Here's something that trips up almost everyone who's never had to think about it closely: two fonts set at the exact same size, say 40px, can look like completely different sizes on screen. Not similar. Different enough that one looks like a headline and the other looks like an afterthought.
That's not a bug. It's the part of typography that "just use a bigger font size" completely fails to explain.
Point size is not the same thing as apparent size
The number in your font size field measures a fixed unit of the typeface's design space, not how big the letters actually look. What actually determines how big something reads is the x-height, how tall the lowercase letters are relative to that unit, how wide the letterforms are, and how heavy the strokes are.
A geometric sans with a tall x-height and wide letterforms can look 20 to 30 percent larger than a narrow, low-x-height serif at the identical point size. This is why swapping fonts on a template, without touching a single number, can suddenly make text that used to command attention feel like it shrank.
The practical result: don't trust the number. Trust your eyes, at actual size, on the actual device. If a font looks small at 40px, it might just be a font with a small apparent size, and the fix might be the typeface, not the number.
Hierarchy is a ratio problem, not a size problem
Most advice says "make your hook big." The more useful version is: the jump between your hook and your supporting text is what your eye actually registers, not the absolute size of either one.
A hook at 60px next to supporting text at 55px barely reads as a hierarchy at all, your eye can't tell what to look at first, even though technically "the hook is bigger." A hook at 60px next to supporting text at 28px reads as an obvious, instant hierarchy, because the ratio between them, roughly 2:1 or more, is large enough for your visual system to register a clear difference without consciously measuring it.
As a working reference point, a jump of less than about 1.3x between two text elements tends to read as "basically the same size" to a casual viewer, even if you can see a difference staring at it closely. If you want something to feel unmistakably more important, the gap usually needs to be closer to double, not a modest bump.
Your editor is lying to you about size
Design tools and phone previews inside editing apps are usually rendered at a zoom level and pixel density that doesn't match how the final file actually displays on a real phone, held at a real distance, outdoors, in a moving feed next to five other posts. Text that looks perfectly sized in a full-screen editor preview can shrink to genuinely uncomfortable once it's compressed into an actual 9:16 export and viewed at native size.
This is the single most common reason people are shocked their captions are unreadable after publishing, despite everything looking fine while they were making it. The editor's job is to let you edit. It was never designed to accurately preview real-world reading conditions.
A better test than any size chart
Export the real file. Open it on your own phone, at arm's length, at the brightness you'd actually be scrolling at, next to other content in your actual feed for comparison. Rules of thumb and size charts are a decent starting point, but they can't account for the specific font you picked, the specific screen your audience uses, or the specific context it's competing against. Your own phone, at native resolution, is the only test that can't lie to you.
Where this is heading
Apparent size, hierarchy ratios, and real-device rendering are exactly the kind of judgment calls that are easy to get right once and forget under deadline pressure the next ten times. That's part of what we're building into Varnam, checking real, rendered size and hierarchy against how it will actually appear, not just the numbers you typed in.
If that sounds useful, the waitlist is open.