Does Perplexity weight recency more heavily than ChatGPT does when deciding which product page to cite, and should that change how often we update content?
Perplexity refreshes faster than ChatGPT on user-initiated queries, but neither company publishes a recency weight. Perplexity-User fetches pages live at query time and generally ignores robots.txt (per Perplexity's bot documentation, September 2026), while OpenAI notes robots.txt changes take about 24 hours to register (per OpenAI, September 2026). Refresh cadence should follow product truth, not engine folklore.
What is documented, and what isn't
| Documented | Not documented | |
|---|---|---|
| Perplexity | PerplexityBot surfaces sites in search results and is not used for foundation-model training; Perplexity-User fetches on demand and generally ignores robots.txt (per Perplexity, Sep 2026) | Any recency coefficient, crawl interval, or freshness score |
| ChatGPT | OAI-SearchBot powers ChatGPT search; blocking it removes you from search answers; robots.txt changes take ~24h (per OpenAI, Sep 2026) | Any recency coefficient or recrawl cadence |
Every published "Perplexity weights freshness 2x" figure traces back to third-party observation, not to either vendor. We are not going to restate one as though it were a specification.
What the architecture does tell you
A live fetch at query time reflects the page as it is right now. An index built by a scheduled crawler reflects the page as it was at last crawl. That difference is real and structural, and it means a same-day price change is visible to a Perplexity user-initiated query before it is visible to a ChatGPT search answer. It does not mean Perplexity prefers fresh pages over good ones.
The verdict, and who it's for. Do not build a content-refresh cadence around an engine. Build it around facts that expire: pricing, stock, compatibility, policy. A merchant selling 400 SKUs with seasonal ranges needs a quarterly review of dated claims and a same-day path for price and availability, which is what Shopify Catalog already does automatically for price and availability (per Shopify's Help Center, September 2026).
When cosmetic refreshing actively hurts: editing a page's date without editing its substance. That practice degrades the one signal (a visible, truthful last-updated date) that makes a dated claim trustworthy in the first place.
The Deploi point of view
Our own position, from building on Shopify. Separate from the facts above.
- Our take: Refresh cadence is a data-hygiene question wearing an AI costume. The stores that win on freshness are the ones whose facts live in metafields and update once, everywhere.
- What we’ve seen: Across 18 metafield-driven PDP builds, the pages that go stale are the ones with facts typed into rich-text fields. A spec written into a paragraph has no owner and no expiry; the same spec in a metafield shows up in a report when it's empty.
- Where we disagree: The AI-visibility tooling market sells refresh frequency as a ranking lever and charges monthly for it. We have seen no vendor documentation supporting a recency weight, and we will not price work against a number nobody publishes.
- What this page adds: that the Perplexity/ChatGPT difference is a live fetch versus a scheduled index, and that no recency weighting is published by either.
Reviewed by Martin Dejnicki, Director of SEO & AI Search. Facts verified 2026-09-13.