Amazon's first serious brand overhaul in two decades has been rolling out since spring 2025: a softer smile, a single "Smile Orange", new typefaces, one icon language and a family structure for everything from Prime to One Medical. Most coverage focused on the smile. The more useful part is the structure underneath it.
Every growing company eventually hits the problem Amazon was solving. You launch a second product, then a third, a new market, a premium tier. Each team makes sensible local choices, and three years later the brand looks like five companies that happen to share a name. I think Amazon's answer, sorting sub-brands into families with shared traits, is one of the clearest models a founder can borrow, at any size.
01The problem Amazon actually solved
Start with what the refresh was for, because it was a real problem and Koto, the studio behind it, solved it well.
After twenty years of expansion, Amazon had become, in the words of Koto's founder James Greenfield, "at times lots of different companies". Sub-brands had developed their own looks. Ember, a typeface first drawn for Kindle screens, had spread into places it was never designed for, often mixed with other fonts. Icons were produced by separate teams, which left what Koto called "a patchwork of styles and visual languages". The brief was explicitly not to reposition Amazon. It was to help people "navigate something that was increasingly hard to".
The answer is a system:
- the smile redrawn with softer curves;
- one core color, Smile Orange, with related colors for each family (green for Fresh, turquoise for One Medical, a brighter blue for Prime);
- two new typefaces, Ember Modern for everything and Amazon Logo Sans for the wordmark;
- one icon language across teams and regions;
- sub-brands grouped into "cohorts" with a shared character: entertainment, healthcare, devices, grocery.
For a company that touches people through a shopping app, a streaming service, a doorbell, a pharmacy and a grocery store, that is the right kind of work. It makes the brand easier to recognize, cheaper to run and harder to break. On its own terms, it's a success.
02Families instead of one-offs: the cohort model
The most transferable idea in the whole system is the cohort. Instead of designing each sub-brand on its own, or forcing all of them into the same uniform, Amazon grouped them into families that share a character:
| Cohort | Character | Likely members |
|---|---|---|
| Entertainment | Lively | Prime Video, Music |
| Healthcare | Trustworthy | One Medical, Pharmacy |
| Devices | Modern | Alexa, Echo, Kindle |
| Grocery | Fresh | Amazon Fresh |
Every member of a family inherits the master assets (the smile, the orange, the typeface, the icon style) and then adds the traits of its cohort: a color range, a level of energy in the motion and photography, a tone of voice. Prime got a more festive treatment. Alexa and the devices got a more polished one. House brands like Basics and Essentials were folded into the same structure.
What this solves is the classic multi-product headache. Brand teams usually swing between two bad options: total freedom, which produces the patchwork Amazon had, or total uniformity, which makes a pharmacy feel as playful as a streaming service. Cohorts sit in between. They give each product the personality its customers expect, while keeping the family resemblance that makes people trust the next product you launch.
03Five signs your brand has outgrown its first logo
You don't need fifty sub-brands to have Amazon's problem. It usually shows up much earlier, around the second or third product. These are the symptoms worth watching for:
- Every new launch starts with a design debate. Nobody knows whether the new offer should look like the main brand, a cousin, or something new, so the question is reopened each time.
- Your sales deck, website and product don't look like the same company. Each was made by a different person at a different time, and nobody owns the whole.
- The logo fails somewhere important. An app icon, a favicon, a social avatar, an invoice. If it needs a special version every time, the system is missing a level.
- Customers ask whether two of your products are related. That's the clearest sign the family resemblance has gone.
- Your typeface or colors were chosen for one medium and now live in ten. Amazon's Ember was drawn for Kindle screens and ended up on boxes and billboards. Smaller brands do the same with a font picked for a pitch deck.
If two or more of these sound familiar, a new logo will rarely be enough. What you need is a structure: what every product shares, what each family adds, and what stays free.
04A one-page brand system for a company with several offers
Here is the Amazon structure scaled down to something a team of ten can build in a week. Fill in three levels, and write them on a single page that anyone launching something new can read in five minutes.
| Level | What goes in it | Rule | Example for a studio with three offers |
|---|---|---|---|
| Shared core | Name, logo, main color, typeface, icon style, a few tone-of-voice rules | Never changes from one offer to another | Same wordmark, same blue, same font everywhere |
| Family traits | A color range, image style and energy level per group of offers | Defined once per family, reused for every member | Strategy offers get calm, editorial visuals; training offers get brighter colors and more motion |
| Free zone | Campaign visuals, seasonal ideas, social formats | Anyone can experiment, as long as the two levels above stay intact | A launch campaign can borrow a new illustration style for a quarter |
Three habits make the page useful in practice:
- Decide the family before you design the product page. When a new offer comes up, the first question is which family it joins. If it fits none, that's a strategic conversation worth having before anyone opens Figma.
- Name an owner for the shared core. One person approves changes to the top row. Everything else can be delegated.
- Test the system at its smallest size. If your logo and colors still work as a 32-pixel app icon and a one-line email signature, they'll work on a billboard.
Amazon needed 18 months and 15 markets. Most companies need an afternoon, a whiteboard and the discipline to stick to what they wrote.
05The next product is the real test
A brand system proves itself on the day something new launches. In a few years, Amazon will add services nobody has heard of yet, and if the cohorts work, each one will look like it has always belonged. That's the payoff of the whole exercise: every launch starts with a head start in recognition and trust, instead of from zero.
The same is true at any scale. Your second and third offers will benefit from everything the first one built, as long as you decide early which parts are shared and which parts are free.
