Are you actually using keywords / meta / tags in a structured way?

[Magnus Hovik]

Hey!

I’m a bit confused about the whole metadata / keywords / tags topic in ServiceNow KM.

In different videos and material I see things like “Use meta and tags on knowledge articles for relevancy. AI Search utilizes meta and tags for relevance.”

Sounds good. But what does this look like in practical terms? 

If you have 1,000+ knowledge articles across one or multiple knowledge bases:

• Are you actually adding keywords / meta / tags on articles?

• Is it structured?

• Can everyone add them?

• If yes, doesn’t that just turn into tag chaos?

• If no, who governs it? The knowledge manager? A central team?

• How much operational admin does that create?

To me, the bigger impact seems to come from: Knowledge actually existing, clear ownership, single source of truth, good structure and formatting, strong titles and problem statements, etc...

Of course, if one could it all then sure. But is it worth it over other things? I'd love some feedback and thoughts!

Does (structured) tagging/meta actually matter for AI Search and Now Assist relevance in your experience?

Thanks!

[RMiller19]

I agree. Tool agnosticism is preferred, but this question was about the ServiceNow platform.  That said, many mature KM programs now treat tags primarily as a governance and reporting layer, not the primary driver of search.

For years many knowledge bases used tags like service_desk or password_reset because older search tools relied on exact keyword matching.

Modern enterprise search works very differently. Platforms like ServiceNow AI Search tokenize language and analyze meaning, so service desk, service-desk, and service_desk are usually interpreted the same way.

Because of that, the conversation around tagging is shifting. What matters more today is consistency, controlled vocabularies, and well-structured articles, not punctuation in the tag itself.

A clear title, strong problem statement, and structured troubleshooting steps influence search far more than underscore formatting.

In other words, the best practice isn’t really underscore vs no underscore anymore, it’s clean knowledge architecture and disciplined tagging governance.

[Aditya_hublikar]

Hello @Magnus Hovik ,

This is actually a very valid question — and something many teams face in real-world implementations.

In practical terms, yes, metadata and tagging do help with AI Search and Now Assist relevance. They provide additional context and signals that improve search accuracy, especially in environments with 1,000+ knowledge articles.

However, they work best when properly structured and governed. Without control, tagging can quickly become inconsistent and less effective.

Additionally, there is another useful feature in Knowledge Management — Featured Content. This allows you to associate specific keywords with an article. When users search for those keywords, the article is promoted or becomes more visible in the results. It acts as a way to boost important or high-value articles for specific search terms.

That said, strong content (clear titles, good structure, and proper ownership) still has the biggest impact. Metadata and featured content should be used to improve articles search operation.

keywords and tags are added in many implementations, especially when using AI Search, but they are usually structured and predefined. Not everyone is allowed to add them — typically a Knowledge Manager or central team governs tagging to avoid chaos and keep it manageable.

If this helps you then mark it as helpful and accept as solution.

Regards,

Aditya

[Aditya_hublikar]

Hello @Magnus Hovik ,

 I hope you are doing well. Does my response helps you ?

If my response helps you then mark it as helpful and accept as solution.

Regards,

Aditya

[Andy Guerriero]

We supply our authoring community with guidance for meta and the impact of the meta they apply to knowledge articles. In practice we are not staffed to monitor metadata usage and it's largely self-directed by the knowledge managers for those knowledge bases. 

Each search engine (including AI Search) may use/weight meta differently in its results presentation and that's to some degree configurable in the search profiles. 

More important, LLMs and other tools that may consume your knowledge will use metadata in different ways and it's important to understand this as the number of agents accessing/using your knowledge changes.

[theRadioMic · Accepted solution]

@Magnus Hovik,

Have to say that I'm a bit surprised that nobody has mentioned that as soon as you select the Save button on an article, even while in Draft, everything in the article is indexed. With that said, the first rule of using Meta / keywords is don't add anything that is already in the article. Think of it as unnecessarily over-saturating the database and will slow the search down in the portal and backend. The only content needed when using Meta / keywords are terms, keywords and key phrases that are not already in the article.

Thought I'd give answering your questions a go...

If you have 1,000+ knowledge articles across one or multiple knowledge bases:

Are you actually adding keywords / meta / tags on articles?

Only if there is additional, useful content not already in the article that will help the customer locate the article easier.

Is it structured?

If you're referring to the Meta / keywords, the list should be separated by a comma, while key terms and phrases should have underscores between the related words and be separated by a comma as well.

Can everyone add them?

No. Typically only those team members who have been provisioned the role(s) to create and edit articles.

If yes, doesn’t that just turn into tag chaos?

I would say see above but what really causes Meta / keyword chaos is reiterating the article in the Meta field ( been there, seen it and still didn't get a t-shirt for working to clean it up! ).

If no, who governs it? The knowledge manager? A central team?

You had said "one or multiple Knowledge Bases" and centralization would depend on how the Knowledge structure has been set up and the direction leadership wants to take. Not a fan myself because I feel a team member's experience in their area of familiarity helps build the unified voice the content should be delivered to the oraganization in.

Now governance starts with the Knowledge Admin who oversees it all and works with the Knowledge Base Knowledge Manager(s), each who have been provisioned the roles to create, edit, approve and publish articles into the respective Knowledge Base.

The team members who have been provisioned the role(s) to create and edit articles can be limited to doing so in specific categories in a Knowledge Base, because they might be SMEs in that area, or to work on articles in a specific Knowledge Base regardless of category.

How much operational admin does that create?

Not as much as you think it might, especially when team structure is ideally organized and kept up-to-date through regular SCRUMs, while the workflows are optimally processing the requests. The not-so-fun part is to get leadership to agree and allow you to build the underlying structure and processes properly. ( again... still no t-shirt! )

Wait until you start using Knowledge Blocks and break into creating KCS articles!

Trust the information has been at least somewhat helpful and don't hesitate reaching out if you have any additional questions.

Michael