A version of this post first appeared in my newsletter. Subscribe to receive posts like this in your inbox every other Sunday.

Clear labels are essential to good information architecture. Think back to a time when you tried a new website or app and found it easy to use. Labeling likely played a key role.

Labels are words, phrases, and icons that represent larger chunks of information in the website or app. You experience them as choices in navigation systems, labels on input fields and buttons, copy headings and subheadings, etc.

These phrases and symbols offer clues about what content might be there and how it might be organized. Besides making things clear for humans, they also help search engines know what’s most important on particular pages.

Labels also play an essential role in creating context. When visiting an unfamiliar website, you can deduce what kind of place it is from the words in its primary nav. Context also changes your perception of what labels mean. For example, “save” means different things on a bank’s website than on a computer application.

Mindful labeling leads to more understandable, useable, and accessible websites, products, and apps. But good labels can be surprisingly hard to write. Here are some suggestions to get you started.

Use clear and concise language

Labels should be understandable to your users. When defining labels for navigation, avoid jargon, technical terms, marketing phrases, and other terms that may be unfamiliar to users. Research, research, research!

Labels should work as a system

Labels should work as a system, not as individual loosely-related parts. For example, menu choices may start with verbs or nouns — but mixing the two might be weird. Also, consider label length: it’d be awkward to have one choice be a three-character word and another be a five-word phrase.

Be consistent

Don’t use different terms to refer to the same concept in different parts of the system. Conversely, don’t use the same term to refer to different concepts. Modeling the whole system will let you spot inconsistencies, and developing a style guide will keep folks aligned.

Consider context

As I mentioned above, labels create contexts — and contexts change how people understand what labels mean. Be mindful of where labels appear within the system; they might take on subtly different meanings simply due to where they’re experienced.

Consider internationalization

Among other challenges, labels will have different lengths in different languages. Work with native speakers to ensure labels can be adapted without losing clarity or meaning.

Consider accessibility

Labels play a crucial role for people using screen readers. Work with your front-end developers to ensure labels create an accessible and inclusive experience.

Icons are labels too

Labels don’t need to be words. Icons are useful when targeting phones and other devices with small screens. But icons require special care: images can be more ambiguous than words; they rely more heavily on context for their meaning than text labels.

Expand your vocabulary

Naturally, the broader your vocabulary, the more effective you’ll be at recommending suitable labels. A good thesaurus is an invaluable tool. Reading a lot also helps. Note new words you come across — but again, beware of using obscure terms in labeling systems.

Test and iterate

As with all other aspects of IA, your labeling system will evolve. Use tree tests to evaluate the usability and understandability of particular labels. Examine site analytics. If you find labeling issues, iterate and test again.

Mind your labels

I often joke that my job entails designing places made of language. Most websites, apps, and other digital product UIs are primarly collections of words and phrases arranged in particular ways to allow you to do or learn specific things.

Labels are the foundation. Mindful labeling makes the difference between a functional yet unusable system and one that users can easily pick up. Mind your labels, and you’ll be on your way toward architecting systems that meet user needs.