Tags vs Folders

Many of the tools that I use everyday are based on tags, or integrate tags in some ways. In the upcoming posts, I plan to review and describe how these tools are part of my workflow, so it is a good idea to explain why a tag-based system is superior to a classical tree structure.

The goal of an organization system, be it based on a hierarchical structure of folders, or on a flat structure of tags, is to help us finding specific objects in our information system, quickly and effectively. A beautiful organization system must not become an end in itself. It is a mean to access our information.

An example

The best way to prove that tags are superior to folders is probably an example. Let's suppose that we work with different Companies, some of which are Clients, while some are Partners (for simplicity, let's assume that a client cannot be also a partner). With Clients, we can have documents relative to three activities: Pre-sales, Training, and Support. With Partners, we can only have documents related to Pre-sales and Training.


No matter how we organize our folder structure, some duplication is unavoidable. In this first example we start from the company type: Client or Partner, and we drill down to the activities, which are replicated for each of the four companies:

[caption id="" align="aligncenter" width="374" caption="Click on the image to view it full size"][/caption]

We can organize things differently, starting from the activities, but in this case we must replicate the companies within each activity:

[caption id="" align="aligncenter" width="380" caption="Click on the image to view it full size"][/caption]

Things can get worse in case a document can be related to multiple companies, or multiple activities: this obliges us to replicate the document as well, not only the folders. We can try to find other creative solutions, like using links to the single instance of a document, but there is no clean way to manage this case with a hierarchical structure.

In most common cases, a tree structure brings information replication.

If wrongly categorized, finding an object in a tree structure is more difficult than in a flat structure.


Using tags instead of folders, no replication is necessary (neither of folders, nor of documents and other objects), as there are no hierarchical relationships among our attributes (company type, company, and activity). We can freely tag a document with whatever attribute we want to give it:

[caption id="" align="aligncenter" width="380" caption="Click on the image to view it full size"][/caption]

Using tags is not so easy at the beginning, and requires some discipline; the biggest risk is to end up with hundreds of tags, many of which are duplicates of each other, with a slightly different spelling (i.e. travel vs traveling, partner vs partners, work vs Work). In this situation, tags are worthless, as they do not help to find the information.

Coming from a long experience with a classical hierarchical structures, it took me a long time to be able to use effectively to a tag-based system. Looking back, I do not regret having switched and I now use tags to organize everything: email messages, documents, photos, music, bookmarks, and whatever needs to be organized in some way.

Some tools allow us to define relationships among tags. While this is not necessary, it can help to tag objects more quickly. For example, if we can define that Singapore is part of Asia, whenever we use the tag "Singapore", "Asia" will be automatically added.


  • If you are starting with tags, choose flexible tools, which allow you to rename, delete, and re-organize tags. You are going to change your mind many times...
  • Set some rules to create your tags: for example, singular nouns vs plural, all small-capitals, etc. This helps to avoid duplicates;
  • Periodically review your tags and to some clean-up when necessary: if you realize that you have attributed a tag to a single document, you might decide to remove it or merge it with another tag;
  • Never forget the goal: tags (or folders) are there to help us finding specific objects quickly. Modern systems allow to index the information and to perform search operations effectively. With this in mind, it is not necessary to give too specific attributes to your objects: all the words in a document will be indexed and will be valid search terms. The tags should indicate conceptual categories (i.e. a legal document will probably not contain the word "legal", but this word would be an appropriate tag for the document).