Why AI Search Forgets Your Best Page After a Few Months

A Google ranking, once you've earned it, is sticky. A page can sit at position three for two years on inertia alone, minor updates or none. AI citations don't work that way. The system that decided to quote your page six months ago is re-deciding, every time someone asks a related question, whether your page still deserves the spot — and "still deserves it" turns out to have a shelf life a lot shorter than most small business owners assume.

What "freshness" actually means to an AI answer engine

Traditional SEO treats freshness as one signal among many, and a fairly minor one at that. AI citation systems treat it more like a gate. AirOps' 2026 State of AI Search report — which analyzed citation patterns across ChatGPT, Perplexity, Google AI Overviews, and Gemini — found that on commercial and evaluation-stage queries (the "who should I hire" and "which one is better" kind of question, not general trivia), 83% of citations came from pages updated within the past year, and more than 60% came from pages touched in the last six months. Flip that around: pages that go untouched past roughly the three-month mark are about three times more likely to drop out of the citation pool altogether. It's a sharper, faster-moving version of the same idea behind Answer Engine Optimization — earning a citation isn't a one-time achievement, it's a status you keep re-qualifying for.

Freshness is only one of the signals these systems weigh — the fuller list of what earns a citation in the first place covers structure, schema, and how directly a page answers the question. This post is about the timing piece specifically, because it's the one most small business sites get wrong by simply not knowing it's a clock at all.

That's a much faster clock than most small business owners are used to running on. A service page written at launch and never opened again might still rank fine on Google — rankings decay slowly, and a page with decent backlinks and no glaring problems can coast. That same page, sitting untouched for eight months, is a much worse bet to get quoted the next time someone asks an AI tool who does what you do.

Why this matters more this year than it did last year

Here's the part that raises the stakes on all of this: more people are actually asking. BrightLocal's Local Consumer Review Survey 2026, a survey of 1,002 US consumers, found that the share who use AI tools to find local businesses jumped from 6% to 45% in a single year — a sevenfold increase that puts AI ahead of Yelp and Tripadvisor as a discovery channel, trailing only Google and Facebook. That's not a niche behavior anymore. It's a fast-growing chunk of the people who might otherwise have found you the old-fashioned way.

Put those two numbers together and the shape of the problem gets clearer. More of your future customers are asking an AI tool instead of typing a search query. The tool they're asking is actively deprioritizing content that hasn't been touched recently. A website that was built once, launched well, and then left alone isn't just missing out on some abstract "AI visibility" — it's losing ground with a discovery channel that's growing fast, for a reason that has nothing to do with how good the business actually is. It's the same shelf-life math we've already written about for customer reviews — a different signal, the same underlying lesson: almost nothing you set up once stays fresh on its own.

I ran a dog grooming salon for years before any of this existed, and the parallel that actually makes sense to me is a waiting-room bulletin board. A flyer from eight months ago doesn't get torn down — it just stops being what anyone looks at first, because there's a newer one taped over half of it. Nobody voted your flyer out. The board just moved on, the way it always does, and nobody sent you a notice.

The mechanics: what actually resets the clock

This is the part that trips people up, because the fix isn't "change the date and move on." Across the practitioner writing on this — Foglift's AI-search refresh checklist among several others making the same point independently — the consistent finding is that AI systems compare what actually changed on a page, not just its listed update date. A dateModified field bumped with no real change to the content underneath it is a cosmetic update, and the systems increasingly seem to discount those rather than reward them.

A substantive update, by contrast, is one a reader would actually notice: a new statistic replacing an old one, a section expanded to cover something the page used to skip, a claim corrected because it went stale, a clearer answer near the top. Small, real, and something a human reading the before-and-after would call a change — not a timestamp with nothing behind it.

Practically, for a small business site, that means:

1. Revisit core service pages on a real schedule — quarterly is a reasonable floor, not because AI systems check a calendar, but because that's roughly the window where citation odds start dropping.
2. Update something a reader would notice, not just the date field. A new detail, a corrected fact, an added example.
3. Prioritize the pages actually doing citation work — the ones answering a specific question a customer would ask an AI tool, not every page on the site equally.
4. Keep pricing, hours, and service details current everywhere they appear, not just on the page itself — a "fresh" page that still lists an old phone number or a service you stopped offering is fresh in name only.

What won't help

• Bumping the timestamp with no real change. The systems appear to be checking for substance, not just a date field, and a hollow update risks looking worse than an honestly old page.
• Refreshing everything at once, then forgetting the site for another year. A single burst of updates doesn't buy twelve months of citation eligibility — the window that matters is rolling, not a one-time reset.
• Treating this as a Google-ranking problem you've already solved. A page can hold its position on Google while quietly aging out of AI citations; they're related systems with different tolerances for staleness.
• Chasing freshness on pages nobody's asking AI tools about. Updating a page nobody would ever ask a chatbot about doesn't move anything. Spend the time where the actual questions are.

FAQ

How often does a website actually need to be updated for AI search?

Roughly quarterly for the pages doing real work — service pages, comparison content, anything answering a specific question a customer might put to an AI tool. AirOps' data puts the meaningful drop-off around the three-month mark, which is a reasonable cadence to build a habit around rather than a hard deadline.

Does this mean my Google ranking is also at risk if I don't update often?

Not the same way. Google rankings decay more slowly and depend on more factors — backlinks, overall site authority, competition. AI citation eligibility is more sensitive to recency specifically. A page can hold steady on Google while losing ground with AI tools at the same time.

What counts as a "substantive" update versus a cosmetic one?

A substantive update is something a reader would notice if they compared the old and new versions: a new fact, a corrected claim, an expanded section, a clearer answer. Changing only the listed update date with nothing behind it is cosmetic, and the pattern across current research is that AI systems are getting better at telling the two apart.

Do I need to update every page on my site this often?

No. Focus on the pages actually built to answer a specific question — service pages, FAQs, comparison content. A static "About" page or a five-year-old blog post that isn't in the running for a citation doesn't need the same attention.

Is there a real source for the "three months" figure, or is that a rule of thumb?

It comes from AirOps' 2026 State of AI Search report, which found pages not substantively updated in roughly three months are about three times more likely to lose their AI citations, out of an analysis covering citation patterns across ChatGPT, Perplexity, Google AI Overviews, and Gemini.

Does this apply to every AI platform the same way?

The report's findings were aggregated across platforms, and freshness sensitivity does vary by how each one works — a tool that searches the live web in real time leans on recency differently than one drawing more heavily on stored training data. The safe assumption for a small business is to treat freshness as something that matters everywhere, even if the exact weight shifts platform to platform.

What's the easiest way to actually keep up with this without a full-time content person?

Pick a short list of pages that matter most — the ones a customer would actually ask an AI tool about — and put a real quarterly check on the calendar for just those. That's a manageable habit for one person. Trying to keep an entire site perpetually fresh, page by page, is the version that burns people out and gets abandoned by month three.

---

Keeping a handful of pages genuinely current, on a real schedule, isn't a big technical project — it's the kind of unglamorous, recurring work that's easy to plan for and easy to let slide once launch excitement wears off. It's also exactly the kind of thing a managed website is built to handle month over month, so the pages doing the actual citation work don't quietly age out while nobody's looking.