The Architecture Behind the Image
How Unsplash uses metadata, taxonomy, and navigation to turn user-uploaded photography into findable information.
What Happens to a Photo After You Upload It?
When I uploaded a photo of a mural along the Milwaukee River to Unsplash, I expected the hardest part to be choosing the photo. Instead, I found myself thinking about words.
How would someone else describe this image? What would they search for if they wanted to find something like it? Is “Milwaukee” more useful than “Midwest”? Does “riverwalk” actually help someone find the photo, or am I just describing what I can see?
I was putting on my information architect (IA) hat.
What looks like a simple photo-sharing platform is actually organizing a large collection of user-generated information. The photograph is the starting point, then metadata, tags, categories, navigation, search, and user behavior give that photograph meaning and create pathways to find it.
That made me curious about what happens behind the image. How does Unsplash take an individual photograph from upload to discovery? How much of that organization is controlled by the contributor, and how much is controlled by the platform? And as the collection grows, how does Unsplash maintain enough structure and context for users to trust what they find?
To answer those questions, I followed one photograph through the system and looked at Unsplash through the lens of information architecture.
From a Photo Dump to a Searchable Collection
Unsplash started as a side project in May 2013 as a way to share better stock photos. At the time, most stock photography was either behind a paywall or lacked the artistic quality the founders wanted. Originally hosted through Tumblr with photos distributed through Dropbox links, Unsplash moved to its own custom site by September 2014 with 766 user-submitted photos.
In January 2015, Unsplash introduced search, powered by more than 100,000 tags contributed by community members. I actually remember receiving the call-to-action email asking users to help tag photos, and I happily participated by tagging and voting. Looking back, this was an important shift from simply collecting photographs towards a searchable platform.
Unsplash also historically screened submissions for quality and subjects. A 2018 post from the editorial team noted that reviews could take 24–36 hours or longer, with established contributors more likely to have photos selected for additional discovery.
This early history shows that Unsplash’s information architecture developed through community-generated metadata, editorial review, and structured discovery.
The Upload Is Part of the Architecture
Today, submitting a photo to Unsplash requires relatively little information from the contributor. After creating an account, users can upload a photograph and add its location, up to 20 tags, and a short description. None of these fields are required, but the tagging process is particularly interesting from an information architecture perspective. As users begin typing, Unsplash suggests existing tags rather than allowing contributors to create their own. This gives contributors some control over how their photos are described while helping maintain a consistent vocabulary.
Once a photo is submitted, additional information surrounds the image and becomes part of how it can be organized and discovered. This includes the location, tags, description, posting date, camera information when available, and the contributor’s name, username, and profile. Categories, featured placements, views, and downloads add even more context to the photograph within the larger collection.
What stands out to me is how little information the contributor is required to provide compared with how much information eventually surrounds the photograph. The image starts as a single piece of content, but metadata, contributor information, categorization, and engagement data create additional relationships and pathways for someone else to find and understand it. The upload process isn’t just about adding a photo to Unsplash; it is the first step in organizing that photo for everyone who comes after.
How Do You Describe a Photo Someone Else Needs?
When uploading my own photo of a mural along the Milwaukee River, I put on my IA hat and thought about what keywords would make good tags for discovery. I started with location-based words — Milwaukee, Wisconsin, Midwest — then moved to subject matter: mural, street art, and public art. From there, I thought about the environment: urban, city, river, riverwalk, and bridge. I considered adding observations like grey sky or a light pole, but struggled with relevance. Would this photo actually make sense to show up under those searches? I ended up using 14 of the 20 available tags.
That exercise made me think about the difference between describing an image and organizing an image. Just because something appears in a photograph doesn’t mean it is useful metadata. A light pole may be visible, but if someone searches for “light pole,” is my photo actually a useful result? Tagging requires contributors to think beyond what is visible and consider what information will help someone else find and use the image. In this way, the contributor becomes part of the information architecture.
Unsplash’s taxonomy appears simple: categories, tags, and location. But with up to three categories, one location, and 20 tags per photo, each image can connect to dozens of concepts and search paths. Unsplash’s suggested tags also help create consistency across those contributions, connecting individual descriptions to terms other users are likely to search.
Get Chandra Rudolph’s stories in your inbox
Join Medium for free to get updates from this writer.
Users can also create their own collections, contributing to the platform algorithm. Unsplash provides the larger taxonomy, while users can create smaller taxonomies based on what they want to save, reference, or return to.
Different Paths, Same Information
Unsplash provides several pathways from its homepage: Search, Categories, Trending, and Browse. Search is the direct option for users who already know what they want, while the 12 categories provide a higher-level layer of the taxonomy for guided exploration. Trending and Browse shift toward discovery, allowing users to explore without knowing exactly where to go.
These pathways support different levels of user intent: Search supports a known need, Categories provide guided exploration, and Trending and Browse support discovery. Rather than forcing every user through the same path, Unsplash provides multiple pathways into the same collection of photographs.
The simplicity of the homepage is also important. Limiting the visible categories gives users direction without exposing the deeper taxonomy all at once. For me, this is where taxonomy and navigation come together. The categories and tags create the structure, but navigation determines which parts of that structure users encounter.
Metadata & Trustworthiness
The information surrounding a photograph does more than help users find it; it helps them decide whether it is relevant and trustworthy. Location, tags, descriptions, photographer information, camera data, and engagement all add context and create relationships between the image and the person searching for it.
This connects back to my experience tagging the mural photo. I had to decide whether “riverwalk” or “bridge” were useful search terms rather than simply listing everything visible in the image. When thousands of contributors make similar decisions, those choices become part of the larger information architecture. Inconsistent language can make the taxonomy less useful and search results less relevant.
Unsplash combines this user-generated metadata with editorial review, creating another layer of data governance. Contributors help describe and organize the content, while the platform’s review process helps maintain quality and subject matter standards. My decision not to tag the photo with “light pole,” is a small example of why more information isn’t necessarily better. The goal is to provide information that helps someone understand, find, and use the content.
Metadata therefore helps move a photograph from content to useful information. At scale, maintaining relevance and consistency becomes just as important as collecting it.
What Happens When the Content Changes?
Unsplash has recently expanded beyond photography by adding user-generated illustrations. While this broadens the type of content being organized, it also creates an information architecture challenge: how does naming, categorization, search, and trust adapt when the content changes?
This becomes especially interesting as AI-generated images become more common. For creative teams searching for visual assets, knowing how an image was created could become as important as knowing what it contains. Where it came from, who created it, and how it was produced all become additional pieces of metadata that can help users decide what they want to find. Currently, photographs and illustrations are kept separate, while search filters are limited to relevance, orientation, and free versus premium content. As the platform brings together more types of media, I could see provenance becoming another useful way to filter and navigate the collection.
Unsplash already has some of the structure needed for this through contributor profiles, editorial review, and content organization. These existing systems could evolve to give users more transparency and control over what they discover, including whether content was photographed, illustrated, edited, or generated.
This brings the evolution of Unsplash back to where it started. The original challenge was making high-quality photographs easier to find and use. As the collection grew, the challenge became organizing more content without losing relevance or quality. The next challenge may be organizing an increasingly complex creative ecosystem while giving users enough context to decide what they can trust.
Conclusion: The Architecture Behind the Image
The photograph is only the starting point; the real work happens around it. A contributor adds tags and location. Unsplash places the photo within categories and other collections. Search and navigation create different pathways to find it. Metadata provides context, while editorial review and other governance mechanisms help maintain the quality and relevance of collections.
My mural photo also showed me that information architecture isn’t only something designed by platforms. Contributors participate every time they name, tag, organize, or describe what they upload. A seemingly small decision — whether “light pole” is a useful tag — can become part of the larger system someone else relies on to find information.
This is the most interesting part of Unsplash’s architecture. It takes something inherently visual and subjective and creates enough structure to make it searchable, discoverable, and useful. As the platform expands and the types of content change, that structure will have to evolve as well.
I started with a photograph and a question about what words would help someone else find it. I ended up seeing how much information architecture is occurring behind the image.
Works Cited
Cho, M. (2013, May 29). Coffee vs. beer: Which drink makes you more creative? The Startup. Medium. https://medium.com/swlh/coffee-vs-beer-which-drink-makes-you-more-creative-f7fcb3b786b1
Pixelhaze Mainframe. (2025, August 22). Who created Unsplash and its impact on web design. Pixelhaze Mainframe
Spratt, A. (2018, June 20). Unsplash submissions 101. Unsplash. https://unsplash.com/blog/unsplash-submissions-101/
Unsplash. (n.d.). Our story so far. Unsplash History
Unsplash. (n.d.). Red [Search results]. https://unsplash.com/s/photos/red
Unsplash Team. (2025, January 29). Unsplash now accepts illustrations. Unsplash. Unsplash now accepts illustrations
