Should I delete or noindex old pages with outdated facts?

An engine that searches the web before answering picks sentences from the pages it retrieves. If your pricing page says $79 and a 2023 blog post says $49, both are candidates. You can't control which one gets picked. You can control whether the wrong one exists in a form the engine can fetch.

So the order of preference is:

1. Update the page when the topic is still useful. Replace the old fact with the current one, stated plainly. A specific, fetchable sentence (a price, a number, a one-line definition) gives the engine something correct to quote, and that is the only positive on-page signal in the evidence we work from.
2. Redirect with a 301 when a newer page fully replaces the old one. Crawlers that refetch the old URL land on the current page.
3. Noindex when the page has to stay live for people (a legal archive, an old announcement) but should not feed search or AI answers.
4. Delete only when the page has no value and nothing better to point to.

If the wrong fact is a price, the reasoning behind publishing current prices on a fetchable page applies directly.

The noindex tag goes in the page head:

<meta name="robots" content="noindex">

Google has to fetch the page to read that tag. If robots.txt blocks Googlebot from the URL, the tag is never seen. If you only want the page out of answers but still searchable, the softer controls compared in noindex vs nosnippet vs data-nosnippet may fit better.

Google states that a page must be "indexed and eligible to be shown in Google Search with a snippet" to appear as a supporting link in AI Overviews or AI Mode. A noindexed page fails that test. Google and Bing both document noindex, nosnippet, max-snippet:0 and data-nosnippet as the on-page controls for AI answer display. Bing says content marked data-nosnippet stays indexed but is excluded from snippets and AI summaries.

Three limits matter:

• Each engine has its own index. In Otterly's 2,000-post experiment (published 2026), Google deindexed both test sites and its AI surfaces almost stopped citing them. Bing kept them, and Microsoft Copilot kept citing them. Removing a page from one index says nothing about another, so check each engine separately.
• Training data is frozen. An answer given with no live search comes from what the model learned before its cutoff. No tag, redirect or deletion reaches it this quarter. If the wrong fact predates the cutoff, see correcting outdated facts past a knowledge cutoff.
• Other sites repeat you. A review site quoting your 2023 price is outside your noindex. When engines cite those sites, getting the page owner to update is the route, and old news stories work the same way.

Before choosing a fix, confirm the error really comes from an old page. Hallucination vs outdated answer covers how to tell the difference.

Changelogs and release notes

These are supposed to hold old facts. Keep them, but date every entry and link each one to the current state. If a changelog keeps surfacing in answers, data-nosnippet on the outdated block is narrower than noindexing the whole page.

Old blog posts with traffic

Update the figure in place and add a dated note saying what changed. Redirecting a post with its own audience throws that audience away.

Retired plans or products

Redirect to the plan or product that replaced them. If nothing replaced them, say so on a short current page, so the engine has a true sentence to quote instead of silence.

Rebrands

Old names live on in URLs, titles and image alt text. The same audit applies, and AI using your old company name covers the extra steps.

1. List every fact that changes. Prices, plan limits, feature names, headcount, locations, integrations.
2. Search your own site for each one. The pricing page is rarely the problem. Old posts, help articles, PDFs and footers usually are. The free brand fact consistency check flags pages that state the same fact differently.
3. Pick one action per page from the table above, and fix internal links that still point to retired URLs.
4. Check index status engine by engine in Google Search Console and Bing Webmaster Tools.
5. Recheck the answers. The free checker on this site shows what AI engines say for up to three buyer questions, which tells you whether the wrong fact still appears. It is a snapshot, so rerun it after crawlers have had time to revisit.

None of this guarantees an engine will quote the right figure. It removes the wrong one from the pages you control, which is the part you can fix.

Will noindexing old pages stop ChatGPT from quoting them? Not necessarily. Google and Bing document noindex as an on-page control for their own search and AI features. ChatGPT search uses OpenAI's crawler, OAI-SearchBot, and its own sources. An answer given without search comes from training data, which no tag changes. Updating or redirecting the page removes the wrong sentence for every crawler that refetches it, which is why it beats noindex for most cases.

Should I block outdated pages in robots.txt instead? Not if you want Google to drop them. A robots.txt block stops Googlebot from fetching the page, so it never sees a noindex tag on it. Google can still list a blocked URL it learns about from links. Robots.txt rules are also per bot, so a block for one crawler leaves the others untouched. Use noindex, a redirect or an update for outdated content.

How long until AI answers reflect the change? Nobody publishes a timeline. Searched answers can change once the engine recrawls the page, and that schedule belongs to each engine. Answers given from training data change only when a new model ships with a later cutoff. Check your index status in Search Console and Bing Webmaster Tools, then recheck the answers themselves rather than assuming a date.

Is a 404 or a 301 better for a retired pricing page? A 301 to the current pricing page is usually better. Anyone following an old link, human or crawler, lands on the right number, and you keep the links pointing at the old URL. Return a 404 or 410 only when nothing current covers the same topic. Either way, update any internal links that still point at the retired page.