Knowledge Academy - Deep Dive into Article Optimization in Knowledge Center

[rob_martoncik]

Stop Publishing Broken Knowledge: Deep Dive into Article Optimization in Knowledge Center

📝 Topic

A technical deep dive into Knowledge Center’s Article Optimization features—covering automated scans, AI-powered recommendations, custom rules, and how to keep your knowledge base clean, searchable, and AI-ready.

Deep Dive: Article Optimization in Knowledge Center

Knowledge quality directly impacts search accuracy, AI responses, virtual agent performance, and overall user experience. In this deep dive session, the Platform Knowledge team explored the technical details behind Article Optimization in Knowledge Center—a major step forward in proactive knowledge governance.

Why Article Optimization Matters

Traditionally, knowledge cleanup has been reactive. Articles get published, then problems surface later—broken links, poor titles, inaccessible images, or structural issues.

Article Optimization changes that by introducing automated scans that continuously evaluate content health.

Instead of waiting for issues to surface, Knowledge Center now flags them before they impact users.

How Article Optimization Works

At a high level, the feature is built around three core components: Jobs, Scans and Findings

1. Jobs

Jobs define when optimization runs. These can be:

• Scheduled (weekly, monthly, etc.)
• Run against specific tables
• Configurable per environment

2. Scans

Scans define what the system is looking for. There are two types:

Script-Based Scans (Available to All Customers)

These evaluate deterministic conditions, such as:

• Broken internal and external links
• Missing image alt text
• Improper heading structure (e.g., multiple H1 tags)

These scans improve accessibility, SEO, and technical correctness.

AI-Based Scans (Now Assist Required)

These use LLM-based evaluation to analyze content contextually, including:

• Title relevance compared to article content
• Suggested alt text for images
• More advanced readability and semantic evaluations (future roadmap)

These scans don’t just identify issues—they recommend improvements.

3. Findings: How Issues Surface

When scans detect problems, they create findings.

Findings:

• Are stored as records
• Surface directly in the Advanced Editor
• Group under their corresponding scan
• Can be ignored or resolved

This tight integration inside the editing experience means authors don’t need to leave the article to fix problems.

Optimization in Action

During the demo, the team showed:

• A mistyped article title triggering a title relevance warning
• Multiple images flagged for missing alt text
• Broken links identified automatically
• Immediate feedback appearing after saving

Authors can review each finding and:

• Accept AI recommendations
• Manually correct issues
• Ignore findings intentionally

This makes optimization collaborative—not forced automation.

Custom Scans

One of the most powerful capabilities covered was the ability to create custom optimization scans.

Organizations can define:

• Script-based custom logic
• AI-based prompts and skills (for Now Assist customers)
• Organization-specific compliance rules

This allows teams to enforce brand guidelines, regulatory requirements, or internal writing standards at scale.

Preparing Knowledge for AI

Article Optimization is not just about cleanliness—it’s about AI readiness.

Search, Now Assist, and Virtual Agent all depend on:

• Clear structure
• Relevant titles
• Accessible media
• Valid links
• Clean semantic markup

By proactively improving these areas, Knowledge Center ensures knowledge performs better across the platform.

Final Thoughts

Article Optimization represents a shift from reactive knowledge maintenance to proactive governance.

Instead of periodic audits, Knowledge Managers now have continuous quality monitoring built directly into their workflow.

If you’re investing in AI on the platform, optimization isn’t optional—it’s foundational.

⏱️ Chapters

(Chapters structured for a deep dive format — adjust timestamps if needed to match final runtime.)

0:00 – Welcome & Session Overview
3:10 – Why Article Optimization Matters
6:45 – Architecture Overview: Jobs, Scans & Findings
11:30 – Script-Based Optimization Scans Explained
16:20 – Broken Link Detection Demo
21:10 – Missing Alt Text & Accessibility Improvements
25:40 – Heading Structure & SEO Validation
30:15 – AI-Based Title Relevance Scan (Now Assist)
35:30 – How Findings Surface in the Advanced Editor
40:00 – On-Change vs Scheduled Scan Execution
44:20 – Ignoring vs Resolving Findings
48:50 – Creating Custom Optimization Scans
55:10 – Preparing Knowledge for AI & Search
59:30 – Q&A and Closing Thoughts

Additional Resources

{Coming Soon}

Presentation PDF

Q/A Responses PDF

Search Tags

ServiceNow, Knowledge Center, Article Optimization, Knowledge Management, Now Assist, ServiceNow AI, Knowledge Governance, Advanced Editor, Broken Link Detection, SEO Optimization, Accessibility Compliance, Knowledge Base Health, Platform Academy, AI Powered Knowledge, ServiceNow Developers, Enterprise Knowledge, Virtual Agent Optimization, AI Search, Servicenow Knowledge, Service Now Knowledge, Knowlege Center (misspelling), ServceNow (misspelling), Article Optimisation (UK spelling)

[Deepanjan ghosh]

Dear Rob,

Thank you for the details.

could you please share more details as documents or presentation to understand what are features are available as OOTB to use and also if required which all can be configured based on the business prospective.

Thanks and Regards,

Deepanjan Ghosh

[maxmueller]

Does Article Optimization considers content in Knowledge Blocks?

[Connie Hernande]

 We are are trying to implement the article optimization to our process, however, we test the OOB, but they are not running properly, either all the card suggestions for each of the OOB shows up even tho, is not a issue on the KB,  or there is an "Error Retrieving Article",  with different templates it only recognizes 1 OOB scans, any suggestion on what should I'd be looking at or how to make it work properly?

[MaryG]

Hi @Connie Hernande 

A couple of things I would check before changing any of the OOB scans.

First, confirm which Article Optimization scans you expect to run. Article Optimization includes both script-based and AI-based scans, and the available behavior can depend on the applicable AI/ServiceNow Otto configuration and entitlement.

I would then review the OOB Article Optimization jobs and their associated scans under Knowledge > Knowledge Center > Lists > Article Optimization > Jobs. Check the query conditions, associated scans, and whether any of the OOB records have been modified.

Since you are seeing different behavior depending on the article template, I would also compare the underlying fields used by those templates. If a scan expects content from a particular field and that template stores or presents the content differently, that could explain why only one scan is producing a result.

The two symptoms you mentioned are worth separating:

• If a suggestion card appears when the condition clearly does not exist in the article, I would look at the individual scan configuration/script and determine exactly what condition it is evaluating.

• If you receive “Error Retrieving Article,” I would treat that as a technical issue rather than an optimization result and check the system logs/browser console for the associated error.

I would reproduce this first with an OOB article template and an OOB Article Optimization job/scans before customizing anything. If the same behavior occurs with completely OOB configuration, I would consider opening a ServiceNow Support case with the specific template, scan, and article examples.