What is citation decay?

Citation decay, also called AI citation decay, is a lasting drop in how often AI answers cite a URL for the same prompts, after allowing for normal run-to-run variation. One missing citation is not decay. Decay is a decline that persists across several checks with the same prompts, engines and counting rules.

Why it matters for AI visibility

AI answers are unstable. A page can be cited on Monday and absent on Tuesday because the engine retrieved a different set, not because the page failed. Calling every dip decay causes teams to rewrite pages that do not need it. Real decay still matters: a source that used to support your category answers can be replaced by fresher reviews, a changed product, or a page that no longer matches the question. You only see it in a series. Chart whether each URL is cited on each check, and compare a recent window with an earlier one of the same length. Look for a drop that survives the usual noise. Then read the answers that stopped citing you and see which sources took the place. Refresh the page when the substance is stale. Changing only the visible date, without a real edit, is not a remedy you can honestly credit. If decay has already cost you citations, our guide on how to get cited by AI covers how to earn them back.

Example

Illustrative example: a pricing guide is cited in 9 of 12 weekly checks in one quarter (75%), then in 1 of the next 4 checks (25%), while a newer comparison page starts being cited for the same prompts. A drop that size, held for a month, is decay. One empty week between cited weeks is not. Single checks swing a lot: in Kollavo's November 2, 2026 ChatGPT panel, 83% of cited URLs appeared in only 1 of 3 runs of the same question.

Related terms

Prompt tracking produces the series you need. An AI citation is the event that becomes less frequent. Earned media can decay too, when a third-party page that used to be cited is no longer retrieved. Do not treat a one-day gap as a trend.

Related on Kollavo

FAQ

What does AI citation decay mean?

AI citation decay means a page is cited less and less often in ChatGPT, Perplexity, Gemini or Google AI answers for the same prompts over several weeks, not just missing for a day. It is measured as a citation rate over a fixed prompt set and compared between two windows of the same length.

How is citation decay different from normal AI answer volatility?

Volatility is day-to-day churn: a URL is cited on some checks and not others around a steady average. Decay is when that average itself drops and stays lower in the next window. Check both with the same prompts and engines before you rewrite a page.

How do you measure citation decay?

Keep a fixed prompt set and run each prompt on the same engines several times per check. Record whether the URL is cited in each run, then compare its citation rate in a recent window with an earlier window, for example the last 4 weekly checks against the previous 12. A drop that holds after normal run-to-run noise is decay.