Visual form makes claims faster than labels do. Same size reads as same importance, side by side reads as alternatives, an arrow reads as cause. Review figures for what their arrangement implies, with someone who does not already know the answer.
The pattern
Every element of the figure is correct. It has been checked by three scientists. And the audience comes away believing something the company did not intend.
They think the second program is at the same stage as the first, because the pipeline bars are the same length. They think two mechanisms are alternatives, because the layout put them side by side. They think a step is optional, because it is drawn in a lighter gray that was chosen for contrast, not for meaning.
None of those readings is stated anywhere in the figure. All of them are supported by it.
Why it breaks
Visual form makes claims independently of labels, and it makes them faster.
Same size reads as same importance. Side by side reads as comparison or as alternatives. Above reads as before. An arrow reads as causation. A shared axis reads as a shared scale. A color used twice reads as one category.
Those readings happen in the first second, before any label is read, and they are not overridden by a caption. Vision research treats features like size, length and color as preattentive, meaning they are picked up in about a quarter of a second, before anything is read. If the visual claim and the written claim disagree, the visual one wins.
Scientific review does not catch this because it reads the figure the way a specialist reads: element by element, checking each against what they know. A specialist supplies the missing context automatically and never notices they did. The investor does not have that context and reads the shape instead.
What changes
Review figures for what they imply, not only for what they state, and do it with someone who does not already know the answer.
Practically, that means a second review pass with three questions. What does this figure claim by its arrangement? Which of those claims did we intend? Where the form is saying something we did not choose, is that a labeling fix or a structural one?
Most of the time it is structural. Two things that must not be read as alternatives should not share a row. A pipeline where stage is meaningful needs a stage axis, not equal bars. A step that is required should not be the one drawn faintest.
The practical result
Fewer questions that are really corrections. The recurring ones — “so are these two competing?”, “is the second one further along than I thought?” — come from the figure, not from the science, and they stop when the figure stops implying them.
It also removes a specific and expensive risk: a figure that overstates. A pipeline graphic that makes a preclinical program look clinical is not a design problem. It is a claim, made visually, that the company did not intend to make and would not put in writing.
Accuracy is the floor. What the arrangement asserts is the message.
