The Case for Better Metadata
Yesterday’s article touched on user-generated content and connects to a broader conversation across the industry right now: how will AI tools find, categorise, and recommend the right learning content when most of our digital assets have little to no useful metadata attached?
Metadata itself isn’t new or niche; it’s a foundational part of how the internet organises and retrieves information, and it has been for decades. But most platforms I look at, and most teams I speak with, simply haven’t filled out their metadata fields because it hasn’t felt important or relevant. L&D is a function that never has enough time; we’re always playing catch-up, and when it comes to loading something onto the LMS or the intranet page, we tend to do the minimum to get it working for our end-users and then move on to the next task, which often means skimping on metadata.
That approach is becoming harder to sustain because AI-powered search, recommendation engines, and skills-based platforms all rely on metadata to function effectively; if we want these tools to surface the right content for the right person at the right moment, we need to do the groundwork now.
Most authoring tools, including Articulate Storyline and Intellum Evolve, do capture some metadata at the point of creation, and you’ll typically find fields for title, description, keywords, author, language, and version number; some tools also allow for learning objective tags, duration estimates, and target audience descriptors. LMS and LXP platforms often add their own layers, such as completion status, assigned audiences, due dates, and compliance categories, but these tend to serve administrative functions rather than semantic ones; they tell the system who should see the content and when, but they don’t explain what the content is really about, what skills it develops, what job roles it supports, or what problems it helps people solve.
The honest reality is that most of these fields sit empty, or they’re filled with generic placeholders that provide no meaningful differentiation; we’ve treated metadata as an afterthought, something to tick off before publishing rather than an investment in future discoverability.
I won’t pretend that the current generation of LMS and LXP tools are making full use of this metadata; many aren’t, but the direction of travel is clear, and the organisations that prepare their content libraries now will be far better positioned when AI-driven learning experiences become the norm rather than the exception. That means building habits around consistent, thoughtful metadata entry, including clear and specific descriptions, standardised skill and competency tags aligned to your organisation’s frameworks, accurate duration and difficulty indicators, and explicit connections to roles, tasks, and business outcomes; it also means auditing existing content libraries to fill in the gaps before adopting new technology, rather than discovering the problem after implementation.
If you’re responsible for a training content library, now is a reasonable time to start asking: what metadata do we capture, what do we leave blank, and what would we need to add if a smarter system came knocking tomorrow?