XML Sitemap Priority and Changefreq for AI Search Crawlers

TL;DR

Keep your sitemap minimal: plus an accurate . Drop or de-emphasize and . Ping IndexNow on every meaningful update so Bing and ChatGPT search index quickly. Validate canonical-only URLs, split sitemaps at 50,000 URLs / 50 MB, and submit via Search Console and Bing Webmaster Tools.

Field semantics

The sitemaps.org schema defines five elements per :

Gary Illyes confirmed Google "mostly ignores" because most sites lie about it (SE Roundtable, 2015). When accurate, however, it remains the single most useful sitemap field. Format as ISO 8601 (2026-05-03T12:00:00+00:00). Update only when the canonical content meaningfully changes.

and

Google openly states it ignores both. There is no public documentation that GPTBot, OAI-SearchBot, ClaudeBot, or PerplexityBot consult either. Bingbot honors them as weak hints. Setting 1.0 site-wide accomplishes nothing.

Canonical-only rule

Include only canonical URLs you want indexed. Excluding noindex, redirected, blocked, or duplicate URLs is mandatory — their presence weakens the trust of the entire sitemap (Google, 2014).

AI-crawler observed behavior

The practical conclusion: AI crawlers use sitemaps for URL discovery and rely on to decide when to refetch. Field-rich sitemaps with priority and changefreq do not improve AI-side crawl prioritization.

Sample sitemap (recommended shape)

<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url>
    <loc>https://example.com/article-a</loc>
    <lastmod>2026-05-03T09:00:00+00:00</lastmod>
  </url>
  <url>
    <loc>https://example.com/article-b</loc>
    <lastmod>2026-04-30T12:00:00+00:00</lastmod>
  </url>
</urlset>

For sites over 50,000 URLs, use a sitemap index pointing to per-section sitemaps (/sitemap-articles.xml, /sitemap-glossary.xml, etc.).

Sitemap freshness: pings and IndexNow

Google deprecated the /ping?sitemap=... HTTP endpoint in 2023. Today the recommended workflow is:

1. Google Search Console — submit each sitemap once; Google polls them periodically.
2. Bing Webmaster Tools — submit and enable IndexNow.
3. IndexNow — on every meaningful publish or update, ping https://api.indexnow.org/indexnow?url=...&key=.... Supported by Microsoft Bing, Naver, Seznam.cz, Yandex, and Yep (IndexNow.org, 2026; Bing, 2026).

Bing now recommends IndexNow over manual URL Submission as the primary path (Bing Webmaster Tools, 2026). Because ChatGPT Search is layered on Bing's index, IndexNow indirectly accelerates ChatGPT visibility — community write-ups consistently report faster appearance after IndexNow adoption.

IndexNow request shape

POST /indexnow HTTP/1.1
Host: api.indexnow.org
Content-Type: application/json

{

"host": "example.com",

"key": "abcdef1234567890abcdef1234567890",

"keyLocation": "https://example.com/abcdef1234567890abcdef1234567890.txt",

"urlList": [

"https://example.com/article-a",

"https://example.com/article-b"

]

}

Host the key file at the documented keyLocation. Most CMS and static-site tools (Next.js, Astro, Hugo plugins) ship IndexNow integrations.

Audit checklist

• [ ] Sitemap returns 200 with Content-Type: application/xml.
• [ ] All URLs are canonical, not redirected, not blocked.
• [ ] reflects the actual last meaningful change.
• [ ] and are absent or all set to a single benign value.
• [ ] Sitemap submitted in Google Search Console and Bing Webmaster Tools.
• [ ] IndexNow key file hosted; pings fire on publish.
• [ ] Sitemap index used if URL count exceeds 50,000.
• [ ] Each child sitemap is under 50 MB uncompressed.
• [ ] Sitemap: directive present in robots.txt.

Common mistakes

• Setting 1.0 on every URL.
• Bumping daily on unchanged content (it teaches crawlers to ignore it).
• Including URLs that 301 to elsewhere or return noindex.
• Forgetting to ping IndexNow on edits — the protocol is push-based.
• Mixing canonical and parameter-laden URLs in one sitemap.
• Relying on the deprecated Google ping endpoint.

FAQ

Q: Do AI search crawlers read XML sitemaps?

Yes. OpenAI, Anthropic, and Perplexity all use sitemap discovery for URL enumeration. They rely on and ; legacy fields are ignored.

Q: Should I remove and entirely?

You can. They have no documented positive effect for Google or for AI crawlers and they encourage drift. If you keep them, set them once based on your site structure and do not touch them again.

Q: How does IndexNow help ChatGPT visibility?

ChatGPT Search uses the Bing index. IndexNow accelerates Bing indexing of changed URLs, which in turn accelerates ChatGPT freshness. There is no direct IndexNow endpoint for OpenAI (IndexNow.org, 2026).

Q: What is the maximum sitemap size?

50,000 URLs and 50 MB uncompressed per the sitemaps.org spec. Use a sitemap index when you exceed either limit.

Q: Does the deprecated Google sitemap ping still work?

No. Google deprecated the /ping?sitemap=... endpoint in 2023. Submit sitemaps through Search Console and rely on lastmod for ongoing freshness.

Q: Should I submit RSS or Atom feeds in addition to a sitemap?

Yes for news and frequently updated content. Search engines and AI crawlers benefit from feeds for freshness signals while sitemaps anchor full URL coverage.