This post continues the BRIDGE framework, a series on how organizations preserve knowledge continuity when key team members retire, change roles, or move on. Earlier posts in this series established that knowledge must first be surfaced before it can be documented. Documentation alone, however, is not enough.
Drawing on insights from interviews, I explore how these professionals worked to ensure their knowledge did not disappear. In each case, detailed documentation played a key role. The deeper question is whether more could have been done at the level of information architecture underpinning that documentation. What can we learn from their experiences?
At this stage of BRIDGE, document accessibility becomes the central concern. Accessibility is not simply about convenience; it is about ensuring that knowledge can move from one professional to another without losing meaning along the way.
Note: When I refer to information architecture (IA), I mean the structures that make knowledge findable—including taxonomies, classification systems, and metadata.
In a World of AI, is Metadata Still Relevant?
Before turning to the interviews for practical examples, it is worth reinforcing why, despite developments in AI, metadata remains essential.
Metadata as Grammar
- Context: Metadata provides context and structure. It can be understood as part of the grammar of a file, enriching content with meaning that may not be obvious in the text itself.
- Search: Functionality is better than ever. However, metadata is necessary for authoritative identification of concepts, both for human users and AI systems.
- Future-proofing: If organizational knowledge is intended to inform AI tools, clear and consistent information architecture is key. AI depends on structured context to interpret information reliably.
Getting the Metadata Balance Right
Can you have too much metadata? The amount of metadata should be determined by the problem being solved. For example, a law firm may only need enough metadata on matters to identify sector, work type, jurisdiction, and outcome, together with brief notes explaining why a particular strategy or approach was successful.
- Simplicity: Keep structures simple. Taxonomies and keyword sets should support users rather than burden them.
- Balance: Requiring excessive tagging or overly complex classification discourages adoption and reduces consistency.
- Filtering: Metadata also supports tasks that AI may not handle reliably on its own, such as filtering content by practice area, client, date, or project type.
A classification scheme does more than organize content; it explains relationships, signals relevance, and preserves the logic behind how resources are understood within an organization.
Prioritize Critical Content
As noted above, not all content requires the same level of structure. Rather than attempting to tag everything, begin with critical content—the information accessed frequently or directly tied to business or value-forward outcomes.
Information hygiene is equally important. Content that is outdated, inaccurate, or obsolete undermines trust and should not remain in active knowledge repositories simply because it exists. Garbage in, garbage out.
Information architecture is fundamental to effective information organization and retrieval. For library and information professionals, the challenge is often helping the wider organization recognize the importance of IA—not only for knowledge continuity, but for legal tech and AI initiatives more broadly.
Structure as Embedded Experience
The interviews revealed a consistent theme: knowledge remains trustworthy only through application, not storage. There is always a danger of being out of sight, out of mind. Documentation retains authority only when it forms part of a living process in which people continue to work with, question, and adapt over time.
One of my interviewees, Anne C., describes this in practical terms as preserving context alongside content. For her, accessibility is not achieved through formal documentation alone, but through continuous sharing and small acts of recording that prevent knowledge from becoming siloed. Even a brief note explaining how or why something is done ensures that knowledge remains visible and usable beyond the individual performing the task.
Repositories may contain the same content, but newer professionals lack what Anne describes as the “indwelling experience” that allows classification and research decisions to be made quickly and confidently. Structure becomes the mechanism through which experience can be externalized and shared.
Structure as Institutional Memory
Over time, experienced professionals develop ways of organizing and finding information that is not written down anywhere.
Julie’s long career showed that this knowledge—how materials are classified, grouped, and mentally navigated—is part of how work actually gets done.
I identify this as something worth deliberately capturing. The risk is that when an experienced librarian leaves, their documentation remains. Still, the understanding of how things fit together, where to look first, or why things were organized in a particular way disappears with them.
Her expertise was not defined by procedures alone, but by accumulated decisions, shortcuts, and adaptations developed in response to practical constraints. These form a personal “mental taxonomy” of research practice: an understanding of when to follow formal structures and when to adapt them.
The implication is clear. Knowledge continuity requires capturing not only outputs, but the organizing logic that makes those outputs usable.
Structure as Efficiency
Penny’s perspective places classification within a longer technological history. She recalls early cataloguing processes that required dial-up connections to the British Library to download bibliographic records—slow by modern standards, but still an efficiency gain over manual cataloguing.
The objective even then was not storage, but control and discoverability.
Her later work in developing library software reflects the same principle at scale: systems should reduce cognitive effort by embedding structure into the tools professionals use every day. This philosophy aligns closely with Dr. S.R. Ranganathan’s Fourth Law of Library Science, “Save the time of the reader.”
Classification should reduce the effort required to find information. At the same time, it must retain the reasoning that makes that information meaningful and trustworthy.
Structure and Content Lead to Durable Knowledge
The experiences described by Anne C., Julie, and Penny point toward a shared conclusion. Knowledge continuity does not fail because organizations lack information. It fails when the reasoning that gives information meaning is not preserved alongside it.
The growing role of AI makes this distinction more significant rather than less. Search and automation can improve discovery, but they cannot compensate for missing context or unclear structure. Without well-organized knowledge, AI accelerates noise as easily as insight. With strong information architecture in place, it extends existing professional practice by helping knowledge surface at the point of need without losing meaning.
Within the BRIDGE framework, documenting accessibly marks the point at which captured knowledge becomes durable knowledge. Structure and context allow knowledge to remain interpretable by people who were not present when it was created, reducing the need to rediscover or recreate what already exists.
The next stage of BRIDGE moves beyond accessibility to consider ownership: how responsibility for knowledge ensures that what has been made accessible remains current, maintained, and responsive to change rather than gradually drifting out of relevance.
0 Comments