AI SEARCH (search analytics dashboards) search events

[Moliza]

When it comes to our metrics for our dashboards, our numbers always fluctuate. We always stay between 2.75 and 3.00 never lower. To help AI SEARCH, we configured the ai suggestions, allocated the right key words to significant articles, and also boosted articles. what else should I take a look at?

[Isaac Vicentini]

Hi @Moliza,

If my answer was helpful to you, accept it as a solution. Thanks!

Here is the link to Exclude knowledge block content from the AI Search index documentation:

https://www.servicenow.com/docs/bundle/yokohama-platform-administration/page/administer/ai-search/ta...

[Isaac Vicentini]

Hi @Moliza,

If my answer clarified your doubts, please mark it as the solution to help others as well.

[Moliza]

Hi. Issac 

under the boost action, when change the when function to short description, it reverts back to active when i save the page. why is that?

[Isaac Vicentini]

Hi @Moliza,

Does this happen regardless of the value you enter in the "Matches the Searcher's" field?

I tested it on my instance, and this behavior doesn't occur. If it continues to occur for you, I suggest opening a support ticket.

[JessicaLanR]

Hi there,

It sounds like you're already doing a solid job optimizing AI Search by configuring AI Suggestions, applying relevant keywords, and boosting key articles. If your metrics are consistently fluctuating between 2.75 and 3.00, here are a few additional areas you might want to explore to improve precision and relevance:

1. Review Search Logs: Dive into the Search Analytics module to understand what users are searching for and where drop-offs happen. Pay attention to zero-result searches and frequent refinements — these indicate gaps in either content or findability.

2. Metadata Quality: Beyond keywords, ensure your articles have strong, meaningful titles and summaries. AI Search weighs these heavily, so well-structured metadata helps a lot.

3. Content Coverage: Are there topics users frequently search that aren't yet covered by knowledge articles? You might consider mining your incident data for patterns and creating content around common pain points.

4. Training AI Models: Make sure you're leveraging the AI Search Relevance Feedback feature. If users are clicking certain articles consistently, that feedback should be used to retrain your relevance model — but it needs to be enabled and configured correctly.

5. Article Freshness: AI Search can give preference to more recently updated content. If you have excellent articles that haven't been touched in a while, consider updating them with a fresh timestamp and minor refinements.

6. Advanced Boost Rules: If you haven’t already, explore conditional boosting based on user roles, departments, or search terms. Targeted boosts can yield more relevant results per audience segment.

7. Personalized Search Context: If you're on a higher-tier edition, enabling contextual search by persona or use case can significantly improve AI Search performance.

If after these optimizations you’re still stuck below 3.00, I’d also recommend checking the AI Search documentation for your specific version — some improvements or model updates vary slightly across releases.

Let me know how it goes, and feel free to share your AI Search metrics dashboard configuration for a deeper look.