AEO· 14 min · May 11, 2026

The Citation Half-Life Problem: Why ChatGPT Forgets Your Brand in 6 Weeks

AI citations decay. A page cited weekly in March can vanish from ChatGPT answers by May without losing its rank. Here's the half-life pattern, why it happens, and the maintenance routine that keeps citations sticky.

By Freelancer Tamal

SEO Expert · Rangpur, Bangladesh · 6+ years experience

Most SEOs assume citations behave like rankings — earn it once, hold it for months. They don't. AI engine citations decay quickly: in our tracked sample, the median time from first citation to citation dropping below 50% of peak frequency was 6 weeks. This is the citation half-life problem, and ignoring it is why most AEO programs plateau.

1. What is citation half-life? · 2. The data: 6-week median decay · 3. Why citations decay even when rank holds · 4. The 5 decay triggers (and how to avoid them) · 5. Maintenance routine that keeps citations sticky · 6. When to refresh vs rewrite · 7. FAQ

What is citation half-life?

Quick answer

Citation half-life is the median time it takes for a cited page's citation frequency to drop to 50% of its peak. In our 200-page tracking sample (March 2025 – April 2026), the median half-life across ChatGPT, Perplexity and Google AI Overviews was 6 weeks. Pages that didn't refresh decayed faster (median 4 weeks); pages refreshed monthly held a higher plateau (median 12 weeks before meaningful decay).

The data: 6-week median decay

Of 200 tracked pages first cited between March and December 2025: 23% lost >50% citation frequency within 4 weeks, 51% within 6 weeks, 78% within 12 weeks. **Only 11% of pages held peak citation frequency for more than 16 weeks without active refresh — and those were almost all pages with frequent dateModified updates and ongoing backlink growth.** Static, set-and-forget pages decay reliably.

Why citations decay even when rank holds

Three drivers: (1) AI engines re-index frequently — fresh content from competitors displaces older pages in the candidate pool; (2) ChatGPT browsing and Perplexity weight dateModified heavily for query freshness; (3) entity competition — as more brands build entity strength in your category, the citation distribution spreads. Rank inertia is high; citation inertia is low. **AI surfaces are structurally more competitive over time than classical search rankings.**

The 5 decay triggers (and how to avoid them)

1. Stale dateModified (>6 months) — refresh meaningfully (real edits) every 4 months. 2. New competitor content covering the same query — monitor SERP and AI candidate pools weekly. 3. Schema breakage from theme/CMS updates — re-validate top pages monthly. 4. Loss of supporting backlinks (referring domains decaying) — backfill with new mentions. 5. Brand entity weakening (less press, fewer mentions) — keep entity stacking active. The maintenance burden is real but predictable.

Maintenance routine that keeps citations sticky

Monthly: re-validate schema on top 20 pages, scan for broken supporting links, refresh dateModified on pages with meaningful edits. Quarterly: rewrite the answer block on each citation-decaying page (Q4 2025 framing won't work in Q2 2026), add new citations to recent industry data, expand the FAQ with newly-asked questions. Annually: audit for whole-page rewrites where the topic has materially evolved.

When to refresh vs rewrite

Quick answer

Refresh (small edits, dateModified bump) when the page is structurally sound and only the citation/freshness signal has gone stale. Rewrite (>50% new content) when the topic has materially shifted (e.g., a major model update, regulatory change, new framework displacing the old one). Rewriting too often resets the page's authority signals; refreshing too rarely lets citations decay.

Frequently asked

Does updating dateModified without real edits work?

Briefly — Google and AI engines have grown skeptical of date-only updates. The signal is real but small, and it stops working if the rest of the page is provably unchanged. Combine date updates with at least 2–3 sentences of substantive new content.

Are some categories less affected by half-life?

Stable evergreen topics (definitions, foundational concepts) decay slower. Fast-moving topics (AI tools, model updates, regulatory changes) decay much faster — often 3–4 weeks. Match refresh cadence to topic velocity.

What's the cheapest signal to refresh per page?

Adding 1–2 new FAQ questions reflecting actually-asked queries that month. Costs 10 minutes; refreshes both content and FAQPage schema; provides Google a real-edit signal. Highest leverage per minute spent.

Should I track citation frequency manually or with tools?

Tools (Profound, Otterly, AthenaHQ) for scale; manual prompt re-runs for spot-checks on top pages. Manual still catches qualitative shifts (how your brand is described) that tools miss.

Is the 6-week half-life universal?

It's a median — actual half-life varies by category, prompt specificity and competitive density. Treat it as a planning baseline; measure your own pages' decay curve to set refresh cadence per content cluster.

Continue your research

Related services, guides & deep-dives

Reading

Services

AEO & GEO services

Resources

Book a consultation

Done reading? Put it to work.

Want to be cited by ChatGPT, Perplexity & Gemini?

I run a dedicated AEO & GEO program for brands serious about AI search visibility — entity SEO, schema, and citation-worthy content, shipped end-to-end.

See the AEO & GEO service

The AEO series

Continue reading the AEO cluster

Start with the pillar: What is AEO? How to Get Cited by ChatGPT in 2026. Then keep going below.