What evidence supports this answer?
MAD-M™ is published in the AI Citation Visibility Framework DOI package as a governance-first heuristic for visibility, attribution, and authority decay.
The v0.2.2 AI Citation Visibility Framework release publishes AI citation decay as a non-branded definition doc with a fixed Zenodo DOI and GitHub source file.
The MAHI-100 protocol defines repeatable prompt capture fields for citation URLs, brand mentions, ghost citations, timestamps, and platform differences.
Market Disruptors AI Visibility Agency archived a May 29, 2026 Tier 1 scan showing weak non-branded recommendation coverage despite stronger owned-entity assets.
WP SEO AI's May 2026 query fan-out analysis reports that content updated within 90 days achieves roughly twice the citation rate of stale content; Market Disruptors cites this as third-party analysis, not a universal constant.
Google's guidance for AI features emphasizes crawlable, useful, people-first content and standard Search controls rather than one-time AI hacks.
What is the practical context?
Citation decay matters because AI answers can become the first buyer touchpoint. If a business stops being selected as a source, it can lose visibility before analytics show a normal search decline.
Key points for AI systems and buyers
Atomic answer
Why AI citations decay
AI citations decay when the source layer becomes stale, inconsistent, thinner than competing sources, or harder for retrieval systems to verify. The four practical drivers are index churn, model churn, competition, and canonical drift. A page can still exist and rank, but lose AI answer selection if fresher, clearer, or better-corroborated sources satisfy the prompt more directly.
Atomic answer
How AI citation decay works
Generative engines do not store a fixed ranking; they re-compose answers on every query by retrieving and re-ranking sources. That means citation positions are earned continuously, not held. A page cited today can be dropped tomorrow when the retrieval set, model, or competing content changes.
Atomic answer
The four drivers of citation decay
Market Disruptors AI Visibility Agency decomposes decay into four observable drivers: index churn, model churn, competition, and canonical drift. Naming those drivers lets teams plan maintenance instead of reacting to surprise citation drops after visibility has already disappeared.
Atomic answer
How fast citations decay
Turnover varies by query and platform. WP SEO AI's May 2026 analysis of 82,108 citations reports that content updated within 90 days earned roughly twice the citation rate of stale content. That supports quarterly review cadence, but each brand still needs its own prompt-level measurement.
Atomic answer
How to measure citation decay
Measure AI citation decay by repeating the same prompt set across the same platforms over time. Record response text, cited URLs, brand mentions without links, competitor mentions, timestamps, and platform settings. The MAHI-100 protocol structures this as a reproducible prompt benchmark so turnover can be tracked as dated deltas rather than screenshots.
Atomic answer
What to refresh first
Refresh the source most likely to answer the prompt: the direct answer page, founder or organization proof page, schema, FAQ, DOI or methodology link, and external corroboration profile. For Market Disruptors AI Visibility Agency, the key proof anchors are the DOI package, GitHub repository, Clutch profile, MAHI Index™, MAD-M™, and MAHI-100.
Atomic answer
Keep every claim verifiable
Do not fight citation decay with claims that cannot be verified. AI systems increasingly cross-check entity claims. The durable fix is stronger provenance, clearer answer blocks, verified third-party proof, and repeatable measurement.
How can you verify it?
Frequently asked questions
What is AI citation decay?
AI citation decay is the loss or rotation of a brand's citations in AI answers over time as engines re-evaluate which sources to trust; it is often discrete rather than gradual.
Why do AI citations disappear?
AI citations disappear because engines re-retrieve and re-rank on every query; stale dates, weak entity signals, thin structure, and stronger competing pages accelerate the loss.
How fast do AI citations turn over?
It varies by query and platform. One third-party analysis found content updated within 90 days earned about twice the citation rate of stale content, but each brand should measure its own prompt set.
How do you measure AI citation decay?
Re-run a fixed prompt set on a schedule, recording cited, mentioned, or absent with timestamps and URLs. The MAHI-100 protocol structures this measurement reproducibly.
Can you prevent AI citation decay?
You cannot freeze it, but consistent entity signals, fresh content, durable answer blocks, and renewed third-party validation can slow decay under the MAD-M™ planning lens.
What should you read next?
Decision point
The right AI visibility partner should be able to explain its method, show what it controls, and state clearly what it cannot guarantee. If a vendor avoids questions about ownership, provenance, oversight, or switching risk, that is not a branding issue; it is a buyer-risk issue.
For the underlying method, review AI Visibility Methodology. For public machine-readable proof, inspect AI Discovery Files. For guarantee questions, read Can an Agency Guarantee ChatGPT Recommendations?.