Anything Engine Optimization The rolling record of the AI-search industry.

Glossary

Citation drift — also reported as citation decay

Citation drift is the rotation of which sources an AI engine cites for the same prompt over time, with the prompt and the page both unchanged. It is not malfunction — it is how retrieval-based answers behave, and it is why AI visibility is a distribution to move rather than a position to hold.

SISTRIX sampled 82,619 prompts over 17 weeks ending April 8, 2026. SISTRIX measured weekly domain churn-in, not URL loss: 74% for ChatGPT Search in Germany and 56% for AI Mode there. Rates and stable-core behavior differ by engine and country.

The stable-core finding concerns AI Mode; ChatGPT’s median prompt had no domain present in every sampled week. Domain churn and individual URL survival are different measures.

SISTRIX reports 0.17 Jaccard overlap between the cited domain sets of AI Overviews and AI Mode: shared domains divided by their union. It is not the percentage of prompts on which they agree. The 1.4% core rate belongs to the News/Media domain category in SISTRIX’s AI Mode table. It is not the lifespan of every news URL across AI search.

Drift has more than one cause, and they are worth separating because only some are addressable. The engine's preferences can shift — AI Mode's publisher mentions fell nearly by half in thirty days while retailer mentions tripled across 15 of 20 verticals. Its retrieval method can change — Reddit's 86.4% four-day fall in ChatGPT citations traced to a change in how the engine assembles its background queries. And the model underneath can be swapped at any time, which the glossary's citation-share entry has flagged since it was written. None of those is something a page can be optimized against.

Measurement is arriving. Profound shipped a feature called Citation Decay in August 2026 that tracks week-over-week counts for every cited URL rather than reporting a point-in-time total — the longitudinal view this research argues is the only honest one.

MeasureFinding
Weekly domain churn-in — ChatGPT Search, Germany 74%
Weekly domain churn-in — AI Mode, Germany 56%
AI Mode prompts with a stable core (rounded) 86%
AI Mode peripheral-domain weekly rotation 89%
AI Overviews prompts with no source change in 17 weeks 53%
Complement of Jaccard domain overlap (not prompts) 83%
AI Mode News/Media domain-category core rate 1.4%
AI Mode brand queries retaining brand domain all 17 weeks 43%
Citation drift over 17 weeks — SISTRIX, 82,619 prompts, 1,548,213 snapshots, six countries, published 2026-05-01. Vendor-published and not replicated.

Common questions

What is citation drift?

Changes in the sources cited for a prompt over time. SISTRIX measured weekly domain churn-in, not URL loss: 74% for ChatGPT Search in Germany and 56% for AI Mode there. Rates and stable-core behavior differ by engine and country.

How stable are AI search citations?

The stable-core finding concerns AI Mode; ChatGPT’s median prompt had no domain present in every sampled week. Domain churn and individual URL survival are different measures.

Do Google AI Overviews and AI Mode cite the same sources?

SISTRIX reports 0.17 Jaccard overlap between the cited domain sets of AI Overviews and AI Mode: shared domains divided by their union. It is not the percentage of prompts on which they agree.

How long does an AI citation last?

The 1.4% core rate belongs to the News/Media domain category in SISTRIX’s AI Mode table. It is not the lifespan of every news URL across AI search.

What is citation decay?

Citation decay follows how a page’s citation frequency changes over time. It should not be equated with SISTRIX’s domain churn-in or category core rates; these count different things.

The record: ChatGPTGoogle AI OverviewsGoogle AI Mode

See also: Citation shareAI visibilityQuery fan-outGroundingGhost citation