TL;DR
- Roughly half of Perplexity's citations point to content less than 13 weeks old; the other half decays fast after that.
- For breaking or trending queries, the effective window compresses to 48–72 hours.
- A 30–45 day refresh cadence is enough to keep evergreen pages inside Perplexity's active freshness pool for most topics.
- Recently updated pages earn roughly 6 citations on average vs. 3.6 for stale pages — a 67% lift purely from freshness.
- Timestamp signals matter: visible
dateModified, updated schema, and genuinely rewritten sections beat cosmetic edits.
Perplexity is the AI search engine with the strongest real-time bias. Unlike ChatGPT's mixed retrieval or Gemini's index-heavy approach, Perplexity issues live web queries against an index of 200+ billion URLs on nearly every prompt. That architecture makes freshness a first-class ranking signal, not a tiebreaker. Below is what the current body of research actually shows about how recent your content needs to be, how often to refresh, and which timestamp signals move the needle.
The recency windows that actually govern Perplexity citations
Amsive's citation research popularized the 13-week rule: about half of AI citations, across engines, point to content published or substantially updated within the last quarter. Perplexity sits at the aggressive end of that distribution. An Ahrefs analysis of 17 million AI citations published in July 2025 found that cited pages are on average 25.7% fresher than the pages ranking organically for the same queries — the gap is real and measurable.
Perplexity narrows further. A 2026 SE Ranking study of 216,524 pages attributed roughly 44.2% of Perplexity's selection weighting to temporal freshness, higher than any other AI search engine measured. Approximately half of Perplexity's citations resolve to current-year content.
For time-sensitive queries — news, product launches, regulatory changes, model releases — that window collapses dramatically. Anecdotal and log-level testing consistently shows a 48–72 hour half-life on breaking topics. Publish on Monday, and by Thursday your citation share on the same query is often being eroded by newer coverage, even if your piece is objectively better.
Practical implication: if you're competing on anything with a news cycle, "publish and move on" doesn't work. You need a follow-up post, a visible update, or a second angle within days, not weeks.
Indexing and re-crawl speed: how fast Perplexity actually sees new work
The freshness advantage only exists if Perplexity's crawler picks up your update. Two speeds matter here:
- New domains: typically 7–14 days before pages become eligible for citation, sometimes longer if the domain has no external signal footprint (no Wikipedia entity, no mentions in the training corpus, no inbound links from already-crawled sources).
- Established domains: PerplexityBot re-crawls on a rolling basis, often within 24–72 hours for pages on domains it already trusts. Well-structured content on authoritative domains can appear in live Perplexity responses within 24–48 hours of publication.
If you're seeing slower discovery than this, the usual culprits are robots.txt blocks on PerplexityBot, aggressive edge caching serving stale HTML to crawlers, or client-side rendering that hides the update from bots — issues I've covered in SSR vs CSR for AI crawlers and the robots.txt AI crawler audit.
Timestamp signals that Perplexity reads
Perplexity's retrieval stack pulls freshness signals from multiple places. Ranked roughly by influence:
<meta property="article:modified_time">and<meta property="article:published_time">— the most reliable machine-readable signals.- Schema.org
dateModifiedanddatePublishedonArticle,NewsArticle, orBlogPostingtypes. Keep these honest; mismatches between schema and visible dates get discounted. - Visible "Last updated" text near the title or byline. Human-readable dates get parsed and, in my testing, appear to carry independent weight beyond the schema value.
- XML sitemap
<lastmod>— used for crawl prioritization more than ranking, but it triggers the re-crawl that lets the other signals matter. - HTTP
Last-Modifiedheader — mostly relevant for conditional GETs, but consistent values across all four signals reduce ambiguity.
The signal that does not work: changing dateModified without changing the content. Perplexity's diff-aware crawling means cosmetic bumps get ignored, and repeated abuse can suppress the domain's freshness score. Rewrite at least one substantive section — a stat, a paragraph of analysis, a new example — when you update the timestamp.
The refresh cadence that keeps you inside the window
For evergreen commercial content, a 30–45 day refresh cycle is the sweet spot. That cadence keeps your dateModified inside the 13-week window with two full re-index cycles of buffer, without burning editorial capacity on pages that don't need it.
A workable prioritization:
- High-value pages (top 20% by citation potential): refresh every 30 days. Update stats, add one new subsection, verify every claim.
- Mid-tier pages: 60–90 days. Focus on stat updates and fixing anything that's aged badly.
- Long-tail pages: 6 months, or when a substantive change is warranted.
- Time-sensitive pages (pricing, comparisons, "best X in [year]"): review monthly at minimum; refresh immediately when a competitor moves.
The payoff is measurable. The Ahrefs data showed recently updated pages average 6 AI citations vs. 3.6 for outdated pages — a 67% lift attributable specifically to freshness, not new content creation.
What to stop doing
- Stop backdating updates without editing. It's detectable and it suppresses trust.
- Stop treating "published date" as immutable. If you materially rewrite a piece, update both
datePublishedanddateModified, or move to an updated-date-forward display. - Stop refreshing everything on the same cadence. A stale glossary entry costs you nothing; a stale comparison page costs you citations weekly.
- Stop shipping updates without pinging your sitemap.
<lastmod>changes plus a fresh sitemap submission cut re-crawl latency noticeably.
FAQ
How quickly can a new page appear in Perplexity citations?
On an established domain with clean rendering and good schema, 24–48 hours is realistic. New domains typically wait 7–14 days before their pages enter Perplexity's eligible pool.
Does updating dateModified alone improve citations?
No. Perplexity's crawler compares content between visits, and timestamp changes without meaningful content changes are discounted. Rewrite at least one substantive section per refresh.
What's different about Perplexity's freshness bias vs. ChatGPT or Gemini?
Perplexity retrieves live on nearly every query, so freshness compounds directly into citation odds. ChatGPT leans more on its training corpus and structured retrieval, and Gemini blends AI Overview signals — both weight freshness less aggressively than Perplexity does.
