Why content structure and governance are still innovative
In all my years as a consultant, I’ve never seen technology solve a people or process problem.
Using AI is the new pink. (“Whoever said orange is the new pink is seriously mistaken.” - Elle Woods, Legally Blonde) AI developments are fast and furious: Every week there’s a new model, a new announcement, a new lawsuit, a new personality conflict. We hear about companies requiring employees to use AI tools for coding, content creation, and research analysis. We are pushed to use AI to create personalized and accurate content for others.
And yet… Content chaos persists. The underlying content pushed into an AI tool or LLM is out of date and contradictory. There are gaps in the information. Different teams are creating the same content without knowing it.
There are so many ways to improve efficiency and effectiveness without using a new technology. Indeed, to use AI effectively, a tool that ingests content and spits it out in another way, we have to improve our content quality.
This isn’t a technology problem, and AI isn’t going to solve it. This is a structure and governance problem that necessarily involves people.
Content structure and content governance are the real innovation. Companies that align on their content structure and governance will be the real long-term winners. It’s not a shiny new technology, but the unglamorous discipline of laying down the content structure and governance which will give you long term gains for using AI, personalization, and omnichannel content success.
Can’t solve people problems
People cause problems with content and information for all kinds of reasons that can’t be fixed with technology:
They don’t like the rule, so they circumvent it
They weren’t consulted, and the rule doesn’t work for them, so they do something else
They don’t know the rule so they don’t know to follow it
They know there’s a rule but haven’t training on how to execute it properly
They dislike the person who is enforcing the rule, so they circumvent it
The rule wasn’t implemented properly, so they can’t actually follow it
By “rule” in this sense, I mean a guideline, a process, a standard, or something that’s meant to support good process and content.
How do you work with a person who is causing issues for any of the above reasons? One answer is to “restructure” their position, but you can’t do this with everyone. If you do it, you’ll lose valuable knowledge. Many companies don’t address these issues and kick the can down the road.
Companies that do effectively address people problems are the real innovators and create real value for the organization. Potential benefits include: reduced churn, everyone working toward the same goal, and bringing a product to market faster.
Can’t solve process problems
In my work, I find that many content processes are not well defined and documented. Questions abound:
When do you create new content?
How do you take content through a review process?
How do you interview SMEs and get their input?
What kinds of revisions are taking and by whom?
How do you keep content up to date?
How is the owner tracked and held accountable?
What is the process for different types of content?
Are the processes the same across teams?
Many times, we want technology to improve the process or replace part of the process, but often we don’t know what the process is. If you don’t know what the process is and where the problems occur, then any changes are a shot in the dark, throwing darts at a very, very small dart board.
When you can illustrate a process and identify the parts that don’t work, you can adjust those things and improve the process. That creates more efficiency and more value.
Can’t solve content quality problems
Not surprisingly, using AI can’t solve a content quality problem. As I said above, an LLM will ingest whatever content or information you give it. If you give it out of date information, it will answer you with out of date information. On top of that, it can give you hallucinations. You might experience that its lack of memory affects its ability to reason for a longer duration. (This last one really depends on what model you’re using.)
You won’t be able to prevent hallucinations and contradictions, but you can do everything to ensure that the information ingested is accurate, up to date, and complete. This way you know that it’s the LLM that is hallucinating, not the information you’ve given it. It’s faster to troubleshoot AI or LLM issues when you’ve narrowed down the potential causes of the problem.
Can’t solve content structure problems
In a recent meeting about AI, someone asked, “Once we start adding content from across the organization, how will we deal with the same words meaning different things? We know that the way we use some terms is not the same as the way other teams use the same term.”
AI won’t be able to parse out these inconsistencies and they will only create confusion and inaccurate answers. That’s why domain models aka ontologies aka knowledge graphs are becoming so foundational to LLMs.
Essentially, a knowledge graph is a way for you to tell AI how all the things in your area of expertise are related. This means that the AI doesn’t have to guess at the relationships - you’re explicitly defining them.
Creating a knowledge graph is a big lift, and it’s understandable that a company would dive into AI and LLMs and chatbots first. AI is shiny and new and more fun! But without a knowledge graph, an LLM and your chatbot won’t scale, so it’s important to do both at the same time.
When you have a large body of content, your content also needs to use metadata and taxonomy to tell AI what that content is about. Content needs to have good on page design: good titles, descriptions, intro paragraphs, and good headings.
It’s a lot of work
It is a lot of work to effectively implement AI. Other organizations are not doing this work. The foundation to effective AI implementation is content structure and content governance. The sooner your company sets the foundation, the more value you’ll create from your AI work. The sooner you start, the sooner you’ll have the advantage.
Key Pointe works with organizations to untangle content problems like these with stakeholder engagement, user research, content audits, information architecture, taxonomy, and content migration planning. A short conversation is often enough to identify a practical path forward. Please reach out if you would like some guidance.