ready for a client, and FairCharts tells you what to fix before they see it. It stays quiet when a chart is already honest, so a clean bill of health means something.
Three questions we ask of every chart
A chart can be wrong in three different places, and they are not interchangeable. FairCharts checks all three, in order, and tells you which one a problem lives in. This is the same set of questions that decades of visualization research keeps coming back to.
Is it the right chart for the goal?
First, does this chart type actually answer the question being asked. A pie chart asks your eye to compare angles, which it does poorly, so if the goal is to rank five service lines from biggest to smallest, sorted bars do the job and a pie cannot. The most common problem here is a chart that looks fine but cannot support the decision it is meant to inform.
Does it represent the data honestly?
Next, does the chart show what the numbers actually are, or has the data been framed to flatter a story. A time range cropped to start at a convenient low point, lines all rebased to 100 so a shrinking product still slopes upward, percentages that do not add up. The chart type can be perfect and the picture still misleading because of what was left in or out.
Is the encoding distorting?
Last, does the visual mapping match the numbers. On a bar chart the length is the value, so a y-axis that starts at 90 instead of 0 turns a small rise into a towering one. Two unrelated series forced onto two y-axes can be slid into a fake correlation. A rainbow palette can invent boundaries that are not in the data. Right chart, honest data, but the encoding lies.
Quarterly revenue ($M)
Quarterly revenue ($M)
Most charts that get flagged are only wrong in one of these three places. Naming which one is what lets you fix it in one move and explain the fix in one sentence.
Two ways in, and an honesty boundary we will not cross
How much we can fix depends on how much you give us. There are two modes, and the difference between them is the most honest thing about FairCharts.
The corrected chart uses your real numbers
Paste your data table, with or without the chart, and the fixed version is built from your actual values. The chart you get back is one you can drop straight into the deck and defend, because the numbers in it are yours.
The fix shows the corrected design, not invented numbers
Upload only a picture of a chart and FairCharts critiques the design, the chart type, the axis, the colors, the clutter, and shows you the corrected encoding. It will not read exact values off an image and present them as if they were real. So an image-only fix is illustrative. It demonstrates the honest version of the chart without inventing your data.
Most tools would happily guess the numbers off your screenshot and hand you a confident chart built on a guess. We will not, because a chart you cannot trust is the whole problem we exist to solve. When the numbers are real, we say so. When the fix is illustrative, we say that too. That line is the product.
Why you can put your name on the fix
Built on decades of visualization research. The rules FairCharts applies are not house style. They come from a long, established body of work on how people actually read charts, the same principles taught and tested across the field. We do not attach a name or a slogan to it because the authority is the research, not us.
Every fix is tied to a named principle. A flag is never "this looks off." It is "the y-axis starts at 90, which exaggerates the difference, so start bars at zero." You get the reason in plain language, which means you can repeat it in the room and defend the chart yourself.
It stays quiet when a chart is already fine. FairCharts does not invent problems to look busy. If a chart is honest, it tells you so. A tool that flagged everything would be useless, so a clean result is a real result.
What it will and will not do
Review a chart and rank what is misleading
Issues come back ordered by how much they distort the message, each tagged with the level it lives in (the task, the data, or the encoding) and a one-line reason.
Hand back a corrected, presentation-ready chart
Not a sketch. A clean version you can use, with the reasoning attached so you can defend it.
Tell you when there is nothing to fix
Silence is a feature. A chart that passes gets a pass.
Invent numbers from a picture
Image-only mode critiques the design and shows the corrected encoding. It does not read exact values off a screenshot and present them as real. For a fix built on your actual numbers, paste the data.
Make a dishonest chart prettier
This is not a slide-polishing service. The goal is a chart that is correct and clear, not one that is decorated.
Replace your judgment, or your BI tool
FairCharts is not a dashboard, a stats package, or a general design tool. It does one thing, checks a chart for honesty and clarity, and does it well.
FairCharts is in beta and provided free and as-is while we test. See the terms for what that means.
Want to put your own charts through it?
FairCharts is in beta. Leave your email and we will let you know the moment it opens, no spam in between.