More meaningful than font, placement and format of printed documents, metadata in the digital world delivers opportunity that was unknown in the 1990s. Shifting from the print world to the internet, Creekmore learned the potential value of metadata. But effective use requires clarifying goals, applying metadata strategically and purposefully, and standardizing metadata practices to manage document classification. Common pitfalls among those who appreciate what metadata can offer are designing content management systems with too many or too few metadata options. To avoid excessive or insufficient metadata, information architects must fully understand users’ needs, differentiating daily requirements from occasional ones. Building a simple metadata schema for current use is ideal; focusing on unlimited future possibilities or building an elaborate but unwieldy system is unrealistic.


information architecture
information retrieval systems
document schemas
database models

IA Column

The Curse of Metadata

by Laura Creekmore

I first learned the term metadata in the late 1990s. I’d been doing editorial work for several years at that point, and my work focused more and more on the internet. So I quickly moved from just writing and editing to learning about how the context of my work changed my audience’s perceptions. When I wrote for a printed magazine, I could control almost everything about the experience. When my work moved to a digital forum, I began to realize I didn’t control much at all – at least not the things I was used to controlling, like font size, placement, anything to do with look and feel.

But! The magnificent opportunities that the digital world provided quickly made up for an inability to create magazine-quality design. And so I dived into the world of information architecture, and I began to realize that the information about my information – my metadata – was one of the most powerful tools of all.

I clearly remember struggling to understand what metadata was, even though my introduction to the concept was nearly 20 years ago now. The definition is almost nonsensical. Data…about data? Just a few years later, people began to throw around the expression, “That’s so meta!” which by then made innate sense to me – and meant nothing to most people.

While metadata has seeped into the public lexicon in a deeper way than I would have expected, I would not yet say that this concept is widely understood or appreciated. I’d be willing to bet that most people in the public who’ve heard the term metadata only have a vague sense that it’s associated with the U.S. political controversy over government collection of phone records.

But for those of us with a professional interest in the term – metadata love can go very deep indeed.

Perhaps 15 years ago, I helped design my first content management system with significant metadata capabilities. And I made a classic rookie mistake. That CMS was tricked out, with metadata running in one ear and out the other. We could collect data on practically anything related to our content, so we did. A few months in, I realized we had a couple of big problems:

  • Our whole team didn’t classify documents in the same way.
  • We were spending a lot of time on metadata, without using much of it.

So we backtracked, and we clarified what we really needed to capture. We abandoned some of our classification options wholesale and standardized our use of other terms. We got very tactical about when, how and why to use metadata to organize our content. And the system was very effective for several years.

I’ve designed several content management systems since that time, and with each one, I spend as much time on the metadata as anything else. Metadata powers the modern internet, and spending time to get its design right is critical.

Over time, I’ve done a lot of reading and learning about metadata. I’ve always liked this saying by computer historian Jason Scott: Metadata is a love note to the future. Noted content strategist Rachel Lovinger has done a lot of work to popularize this comment, and I largely agree with it. (See Lovinger’s presentation on metadata from Confab 2015 here: www.slideshare.net/rlovinger/metadata-is-a-love-note-to-the-future.)

But I would also argue that we who love metadata the most can make the worst hash of it.

The problem revolves around the issues I identified in my own CMS years ago: If we can collect data, we think we should. It’s all too easy to design a system that can collect everything we know. We might even start out with good intentions, correctly classifying everything for the first week or month. But in the end, no one has the time or budget to run her organization that way. So we end up with a system that doesn’t search effectively – a frustrating user experience with too many search options and not enough results, or search results that can’t be properly categorized and sorted. Ugh.

I’ve also been in the opposite position, which isn’t great, either – realizing after the system has been operating for a while that you need to collect something that you haven’t been, and you have to sort back through thousands of old records to re-categorize.

In all these scenarios, I’ve fallen back on the user experience. What does my audience need to do their work? What do my internal users need to know to do theirs? Which of those needs are daily needs, and which happen only once a year?

Asking those kinds of questions can help right-size your metadata work.

Part of our challenge comes with our future-oriented mindset, I think. “I might need this one day. Let me just categorize it right now and I’ll always be able to find it later,” the thinking goes. When we focus instead on the user experience, we can be more concrete about what we want to do with the information right now – and reach a more realistic metadata schema as a result.

Next time you have the opportunity to work on metadata design, think about how your audience will be able to use the information you’re collecting. If your metadata will only make sense in a future project phase, hold off. If your metadata will tip the scales toward an overly complex and frustrating user experience, re-think. If your metadata will be difficult for administrative users to manage without frequent team meetings and lengthy training, you may be heading in the wrong direction.

Design your metadata with the future in mind – and with your understanding of today’s reality, for you, your administrative users and your audience.

Laura Creekmore is the Bulletin’s associate editor for information architecture. She and her company, Creek Content, develop content strategy and information architecture for companies with complex communication needs. She can be reached at laura<at>creekcontent.com.