[The Engines]
How Often Should Brands Update Content to Stay Visible in AI Searches?
A new survey puts a number on something citation-hungry brands have suspected for months: content older than 90 days quietly drops out of AI answers. Here is what changed and the cadence to fix it.
The short answer
A survey published July 18, 2026 found that content older than 90 days is 60% less likely to surface in AI-generated local recommendations, while pages updated within the last 60 days appeared at nearly double the rate. The fix is not a cosmetic date bump: refresh core local and comparison pages every 30 to 45 days, FAQs every 60, and treat 90 days as the cliff edge, not a soft decline.
What changed in how AI engines treat old content?
The trigger is a survey covered by Influencers Time on July 18, 2026, tracking thousands of local queries across mid-sized metro markets against the update dates of the pages AI tools recommended. The finding: content older than 90 days is 60% less likely to surface in local AI-generated recommendations, and businesses updating content within 60 days appeared in AI answers at nearly double the rate of businesses that had gone six months or longer without a change.
The more useful detail is the shape of the decline. It is not gradual. The survey describes it as a cliff edge, with recommendation rates holding steady and then dropping sharply once a page crosses the 90-day mark. That matters for planning: a quarterly refresh cycle that lands at day 91 is functionally the same as never refreshing at all.
One more wrinkle worth knowing before you touch anything: cosmetic updates, meaning a changed 'last updated' date with no real edit underneath, showed negligible effect. The pages that recovered visibility were the ones with substantive changes: new hours, new pricing, a rewritten FAQ answer, a staff update. AI systems appear to be reading for actual change, not just a timestamp.
Who does this affect most?
Local, multi-location, and service businesses take the direct hit, since the survey's query set was built around 'best [service] near me' style prompts, the exact question shape those businesses depend on. If your citation share lives or dies on AI Overviews and ChatGPT recommending you for a nearby search, this is not a background trend, it is the mechanism deciding whether you show up at all.
A separate analysis from Seer Interactive, published July 24, 2026 and built on 7,683 dated pages carrying 47,097 citations across ChatGPT, Gemini, and Perplexity, found the pain is not evenly distributed by content type either. Marketplace and comparison pages need the most frequent updates, with 77% to 78% of cited pages updated within the past year. Blogs and guides sit lower at 67%. That puts B2B SaaS teams squarely in the blast radius too: the 'best [category] tool' comparison page, the exact asset most SaaS brands lean on for citation share, is one of the freshness-hungriest formats there is.
| State | Content age | Odds of surfacing in AI-generated local recommendations |
|---|---|---|
| Before | 90+ days since last substantive update | 60% less likely to surface |
| After | Updated within the last 60 days | Nearly 2x the surfacing rate of content inactive 6+ months |
What does the before-and-after data actually show?
Strip the survey down to two states and the picture is stark. Before: a page sitting untouched past 90 days, quietly losing 60% of its odds of being surfaced. After: the same page updated within the last 60 days, back in the recommendation pool at close to double the rate of a page left stale for six months or more.
The table below is the before-and-after in one view. Treat it as a diagnostic, not a target: the goal is never to sit exactly at the edge of the cliff, it is to stay far enough inside the fresh window that a missed sprint does not push you over it.
How should you respond, and on what cadence?
The survey's own recommended cadence gives a workable starting point, and it varies by content type rather than applying one blanket rule across a whole site.
The common failure mode is treating this as a publishing task instead of an editorial one. A junior VA changing a date field does not move the needle. What moves it is someone who actually knows the business updating the substance: does the FAQ still answer the question a prospect is asking today, is the pricing page still accurate, has a case study aged out.
- Core local pages (hours, pricing, service areas): refresh every 30 to 45 days
- FAQ content: refresh every 60 days
- Review responses: reply within 48 hours
- Blog and thought-leadership content: substantive refresh at least quarterly
- Comparison and 'best X' pages: treat as core pages given the 77% to 78% freshness rate Seer Interactive found among cited marketplace and comparison content
Does publishing new content beat updating old content?
Not by itself. Seer Interactive's July 24, 2026 analysis found that 75% of the pages LLMs cite were updated within the last year, but only 42% were actually published within that window. More than a quarter of the pages reading as 'fresh' to AI systems were first published more than two years ago. The freshness these engines reward is being manufactured through maintenance, not through a stream of net-new articles.
A separate study from Gander AI Search Intelligence, published April 7, 2026 and built on 194,077 unique cited URLs, frames this as a roughly one-year half-life: each year of age cuts a page's visibility by about 40% to 60%, with two-year-old content operating at around 33% of its peak visibility and three-year-old content below 25%. Read together, the two studies say the same thing from different angles: a well-maintained two-year-old page can outperform a neglected one published last month, but only if the maintenance is real.
What should your content team do this quarter?
Start with an age audit, not a new content calendar. Pull every page that touches a local, comparison, or 'best X' query, sort by last substantive edit, and flag anything past 90 days first, since that is the group already past the cliff described in the survey.
Assign an owner to each cadence tier above, and measure the result the way the survey did: not by traffic, but by whether the page keeps showing up when someone asks an AI engine the question it targets. That is the discipline citation engineering is built around: coverage and freshness sustained at a pace most in-house teams cannot hold on their own, across every engine that matters, not just Google.
Source: Seer Interactive, Study: Content Recency's Impact on AI Visibility in 2026, 2026-07-24
Key takeaways
- A July 18, 2026 survey found content past 90 days without a substantive update is 60% less likely to surface in AI-generated local recommendations.
- The decline is a cliff, not a gradual slope: rates hold steady, then drop sharply past the 90-day mark.
- Cosmetic date changes do not help. Only substantive edits (pricing, hours, FAQ answers, staff changes) recover visibility.
- Maintenance beats net-new publishing: Seer Interactive found 75% of AI-cited pages were updated within a year, but only 42% were published that recently.
- Freshness needs vary by content type: comparison and marketplace pages need updates far more often (77% to 78%) than blogs and guides (67%).
- A workable cadence: core local and comparison pages every 30 to 45 days, FAQs every 60 days, review responses within 48 hours, blog content quarterly.
Omnicite Editorial. "Content Freshness: How Often to Update for AI Search" The Citation Report, Omnicite. https://omnicite.co/blog/how-often-should-brands-update-content-to-stay-v/
Sources
Content older than 90 days is 60% less likely to surface in local AI-generated recommendations; content updated within 60 days surfaces at nearly double the rate of content inactive 6+ months. Influencers Time, 2026-07-18
75% of AI-cited pages were updated within the last year (Gemini 78%, ChatGPT 73%, Perplexity 65%), but only 42% were published that recently; comparison and marketplace content requires the highest update frequency at 77% to 78%. Seer Interactive, 2026-07-24
Content in AI search has roughly a one-year half-life, with each year of age cutting visibility by about 40% to 60%; two-year-old content operates at around 33% of peak visibility and three-year-old content falls below 25%. Gander AI Search Intelligence, 2026-04-07
Frequently asked questions
How old can content be before it stops getting cited by AI engines?
The cliff edge sits around 90 days. A survey published July 18, 2026 found content past that point is 60% less likely to surface in AI-generated local recommendations, with the drop arriving sharply rather than gradually.
Does changing the last updated date without editing the content help?
No. The same survey found cosmetic date changes showed negligible results. Visibility only recovered on pages with substantive edits, such as new pricing, hours, or FAQ answers.
Which AI engines care most about content freshness?
Gemini and ChatGPT skew freshest, with 78% and 73% of their cited pages updated within the past year respectively, according to Seer Interactive's July 24, 2026 study. Perplexity is more forgiving of older content at 65%.
Do local businesses need to update content more often than a SaaS blog?
Yes for core local and comparison content. The 90-day cliff hits local 'near me' style pages hardest, and Seer Interactive found comparison and marketplace pages need updates at the highest rate (77% to 78%) of any content type, which also puts B2B comparison pages in scope.
Is publishing new content better than updating old pages?
Not by itself. Seer Interactive found 75% of AI-cited pages were updated within the last year but only 42% were published that recently, meaning maintained older pages routinely outperform neglected new ones.
What is a safe minimum update cadence to stay visible in AI search?
Refresh core local and comparison pages every 30 to 45 days, FAQ content every 60 days, respond to reviews within 48 hours, and give blog or thought-leadership content a substantive quarterly pass.