Streamlining Search Results in a Single Knowledge Base

[Garima26]

Hi All,

In my organization, we maintain a single knowledge base that caters to end users, the service desk, and support teams. Here’s a bit more context:
• The knowledge base includes ITIL-marked SOP articles for support teams and service management articles for processes like change, request, problem, and knowledge.
• When searching for terms such as “change management,” the results often show numerous unrelated articles wherever the term is referenced, making it challenging to find the most relevant information.

While I understand that having separate knowledge bases might help, we prefer to stick with a single knowledge base. Are there any best practices or configurations we can apply—like tagging, categorization, or search filters—to improve the search experience and streamline results?

I’d really appreciate your insights or suggestions.

[Dr Atul G- LNG]

Hi @Garima26 

As you mentioned, having different knowledge bases is beneficial, but to enhance it further, tags and meta keywords are essential. You should add precise tags and meta descriptions to improve search results, and if AI search is enabled, make sure to optimize that as well.

[dpittard]

The meta field on knowledge articles is specifically for external search engines, correct? It does not help when internal users are using Global Search or Knowledge Search or the Service Portal? The ServiceNow documentation for the meta field says "Tags for optimizing search from external search engines", but I want to verify I am thinking correctly. We don't utilize external search engines since our knowledge base is for internal employees only, so they utilize the Service Portal. If I'm understanding this correctly, meta tag would not benefit us, so there's no need to use this field.

[qprinsloo]

No, tte "meta" field is not specific and limited to external search engines only. It is used by AI search on the platform as well as classic Zing search if still used. Initially when AI search was first introduced the "meta" field was not indexed and tags had to be used. That is no longer the case. OOTB, the "meta" field is now indexed. When you look at how this is weighted, the "meta" field doesn't add any value unless you want the article to appear in results for words that are not contained in either the article title or body (which are also indexed). These sources provide a good overview. This AI academy covered the various search eights around the 11 min mark.

Content Creation for Relevancy in AI Search

https://www.servicenow.com/community/intelligence-ml-articles/content-creation-for-relevancy-in-ai-s...

Now Assist in AI Search Xanadu Enhancements and Guidance​ | AI Academy

https://youtu.be/LC01ES1NtsM?si=fe-MY3FlOWEdnYJy&t=660

[paulclement]

I tweaked the weighting on our KBs so that the meta field was more than the short description, and I bumped the short description up too. Now we add search phrases and keywords (comma separated) into the meta and try to make the short description as accurate as possible. I have also done a bit of work on the categorisation of KBs.

I feel that doing this did help to bring more relevant content to the top of searches, otherwise it can be dominated by in article or document keywords.