GEO timeliness decay:
the 30-day rule
In generative engine optimisation, the older a page gets, the less likely an AI engine is to cite it. Past thirty days without an update, visibility starts to slide. By six months, most of it is gone — unless nothing better has come along to take the slot.
What timeliness decay is
Timeliness decay is the observable pattern that the staler a piece of content gets, the lower its probability of being quoted in an AI-generated answer. The so-called 30-day rule is the shorthand: once a page has gone a month without being updated or republished, its visibility inside AI search begins to fall away.
It is not a penalty and it is not a filter. It is the by-product of how these systems are built — an engine assembling an answer in real time has to decide which of several candidate sources to lean on, and recency is one of the things it weighs.
From golden window to near zero
The shape below describes a contested, time-sensitive topic — news, trends, pricing, industry data. Decay starts the moment you publish; what changes across the stages is how fast it accelerates.
For time-sensitive queries, models lean hard toward material updated inside the last thirty days. This is when a page is most likely to be cited.
Decay actually started at publication. After thirty days the descent picks up speed and citation frequency starts to thin out.
The critical inflection. Content is reported to shed roughly half its AI-search visibility at around the two-month mark.
Six months untouched and the cumulative loss runs 70–90%. The slot has been taken by someone who kept updating, and the engine now treats the page as a stale source even where the facts still hold.
Past the 180-day mark, small patches stop working. Getting the slot back takes a substantive rewrite plus rebuilt freshness signals — and the recovery period is longer than the decay was.
Decay is relative, not absolute
The curve above is not a countdown clock running on your page in isolation, and this is the part most treatments of the 30-day rule get wrong.
AI engines do not discard content because it aged. They pick the best available answer to the question in front of them, and recency is one input into "best" — sitting alongside depth, structure, authority and how directly the page answers. A page loses its citation when something better and newer takes the slot. Where nothing better and newer exists, it keeps it.
A 180-day page still wins when…
- Nobody in the category has published something both better and more recent since
- It is genuinely more thorough, better structured and more direct than anything else on offer
- The topic is evergreen — definitions, history, fundamentals — where recency carries less weight
- Nothing inside it has been superseded; none of it has quietly become wrong
It loses the slot when…
- A rival publishes something comparable or better on the same topic, more recently
- The figures, prices or rankings inside it have been overtaken by newer numbers
- The query itself is time-sensitive — news, trends, pricing, "best X in 2027"
- Its freshness signals read as old even where the substance still holds
Set your update cadence against your category, not against a calendar. In a slow-moving niche, quarterly may be plenty. In a contested one, thirty days is already behind. Quality buys you time; it does not buy you exemption.
What AI actually checks
Crawlers read freshness off three signals. All three have to agree.
The dateModified property
The machine-readable timestamp in your schema markup. The first thing checked, and the easiest to get wrong or leave stale.
A visible "last updated" date
Rendered on the page where a reader — and a crawler parsing the DOM — can see it. Not buried in a byline or a footer.
New data in the body
Whether the text itself now carries figures, citations or conclusions that were not there before.
All three are required. Changing the date without changing the content is a well-understood pattern and the engines detect it — you spend the credibility and get none of the lift.
In traditional SEO, content is an asset.
In GEO, it's milk.
With one qualification: the expiry date is set by the competition, not by the clock. Content spoils when something fresher and better is standing next to it on the shelf — which is why the cadence that protects you is your category's, not the calendar's.
Why GEO decays faster than SEO
In traditional search, a well-linked, well-structured article can hold the top of Google for years without being touched. Generative engines work on different logic: they synthesise an answer at query time, and their underlying incentive is to avoid stating anything out of date. That pushes them toward aggressive discounting of older material.
The practical consequence is that your position is contingent on what everyone else did last week. A rival publishing on the same topic yesterday is enough for an engine to drop your thirty-day-old piece and cite theirs instead.
Traditional SEO
- Ranking persists; a page can sit still for years
- Links and structure compound over time
- Age can read as accumulated authority
- Refreshing is an optimisation, not a defence
GEO
- The answer is assembled fresh at every query
- Recency is discounted steeply against publication date
- Age reads as risk of stating something wrong
- Refreshing is how you hold the slot at all
The three-step maintenance workflow
Three habits, applied to the pages that actually carry traffic and citations — not to the whole library.
Rolling updates
Put your highest-traffic core pages on a standing refresh schedule. Pace it by how contested the topic is, not by the calendar — and never let a commercially important page reach 180 days before you look at it.
Substantive revision
Every update carries at least 15–20% genuinely new material — new figures, new citations, a revised conclusion. Cosmetic edits do not move the signal and are recognised for what they are.
Automated timestamps
Configure the CMS so every page revision writes a fresh dateModified automatically, and surface the same date on-page. The freshness signal should never depend on someone remembering to set it.
In the GEO era, your content's competitiveness is not decided by how well it was written — but by when it was last updated, and by whether anyone has done it better since.