Resources / Writing / The logo goes last

The logo goes last

Nike paid $35 for the swoosh. Here’s how to make your mark, and how to judge it.

Last one in the brand parts: the logo

It’s last on purpose. Everything else gets built together, but the logo waits until type, color, the surrounding assets, and motion are figured out. By then you know what it has to sit next to and what it has to make people feel, and you can judge it against something other than whether it looked good.

An easy way to think of the logo is as the cherry on top to an already cohesive ice cream sundae. The flavors and toppings are jibing, now you need something to signal its completion to the world.

It also helps to remember where a logo’s value comes from. Nike paid Carolyn Davidson $35 for the swoosh in 1971. Everything Nike did afterward is what made it worth anything. Consistent use and effective brand marketing turned the symbol into a global titan. A logo is a tool, not an art piece.


Judge the logo on SAD

Simple, Appropriate, Distinct. The acronym is easy. The catch is that each one only means something against a specific context, and skipping the context is how a logo review turns into a subjective hellscape started by “do I like it?”

  • Simple, judged against every place it lives. Does it hold its integrity at small and large sizes, or do you have a system in place that helps it adapt to different scales? Favicon, social avatar, app icon, reversed on dark, one color, embroidered on a hat, and more. If it is too complicated and uneasy to work with, you need to refine it. It's entirely possible that you have a family of slightly adjusted logos for different sizes, here is an example of a mock brand icon where the gaps between flame licks get bigger as the mark get's smaller.

  • Appropriate, judged against your attributes. Does it align with your creative direction? A bouncy mark is wrong for a brand that needs to feel like a vault, and a severe one is wrong for a kids’ product. If everything else you built is subtle and refined, the logo can’t show up expressive and ready to party til 2am. The logo should look like it belongs with the rest of your creative assets.

  • Distinct, judged against your competitors. The most important of the three. Your logo should never look like any of your competitors’. Identification is useless if your logo can easily be mistaken for a direct competitor using the same concept. Usually, this is the result of choosing a logo too close to the industry (eg. text bubbles for an SMS platform, people for a CRM, or a literal representation of something vertical specific).


How to make it

  • Sketch wide. Explore all three families before you pick one: a wordmark (the letters doing the work), a pictorial mark (a stylized version of a real thing), and an abstract mark (a shape that carries the idea without depicting it).

  • Find the winning concept. The strongest marks take one piece of the brand’s story and pare it down to a single form. DevStats came from combining a compass with a pie chart.

  • Resolve it in black and white. If it doesn’t work in one color, color won’t save it. Color comes after the form is right.

  • Construct it properly. Clean geometry, corrected by eye. Type that’s hand-tracked instead of typed out. A mark and wordmark balanced so neither one is shouting over the other.

  • Build the family. A primary lockup, the standalone mark, a favicon, a reversed version, a one-color version, clear space, and a minimum size.

A note on AI logo concepts: I have tried using AI for the sketching phase over and over again, and I've found that the AI tools are not good at it, at least if you want to get a unique and interesting mark. So, for the exploratory conceptual phase, let go of the AI tools. Take a pencil and a piece of paper or an iPad and just let your mind wander. No idea gets left off the table.

You'll come up with something far more interesting than the robot would. I guarantee you.

Tools:

  • A sketchbook and a pen or a tablet for the first fifty (yes, fifty) ideas
  • Flora (or a comparable tool) to run Quiver's Arrow for the first vector iterations.
  • Illustrator or Figma to construct the final vector
  • Claude Code for packaging every logo format (it’s really good at this!)

Shrink your logo down to favicon size and turn it black and white. If it doesn’t hold up, it fails. Good logos are simple.

If it doesn’t mesh with the rest of your branding, it fails. Good logos are appropriate.

Put it next to your three closest competitors’. If you can’t tell whose is whose, it fails. Good logos are distinct.

Every good logo is… SAD.

— Zach