AI search visibility

The 13-Week Half-Life of an AI Citation (And What It Means for You)

Getting cited once is not winning. AI citations decay with a roughly 13-week half-life, so visibility is a cadence, not a finish line. Here is the upkeep that keeps you quoted.

Updated 2026-06-017 min

What does a 13-week half-life actually mean?

Half-life is borrowed from physics, and it travels well here. It does not mean a citation vanishes after thirteen weeks. It means that, left untouched, the visibility you hold today decays to roughly half over a quarter, then half again over the next, on a steady curve down.

The practical reading is simple. A page that earns an AI citation in January is materially weaker by April and faint by midsummer, unless something keeps it current. Citation is not a trophy you win once. It is a position you have to keep defending.

Why do AI citations decay in the first place?

Answer engines reward freshness, and the web is competitive. Newer sources appear, rivals update their pages, and the model has more recent, more specific text to pull from. Your once-quoted passage simply stops being the best available answer.

There is also a mechanical reason. These systems rebuild their indexes on their own cadence, and each rebuild is a fresh contest for the slot. A page that has not changed since the last contest enters the next one looking older than it did, relative to everything around it.

  • Fresher rivals publish or refresh and become the better passage to quote.
  • Index rebuilds re-run the contest, and stale pages lose ground each time.
  • Dates and figures age visibly, and an engine reads a 2024 stat as a 2024 stat.
  • Your early reviews stop reflecting the current product, so they read as out of date.

How fast do changes actually take to land?

Weeks, not days. When you refresh a page, answer engines do not see it the moment you publish. They reflect the change on the next index refresh, which runs on their schedule rather than yours.

This cuts both ways. It means a refresh you make today is upkeep for the quarter ahead, not a same-week fix, so the work has to be planned, not reactive. It also means decay is gradual rather than a cliff, which is exactly why a steady cadence beats a frantic one.

What should a quarterly refresh actually update?

Update the things an engine reads as signals of currency: the dates on the page, the statistics you cite, and the reviews you surface. A quarterly pass through these three sustains the visibility you earned, because it puts a fresher version of the page into the next index contest.

The point is not to churn for its own sake. It is to make sure the page that gets re-evaluated each quarter is genuinely more current than it was last quarter, and more current than the rivals beside it.

  • Dates: a visible, honest updated date, and any year references brought current.
  • Statistics: refreshed figures with their source label, replacing anything ageing.
  • Reviews: newer reviews surfaced so the social proof reflects the product today.
  • Claims: any line the engine quoted, re-checked so it still reads as true and specific.

Why does this favour operators who keep at it?

Because decay is the great leveller, and most people stop after publishing. A competitor who writes one strong page and walks away starts losing the slot within a quarter. An operator who returns every quarter to refresh the same page keeps re-entering the contest with the freshest version, and compounds a lead that the one-time publisher cannot match.

This is unglamorous work, and that is the point. The advantage is not a clever trick. It is the discipline of upkeep, which is precisely the advantage that is hardest for a rival to copy.

Where do existing reviews fit into the cadence?

Reviews are the part of the page that ages on its own, even when you change nothing, because the product moves on and the early reviews stop describing it. Surfacing newer reviews each quarter is the cheapest freshness you have, since you are not writing copy, you are letting current customers refresh the page for you.

There is a catch. Reviews only help the cadence if an engine can read and trust them. Most review apps were built for the on-page shopper and stop there, leaving the review text locked in a widget and uncorroborated. Getting your existing reviews readable, corroborated, and cited in search and AI is the gap BetterReviews is built to close, which turns each quarter of new reviews into freshness an engine can actually quote.

What this adds up to

Treat AI visibility as a garden you keep, not a wall you build once. Citations decay on a roughly thirteen-week half-life, changes take weeks to land, and a quarterly refresh of dates, stats, and reviews keeps the page winning the next contest. The work is plain and repetitive, which is exactly why operators who keep at it pull ahead of those who publish and forget.

~13 weeks
Measured half-life of an AI citation, so visibility decays without upkeep
AEO research synthesis, 2025
Weeks, not days
Lag before a refresh lands, because indexes rebuild on their own cadence
AEO research synthesis, 2025
Quarterly
Refresh cadence for stats, dates, and reviews that sustains visibility
AEO research synthesis, 2025
Common questions
How often do I really need to refresh a page to stay cited?
Quarterly is the honest answer, because the half-life is roughly thirteen weeks. A pass every quarter through the dates, statistics, and reviews on a page keeps a fresher version entering each index contest, which is what holds the citation. More often rarely hurts, but less often lets decay win.
If I refresh a page, why do I not see the change immediately?
Because answer engines reflect changes on their own index cadence, which runs in weeks, not days. A refresh you publish today is upkeep for the quarter ahead, not a same-week fix. This is why a planned, steady cadence works and a frantic last-minute edit does not.
Does updating the date alone keep a citation alive?
No, a date change on its own is thin, and an engine reads it as such if nothing else moved. The date is one signal of currency among three. It works when it sits alongside refreshed statistics and newer reviews, so the page is genuinely more current, not just stamped with a new number.
Why do my competitors seem to hold their AI citations longer?
Usually because they keep refreshing the same pages and you published once. Decay hits everyone at roughly the same rate, so the operator who returns each quarter re-enters the contest with the freshest version and compounds the lead. The advantage is upkeep, not a trick.