Do AI answers cite old blog posts, or only fresh ones?
Old posts get cited when they are clear, sourced and still correct. Age alone neither helps nor hurts. Below, five common beliefs set against what we can state with confidence.
Koi Content editorial team ·
Myth versus reality
The belief
What holds up
What we cannot say
Old posts are invisible to AI search
Google's AI Overviews launched in the US in May 2024 and draw on pages that already rank. A well-kept old guide can appear there
How often any given page is quoted
Paywalling my best work keeps it safe from AI
It reduces what a crawler can read, and also what a searcher can find. You lose both
Whether any one tool respects your rules
Republishing everywhere multiplies citations
Copies dilute the signal; one clear source is easier to attribute
Which copy a given system will choose
Backlinks no longer matter
Links still help pages get found and trusted in conventional results, which feed many AI answers
Their weight inside any specific AI product
Fresh dates beat everything
A visible update date on a truly updated article helps. A changed date on an untouched one does not
How any system weighs recency
What a slow writer can actually do
Keep the article correct. That is the dull core of it. A guide with a stale figure in paragraph three loses trust with readers and, we suspect, with the systems that summarise it.
State things plainly, with numbers and sources, early in each section. Summaries pick up crisp answers more easily than winding ones, and human readers prefer them too, which is the better reason to write that way.
Then give the page a fair chance of being found. A single placement on a relevant site, ordered through the marketplace, is one honest way to do that. Short, and cheap. The patience calendar explains how long to wait before you read anything into the result.
A second habit helps more than any tactic. Keep a dated update note at the bottom of each long post: readers like to see that a 2022 tutorial was rechecked in October 2026, and it gives you a reason to open the file every few months.
Ahrefs and Semrush both publish studies on content refresh that are worth reading with a sceptical eye.
We do not know how ChatGPT, Perplexity or Google choose their sources in detail, and nobody outside those companies does. Anyone selling certainty about it is guessing.
Dates worth knowing
Helpful content update
Google rewarded pages written for readers, and it set the tone for how thin or padded writing was treated afterwards.
Core update with spam policy changes
Helpful-content signals joined the core system, and site reputation abuse was named as a policy.
AI Overviews in the US
Generated summaries began to appear above standard results for many queries.
Questions we hear from long-form writers
Should I write shorter posts for AI answers?
No. Write as long as the topic needs, and open each section with a direct answer. Length is not the lever. Clear structure and correct facts are what let a long article be quoted in part.
Do I need special markup to be cited?
There is no confirmed markup that earns a citation. Clean headings, honest dates and visible authorship are sensible whatever happens. Spending a weekend on tricks is less useful than updating your three oldest posts.
Is it worth buying a link for an AI-era post?
Maybe, for the pages you most want found. One placement is a modest test. Read the cluster budget before deciding how much, and judge the result after a season.