# Citation Decay

Citation Decay is the rate at which AI engine citations of a brand decrease over time without sustained earned media activity. AI engines continuously re-evaluate source freshness and authority, and brands that stop generating new high-quality signals see their citation presence erode as competitors produce newer, more relevant content.

Canonical URL: https://machinerelations.ai/glossary/citation-decay
Category: metrics
Attribution: Coined by Jaxon Parrott.

## Source Body

## How Citation Decay Works

AI engines prioritize fresh, authoritative sources. When a brand stops producing new earned media signals — Tier 1 placements, structured data updates, citation-worthy content — the engines gradually shift citations toward brands with more recent, more relevant material.

The mechanism is not punitive. AI retrieval systems are not penalizing inactive brands. They are selecting the most relevant, most recent, most trustworthy sources for each query at inference time. When a competitor publishes fresher, stronger material on the same topic, the older source drops from the retrieved set. The brand's citation presence erodes not because it did something wrong, but because it stopped doing what was right.

Research from Seer Interactive confirms the structural bias: AI-cited content is 25.7% fresher on average than content cited in traditional organic Google results ([Seer Interactive, 2026](https://www.seerinteractive.com/insights/study-ai-brand-visibility-and-content-recency)). Pages not updated quarterly are 3x more likely to lose citations, and roughly half of all AI-cited content is less than 13 weeks old ([SalesPeak AEO Research](https://salespeak.ai/aeo-news/content-freshness-ai-search)).

That 13-week window is the practical half-life of a citation-eligible page. After that threshold, decay accelerates unless new signals refresh the brand's authority.

## Citation Decay Across AI Engines

Citation Decay operates differently across engines because each uses different retrieval architectures and freshness weighting.

| Engine | Retrieval Model | Freshness Sensitivity | Decay Speed |
|---|---|---|---|
| **Perplexity** | Real-time web indexing | Highest — content updated within 12 months earns 3.2x more citations | Fast (weeks) |
| **ChatGPT Search** | Browsing + training data | Moderate — 76% of top-cited pages updated in last 30 days, but 29% of citations are from 2022+ when authority outweighs recency | Medium (1-3 months) |
| **Google AI Overviews** | Search index + generative layer | Moderate — inherits some of Google's freshness signals | Medium (1-3 months) |
| **Claude** | Training data snapshots | Lowest for base model — but high for retrieval-augmented queries | Slow (training cycle dependent) |
| **Gemini** | Google index + model knowledge | Moderate to high | Medium (weeks to months) |

The practical consequence: a brand that goes dark on earned media will lose Perplexity citations first, ChatGPT and AI Overview citations next, and base-model mentions last. But by the time base-model citations erode, the competitive gap is often too wide to close quickly.

## Citation Decay vs. Citation Velocity

| Metric | Direction | Measures | Driven by |
|---|---|---|---|
| [Citation Velocity](https://machinerelations.ai/glossary/citation-velocity) | Upward | Rate of new citation accumulation | Active earned media + content production |
| Citation Decay | Downward | Rate of citation loss | Inactivity + competitive displacement |

A brand's net citation trajectory is the difference between velocity and decay. When velocity exceeds decay, [Share of Citation](https://machinerelations.ai/glossary/share-of-citation) grows. When decay exceeds velocity, the brand loses AI visibility — often before anyone notices.

The asymmetry matters: decay is silent. Citation Velocity is measurable through active monitoring. Decay only becomes visible in retrospect, when Share of Citation has already dropped. This lag is why measurement cadence matters — brands that check quarterly often discover decay too late to recover before competitive displacement is entrenched.

## Causes of Accelerated Decay

1. **Competitor activity increases.** New entrants produce fresh, high-authority content that displaces older citations. In active B2B categories, a single competitor's sustained earned media push can displace an incumbent within one news cycle.

2. **Source freshness drops below threshold.** AI engines apply a recency filter during retrieval. Content older than 13 weeks faces progressively steeper citation probability decline. In fast-moving categories like fintech or AI tooling, the threshold is even shorter.

3. **Entity signal degradation.** Outdated information across profiles, press pages, and structured data reduces AI confidence in the brand entity. If the [Entity Chain](https://machinerelations.ai/glossary/entity-chain) weakens, citation-eligible content gets skipped because the engine cannot confidently attribute it.

4. **Relevance drift.** The brand's existing content no longer matches evolving query patterns as the market shifts. A page optimized for "AI marketing tools" in 2024 may not match the 2026 query "AI visibility platform for B2B" — even though the product is the same.

5. **Platform fragmentation.** Only 11% of sites get cited by both ChatGPT and Perplexity ([Discovered Labs, 2026](https://discoveredlabs.com/blog/ai-citation-patterns-how-chatgpt-claude-and-perplexity-choose-sources)). Decay on one platform does not necessarily mean decay on all — but brands monitoring only one engine can miss erosion happening elsewhere.

## What Citation Decay Is Not

Citation Decay is not an algorithm penalty. There is no demotion signal. The brand simply stops appearing because fresher, more relevant sources exist.

It is not the same as SEO ranking loss. A page can maintain its Google position while losing all AI citations, because AI engines apply different selection criteria — including cross-domain authority, entity clarity, and source structure — that traditional SEO ranking does not measure.

It is not irreversible. Resuming earned media production and generating fresh high-authority signals restores citation presence. However, recovery takes longer than maintenance because competitors have filled the gap during the inactive period. The [Algorithm Credibility Moat](https://machinerelations.ai/glossary/algorithm-credibility-moat) a competitor builds during a brand's inactive period raises the cost of re-entry.

## How to Detect and Measure Citation Decay

Track [Share of Citation](https://machinerelations.ai/glossary/share-of-citation) over time with a consistent query set. A declining trend without competitive changes indicates decay.

**Leading indicators:**
- Earned media output has slowed or stopped for 4+ weeks
- No new Tier 1 placements in the current quarter
- Existing cited pages have not been updated in 90+ days
- Entity signals (schema, profiles, press pages) are stale

**Lagging indicators:**
- Share of Citation declining month-over-month
- Brand appearing in fewer AI engine responses for category queries
- Competitors appearing in responses where the brand previously held position

The gap between leading and lagging indicators is the decay window. Brands that act on leading indicators prevent decay. Brands that wait for lagging indicators are already in recovery mode.

---

## FAQ

**How fast does Citation Decay happen?**
Decay rate varies by competitive intensity and engine. In active B2B categories, brands typically see measurable citation drops within 4-6 weeks of stopping earned media activity. In less competitive niches, decay may take 2-3 months to become visible. Perplexity shows the fastest decay; base-model knowledge in ChatGPT and Claude erodes more slowly but more permanently.

**Can Citation Decay be reversed?**
Yes. Resuming earned media production and generating fresh high-authority signals restores citation presence. But recovery is not symmetric — rebuilding lost citations takes longer than maintaining existing ones because competitors have filled the gap. A brand that maintained 30% Share of Citation and let it decay to 15% may need 2-3x the earned media volume to return to 30% compared to what would have been required to maintain it.

**Does updating existing content prevent decay?**
Partially. Refreshing statistics, examples, and substance can extend citation eligibility. But cosmetic edits — changing a date stamp without updating the material — do not work. AI engines evaluate content delta, not just modification dates. Genuine content refreshes extend the citation window; fake freshness does not.

**How do you measure Citation Decay?**
Track Share of Citation weekly with a consistent query bank of 30-50 category-relevant queries across ChatGPT, Perplexity, Gemini, and Google AI Overviews. A declining trend without corresponding competitive changes indicates decay. AuthorityTech monitors this across 45 queries daily.

## Sources

- https://machinerelations.ai/glossary/citation-velocity
- https://machinerelations.ai/glossary/share-of-citation
- https://machinerelations.ai/research/earned-vs-owned-ai-citation-rates-2026
- https://authoritytech.io/blog/optimize-earned-media-for-ai-search-strategy-guide
- https://authoritytech.io/curated/ai-citation-half-life-platform-refresh-playbook-2026
- https://machinerelations.ai/research/citation-freshness-decay-llm-search-2026
- https://machinerelations.ai/glossary/citation-gap
- https://www.seerinteractive.com/insights/study-ai-brand-visibility-and-content-recency
- https://salespeak.ai/aeo-news/content-freshness-ai-search
- https://discoveredlabs.com/blog/ai-citation-patterns-how-chatgpt-claude-and-perplexity-choose-sources
- https://machinerelations.ai/glossary/citation-decay
- https://authoritytech.io/curated/track-share-of-citation-ai-engines-measurement-2026
- https://machinerelations.ai/research/ai-citation-decay-how-brands-lose-visibility-over-time
- https://authoritytech.io/curated/share-of-citation-benchmarks-2026-ai-engines
- https://machinerelations.ai/research/ai-citation-gap-analysis-methodology-2026
- https://machinerelations.ai/research/entity-chain-scoring-measure-cross-domain-authority-2026
- https://machinerelations.ai/glossary/machine-resolution
- https://machinerelations.ai/glossary/ai-citations
- https://machinerelations.ai/glossary/content-freshness
- https://christianlehman.com/blog/why-ai-search-rankings-and-google-rankings-diverge
- https://authoritytech.io/glossary/citation-velocity
- https://authoritytech.io/blog/extractable-content-structure-ai-citations-2026

## Machine-readable related links

### Related concepts

- [AI Visibility](https://machinerelations.ai/glossary/ai-visibility)
- [Entity Chain](https://machinerelations.ai/glossary/entity-chain)
- [RAG Citation (RAG)](https://machinerelations.ai/glossary/rag-citation)
- [Algorithm Credibility Moat](https://machinerelations.ai/glossary/algorithm-credibility-moat)

### Supporting research

- [How to Run an AI Citation Gap Analysis: The Step-by-Step Methodology for Finding What AI Engines Won't Cite (2026)](https://machinerelations.ai/research/ai-citation-gap-analysis-methodology-2026)
- [Citation Freshness Decay in LLM Search: Why Fresh Pages Keep Replacing Older Sources](https://machinerelations.ai/research/citation-freshness-decay-llm-search-2026)
- [Entity Chain Scoring: How to Measure Cross-Domain Authority for AI Citation Eligibility](https://machinerelations.ai/research/entity-chain-scoring-measure-cross-domain-authority-2026)
- [The Content Volume Trap: Why Publishing More Pages Reduces AI Citation Rates](https://machinerelations.ai/research/content-volume-ai-visibility-trap-audit)

### Framework context

- [Machine Relations Index](https://machinerelations.ai/index)
- [Machine Relations Stack](https://machinerelations.ai/stack)
- [Evidence Base](https://machinerelations.ai/evidence)