Content freshness and AI search: how often B2B teams should update pages
Update B2B pages when their facts change, and review the pages that carry your AI visibility on a fixed schedule, rather than changing dates to look fresh. Ahrefs' analysis of about 17 million citations (vendor research) found AI-cited pages were 25.7% fresher than organic results, with ChatGPT showing the strongest preference and Google AI Overviews the weakest, and Microsoft advises regular updates so AI systems reference current content. A quarterly review of money pages and an IndexNow ping after every real change covers most teams.
Key takeaways
- Freshness helps, but cited content still averages almost three years old: substance first, recency second.
- ChatGPT and Perplexity show the strongest recency preference; Google AI Overviews the weakest.
- Only real changes count: new facts, prices, features, data or guidance. Date bumps without changes are a risk.
- Set a cadence by page type and tell engines about changes with sitemaps and IndexNow.
What the data says about freshness
Ahrefs analyzed about 17 million cited URLs across ChatGPT, Perplexity, Gemini, Copilot and Google AI Overviews, compared with organic Google results, and published the study in July 2025. It is vendor research. Cited URLs averaged 1,064 days since publication against 1,432 days for organic results, which the authors call 25.7% fresher. Measured on last-updated dates, the gap was smaller, at 13.1%.
The engines differ. ChatGPT showed the strongest preference for newer content, and both ChatGPT and Perplexity appeared to order their references from newest to oldest. Google AI Overviews showed the least preference: their top citations were on average slightly older than the organic results.
The authors add the caveat that matters most: cited content still averages about 2.9 years old, and frequently updated low-quality content will not move AI visibility. Freshness is a tie-breaker between useful pages, not a substitute for one.
Which engines care most, and what that means
The same page can be fresh enough for one engine and stale for another. Plan for the strictest engine your buyers use.
| Engine | What the evidence shows | Implication for B2B pages |
|---|---|---|
| ChatGPT | Strongest preference for newer content in the Ahrefs study; references appear ordered newest first | Keep cited articles and comparison pages current; outdated facts get replaced by newer third-party pages |
| Perplexity | References also appear ordered by recency; its retrieval prefilter removes stale content | Date facts that change, such as pricing, limits and integrations |
| Microsoft Copilot and Bing | Microsoft recommends regular updates and IndexNow so AI references current versions | Connect IndexNow and keep sitemap lastmod accurate |
| Google AI Overviews | Least freshness preference of the engines studied | Usefulness and eligibility matter more than recency |
For most B2B companies, that means the pages that win ChatGPT and Perplexity answers, usually comparisons, pricing and integration pages, need the tightest review cycle.
What the engines say
Microsoft is the most explicit. Its February 2026 announcement of AI Performance in Bing Webmaster Tools recommends regular updates so AI systems reference the most current version of your content, and points to IndexNow for telling search engines when pages are added, updated or removed. Fabrice Canel of Microsoft put it more directly at SMX Munich: generative AI values fresh content, partly as a reference check of training data, and IndexNow pushes changes as they are published.
Google is less focused on freshness in its guidance for generative AI features, which emphasizes unique, useful content. Its documentation on dates asks for visible, labelled publication or update dates, consistent with datePublished and dateModified in structured data, and says the dates must describe the page itself.
Perplexity is worth a separate note. Its research team describes a prefilter that removes non-responsive or stale content before ranking, which means an outdated page may never reach the stage where quality is judged.
Get an AI visibility read across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews.
What counts as a real update
We use this rule: if a buyer would act differently after reading the new version, it is an update.
| Change | Real update? | Change dateModified? |
|---|---|---|
| New price, plan, limit or feature | Yes | Yes |
| New data, benchmark or source replacing an old one | Yes | Yes |
| New section answering a buyer question | Yes | Yes |
| Corrected wrong fact | Yes | Yes |
| Typo fixes, new image, design change | No | No |
| Changing the year in the title with no other change | No | No, and do not do it |
A refresh cadence by page type
Not every page needs the same schedule. Tie the review to how fast the facts behind it change.
| Page type | Review cadence | Trigger for an immediate update |
|---|---|---|
| Pricing and packaging | Quarterly | Any price, plan or limit change |
| Product and integration pages | Quarterly | Release, deprecation or new integration |
| Comparison and alternatives pages | Quarterly | Competitor pricing, product or positioning change |
| Documentation | With each release | Behavior change or deprecation |
| Articles and guides that carry citations | Every six months | Newer data or a platform change |
| Case studies | Yearly | Client relationship or result changes |
Show freshness honestly on the page
Engines and buyers both need to see what changed and when. Google asks for a visible, labelled date such as "Last updated" that matches dateModified in your structured data, and for dates that describe the page rather than the events in it.
On pages with facts that change, such as pricing, limits or integrations, we also date the facts themselves: "Supported regions as of November 2026" tells a reader and a model exactly how current the statement is. For articles built on research, replace old figures with newer ones from the same or better sources, and say which year the data covers.
If a page is no longer accurate and you cannot update it, redirect it to the page that is, rather than leaving an old answer for engines to find.
Run the refresh as a process
Freshness fails when nobody owns it. This is the process we set up with clients.
- List the pages that carry your AI visibility from the Search Console generative AI report, Bing AI Performance cited pages and your prompt set citations.
- Assign an owner and a review date to each page, using the cadence table.
- Keep a short change log on each page or in the CMS: what changed and why.
- When a page changes materially, update the visible date and dateModified together.
- Update the sitemap lastmod and send an IndexNow ping, so Bing and other participating engines learn about the change.
- Re-run the relevant prompts four weeks later and record whether citations or descriptions changed.
Mistakes that make freshness backfire
These patterns look like freshness and undermine trust.
- Bumping dates site-wide with no content change.
- Adding next year to titles before anything in the article reflects it.
- Rewriting a page that is already cited and losing the passage that earned the citation.
- Deleting old articles instead of updating or redirecting them, which breaks the URLs engines cite.
- Leaving retired products and prices live on old pages that still get cited.
Where to start
Start with the 20 pages that carry the most AI visibility and the pricing and product pages behind your pipeline. Review them now, set the cadence and connect IndexNow.
Our generative engine optimization service builds the refresh calendar into the monthly program.
Our B2B SEO team handles sitemaps, IndexNow and dates as part of technical SEO.
If AI answers still describe last year's product, talk to our team.
Frequently asked questions
Does content freshness matter for AI search?
It helps as a tie-breaker. Ahrefs (vendor research) found AI-cited URLs were 25.7% fresher than organic results, with the strongest preference in ChatGPT, but cited content still averaged about 2.9 years old.
How often should we update B2B content for AI search?
Update when facts change, and review pricing, product, integration and comparison pages quarterly, cited articles every six months and case studies yearly.
Should we change the published date when we update?
Keep the original publication date and update a labelled last-updated date and dateModified when the change is material. Google asks for dates that describe the page and match the structured data.
What is IndexNow and does it help AI visibility?
IndexNow is a protocol that tells participating search engines, including Microsoft Bing, when URLs are added, updated or deleted. Microsoft recommends it so AI systems reference the current version of your content.
Does Google AI Mode prefer fresh content?
The Ahrefs study did not measure AI Mode separately, and found Google AI Overviews showed the least freshness preference of the engines studied. Google's guidance emphasizes unique, useful content over recency.
Sources
- Ahrefs (vendor study): AI assistants prefer to cite fresher content, 17 million citations (July 28, 2025)
- Bing Webmaster Blog: Introducing AI Performance in Bing Webmaster Tools, public preview (February 10, 2026)
- Search Engine Land: Microsoft Bing and Copilot use schema for their LLMs (Fabrice Canel, SMX Munich)
- IndexNow.org: How IndexNow works and participating search engines
- Google Search Central: Influencing your byline dates in Google Search
- Google Search Central: Google's guide to optimizing for generative AI features on Google Search (updated July 10, 2026)
Related service: Generative engine optimization. See also pricing and client results.
We run enterprise B2B demand generation for teams in Montreal, Toronto, Ottawa and the United States.
