Is content decay faster in AI search? My older content is disappearing from citations

Content decay is real for AI, and it does seem faster than traditional SEO in many cases.

Why AI prioritizes freshness:

1. AI systems value recency signals
2. Real-time retrieval systems (Perplexity) pull current content
3. Training data updates periodically, favoring newer content
4. Freshness implies accuracy (especially for evolving topics)

Decay rates by topic type:

Your 2023 content:

Depending on topic, 2023 content is now 2-3 years old. For most topics, that’s past the decay threshold.

The good news:

Updating content resets the freshness clock. You don’t need to recreate everything.

Operations perspective on managing content refresh.

Our refresh prioritization:

The quarterly audit:

1. Export content inventory
2. Add citation tracking data (Am I Cited)
3. Flag content with declining citations
4. Prioritize by business value
5. Schedule refresh sprints

Our refresh workflow:

Week 1: Identify refresh candidates (30 minutes) Week 2: Research updates needed (1-2 hours per piece) Week 3: Execute updates (2-4 hours per piece) Week 4: Publish and monitor

Monthly capacity:

Our team refreshes 10-15 pieces monthly. That’s enough to keep our high-priority content fresh.

Evergreen content perspective.

Some content IS more decay-resistant:

Decays slowly:

• Fundamental concepts (“What is marketing?”)
• Basic how-tos (“How to create a budget”)
• Definitions (“What does ROI mean?”)
• Historical content (intentionally dated)

Decays quickly:

• “Best X of 2023” (obviously dated)
• Industry predictions (become irrelevant)
• Tool comparisons (tools change)
• Trend analysis (trends evolve)

Making content more evergreen:

1. Avoid specific years in titles (unless necessary)
2. Use relative dates (“In recent years” vs “In 2023”)
3. Focus on principles over tactics
4. Update tactical specifics separately

The hybrid approach:

Structure content with:

• Evergreen core (principles that don’t change)
• Updateable sections (current examples, data)

This reduces refresh work while maintaining freshness.

Data on refresh impact.

What we measured:

Tracked 100 articles before and after refresh:

The insight:

Even minimal updates help. But substantial updates have much bigger impact.

ROI calculation:

For efficiency:

• Minimal updates = highest ROI per hour
• But biggest absolute gains from substantial updates
• Balance based on content value

Recommendation:

Minimal updates for everything, substantial updates for high-value content.

Strategy for systematic updates.

The refresh calendar:

Set recurring refresh tasks:

The update checklist:

For each refresh:

• Update dateModified schema
• Add visible “Last updated” date
• Refresh at least one statistic
• Add recent example if possible
• Check all links still work
• Update any outdated information
• Add new developments if any

Tracking updates:

Content management spreadsheet: | URL | Last Updated | Next Scheduled | Priority |

This prevents content from falling through cracks.

Technical implementation for freshness signals.

Schema for freshness:

{
  "@type": "Article",
  "datePublished": "2023-05-15",
  "dateModified": "2026-01-04",
  ...
}

Always update dateModified when you make changes.

Visible freshness signals:

<p class="last-updated">
  Last updated: January 4, 2026
</p>

Put this near the top, not buried at bottom.

Sitemap freshness:

<url>
  <loc>https://example.com/article/</loc>
  <lastmod>2026-01-04</lastmod>
  <changefreq>monthly</changefreq>
</url>

Keep sitemap lastmod accurate.

Crawl encouragement:

Fresh content should be:

• Linked from frequently crawled pages
• Included in updated sitemaps
• Internally linked from recent content

Clear action plan now. Summary:

Key insights:

1. Content decay is real for AI - especially for evolving topics
2. Update frequency varies by topic type
3. Even minimal updates help - visible date + schema
4. Substantial updates have more impact but take more time

My refresh strategy:

High priority (monthly):

• Top 20 pages by traffic/value
• Pages with declining AI citations
• Scope: Moderate to substantial updates

Medium priority (quarterly):

• Product/service pages
• Current year blog content
• Scope: Moderate updates

Low priority (bi-annually):

• Older blog posts still getting traffic
• Scope: Minimal updates

Archive:

• Content no longer relevant
• Redirect or de-index

Technical implementation:

• dateModified schema on all updates
• Visible “Last updated” dates
• Accurate sitemap lastmod

Tracking:

• Am I Cited for citation trends by content age
• Flag declining content for refresh priority

Thanks everyone - now I have a system.