Why Does Perplexity Still Cite Pages or Reviews That Were Deleted or Corrected?

Why does old content survive?

• Stale index entries. Perplexity keeps its own index, built by its crawler. Until a page is recrawled, the old version can be retrieved and cited.
• Copies elsewhere. Syndicated articles, news aggregators, review aggregators, scraper sites and web archives all keep versions of the original.
• Soft deletions. A page that shows "This page no longer exists" but returns a success status looks live to crawlers.
• Corrections that are easy to miss. A correction note at the bottom of an article leaves the old claim in the headline and opening lines, which are what summaries draw on.
• Model memory. Answers written without fresh retrieval can repeat what the model learned in training.

How Perplexity retrieves sources in real time is covered in how Perplexity retrieves and cites sources in real time.

Why does Perplexity cite an archived article after we published a correction?

Usually because it cites a copy rather than your corrected page. That copy might be a web archive snapshot, a syndicated version on a partner site, or an aggregator's copy. Or your correction sits at the bottom while the headline still carries the error.

To fix it:

1. Put the correction at the top of the article, and change the headline and opening if they contain the error. Note the date of the correction.
2. Ask syndication partners and aggregators to update their copies, or to remove them and link to the original.
3. For web archive copies, the Internet Archive accepts exclusion requests from site owners by email. Give the URLs, the time period and your connection to the site. It makes no guarantee about the outcome.
4. Recheck the answers after Perplexity has had time to recrawl, and send feedback on answers that still cite the old copy.

Do deleted Google reviews remain available to AI search engines?

Not from Google's own listing once they are removed, but often elsewhere. When a review is removed from Google Maps, it disappears from your Business Profile. Copies may remain on sites that scraped or aggregated reviews, in screenshots shared on social media or in articles, and in the memory of models trained while it was public. Answers built from Google's live data should reflect the removal once the change propagates. Answers built from other sources may not.

If a removed review keeps surfacing:

• Find the copy the answer cites, and contact that site to remove it.
• Report the answer to the assistant, explaining that the review was removed from the original platform.
• Publish a clear, factual response where the complaint is still visible, if it has been resolved.

Why does an AI engine cite a partner's cached page after the partnership ended?

Because the partner's page, or a copy of it, is still live or still in the engine's index. Common sources are a partner directory entry, a joint press release on a newswire, an integration page, or a co-branded landing page.

To fix it:

• Ask the partner to remove or update the page and return a 404 or 410 status, or redirect it to a neutral page. A page left live will keep being cited.
• Accept that press releases stay up. Publish your own dated statement that the partnership has ended, so assistants have a current source.
• Update your own site, removing partner badges, logos and integration pages, or marking them as discontinued.
• Use Google's tool for refreshing outdated content once the partner's live page has changed, so Google's copy updates.

Why does Perplexity cite a removed review in answers about my company?

The same mechanisms apply, but reviews spread especially widely. Review platforms syndicate content to partners, aggregators republish ratings, and complaint sites copy text. Open the source Perplexity cites. It is rarely the original platform. Ask that site to remove the copy, pointing out that the original was removed, and include the platform's removal notice if you have one. If the copied review is defamatory, legal routes exist, but take advice. Then report the Perplexity answer with the cited URL and an explanation. Whether to count citations that do not support their claims is covered in whether to count a Perplexity citation when the linked page does not support the claim.

How do HTTP 410 responses affect old citations in ChatGPT and Gemini?

A 410 status tells crawlers a page is gone on purpose. For Google, it makes little difference whether you use 404 or 410. Google's documentation says all 4xx errors except 429 are treated the same: URLs that return them are not indexed, and indexed URLs that start returning them are removed. Gemini and Google's AI features draw on that index, so the page drops out once it is recrawled.

ChatGPT's search draws on its own crawling and on search indexes, which also drop pages that return errors after recrawling. What neither status code does is change model memory. Answers written without a live search can still mention the old content until newer models are trained.

Practical rules:

• Use a 301 redirect when there is a replacement page, such as a corrected version or a successor product.
• Use 404 or 410 when the content should disappear entirely.
• Never return a success code with a "not found" message, which Google calls a soft 404 and other crawlers may treat as a live page.
• Remove the URL from your sitemap and internal links, so crawlers do not keep being sent to it.

How do you check whether old content is still being used?

Run the questions where the old content appeared across Perplexity, ChatGPT, Gemini and Google's AI features, several times each, every week or two. Record whether the old claim appears and which source is cited. Classify each source as the original, a copy, an archive or none, since each needs a different fix. Stop when the claim is absent from all runs for two consecutive checks.

What do people ask about deleted content in AI answers?

How long does it take for deleted pages to stop being cited?

Often a few weeks once they are recrawled, longer for pages crawled rarely. Copies elsewhere keep them alive until they are removed too.

Should we block Perplexity's crawler from the deleted pages?

No. Crawlers need to fetch the page to see that it is gone. A block can leave the old version in the index longer.

Will a correction note fix the problem?

Only if it sits where summaries read: at the top, with a corrected headline and opening.