The benefits of organising content with tags

Traditionally, organising content into categories has worked well. But for truly massive sites — serious projects that count articles, members or products in the hundreds or thousands — tags provide a better solution to the thorny problem or finding content.

Tags are labels that describe content. Content is the meat of a web site: articles, products, people, or downloadable files.

There are three important differences between tags and categories:

Categories lend themselves to hierarchies. Tags are great at cross-referencing.
Categories are perceived as containers. Tags are seen as appendages.
Categories are vague. Tags are specific.

These differences are not trivial. Using categories, tags or both have tremendous impact on how people use a site—or if they go looking for something better.

Meta data is information about information such as title, author, copyright date, format, genre, model number, duration, expiration, ISBN number and version.

Tags are not quasi-cats

The difference between tags and cats is their relationship with content. In a one-to-many relationship, an item belongs to one group and one group can hold many items. In a many-to-many relationship, an item may belong to more than one group and vice versa. Categories are usually one-to-many. Tags are always many-to-many.

People see categories as fixed buckets. When I worked in news, I created content as a “news article” or a “sports infographic” or an “opinion piece.” The terms were broad: Sports covered football, tennis, swimming, volleyball, golf, team ratings, player profiles, history, the budget for stadium construction and injured horses. “Sports,” all of ’em.

Tags are more specific. People assign or create tags regardless of what they’ve used before. Content can have many tags—even tags of the same type. For example, an article may have several author tags; a product may have several colour tags; a music file may have several genre tags.

Tags tend to fit the content they describe while content fits the categories they fall under. Mostly.

Many web sites that collect massive amounts of information use tags, including Stumble Upon, Delicious and Flickr.

The Dublin Core is a set of independent metadata “tags” invented in 1995. Read about the DC elements.

Uses for tags

The obvious use for tags is to describe the subject(s) of the content. But the potential for tags is incredibly diverse. To name a few:

Why tag when you can add database fields? Some metadata, like titles and creation dates, rarely apply to an item more than once. Granted, a member profile will have one last name and one first name, and these data will only apply to one piece of content. Tags are best used if content might have more than one of the same type of data, and if the data may get reused often.

A recipe will have more than one ingredient, but different recipes may have more ingredients than others. A book may cover one topic or ten. Photos might have tag(s) describing where they were shot, unless they were composed in a studio.

Why tag when you can search? Keyword searches scan for words or phrases in the content itself. On the plus side, you get a comprehensive look of the text. But if you search a wine site for, say, “good merlot,” your results would include an article that says “this is nothing like a good merlot.” And if your content isn’t text based, it can’t be searched at all (until Google invents full audio and video searches).

Tags act like recommended, one-click keyword searches because the author has deliberately described the content.

system

System tags also determine how a site displays content, but for more practical reasons. Content tagged “delete me” could be wiped during daily database sweeps. Tags for “high priority” and “low priority” could cause content to appear at the top (or bottom) of search results.

context

Context tags hides or reveals content depending on the situation. Information tagged “XXX” prevents underage viewers from viewing naughty pictures (if you can prove their age). Content formatted for mobile devices may be tagged “mobile.” Information tagged “rough draft” or “banned” isn’t available to the public. Products tagged “summer sale” only appear during certain months.

implied

Implied information comes from the act of tagging itself. When searching, a system can be told to disregard tags that are only applied once. Administrators may review single-use tags and delete the unnecessary ones.

temporary

Temporary tags evaporate when they’re no longer needed. These tags act as quick selections for acting on certain pieces of content. When organising information in content management systems, content may be tagged as “need to review” until it’s been edited.

point of view

Point of view tags appear only to the right people. For example, site administrators may see tags that mark content to be removed or a note to follow up later. Members of a site may add private tags for their own reference.

spontaneous

Spontaneous tags mark seemingly unrelated information, like different products that work together. Say a tourism site sells tickets, luggage and travel insurance packages. They can mark a certain suitcase, plane ticket and insurance pack with a “Summer Savings” tag to automatically appear together on the sales page.

Otherwise, you can group tags into categories, as it were, such as:

Format as in HTML, PDF, ebook, music, video, paperback
Audience like children, adults, specific political parties, professionals, women, immigrants
Authors for different people
Flavours of food or drink
Location, meaning specific geography or “down the road”
Whatever tags seem to belong together. It varies per project.

Types of types of tags

Extremely sophisticated sets of tags may either use hierarchies of tags allow a tag to belong to more than one type.

Be afraid. This way lies madness.

It’s possible to have tags for “morning,” “afternoon” and “evening.” These might be time of day tags. But you could also have tags for “1970s,” “1980s” and “1990s” — decade tags. When tagging music, the decade it was produced has nothing to do when the time of day you prefer to enjoy it. But both are types of time.

You should only use extreme hierarchies of tags themselves if you have thousands of records to organise. It’s easy to have as many supporting tags as actual content. Not that a multitude of tags is bad—but they require effort to maintain. That’s the problem with such a powerful tool: To avoid making a muddled mess, you have to apply some discipline.

Cross references: Tags’ real strength

When you go to a category, you can expect to see a healthy amount of content organised in a logical manner. As specific labels, a single tag is less useful. Their real strength lies in multitudes.

Looking for content tagged “green” it’s about as meaningful as clicking to the “outdoors” category of a camping gear web site. But look for everything tagged “green,” “backpack” and “men” will yield meaningful results. If the system is sophisticated enough, you can search for “backpack,” “men,” but not “green” if that’s not your colour (and assuming you’re looking for a men’s pack).

Tags are at their best in the masses. As the volume of information grows, well-kept tags provide solid references and sometimes surprising correlations that change as your content changes in the long run.

I’m working on a tag-heavy project with a twist: The most popular feature chooses ten pairs of tags at random. People who tried it loved clicking unexpected combinations just to see what happens. Ask me for a preview.