Google Search Documentation Changes, Oct 15 to Nov 12, 2026

Google's Search Central changelog lists six documentation changes between 16 October and 12 November 2026, and the most useful for most sites is a new page, added 12 November, explaining how to identify Google Search traffic in analytics using the HTTP Referer header. Google also announced a shared AI training control on 30 October. None is described as changing how Search ranks pages. (As of 12 November 2026.)

Key Takeaways

• On 12 November Google added a page on identifying traffic from Google Search using Referer headers. As fetched, it does not mention AI Mode.
• On 30 October Google said it will adopt a shared robots.txt "policy token" for AI training preferences, based on an IETF draft, with support planned for 2027 (Google).
• On 26 October the user-generated spam guidance was extended to tools that let users publish shared pages or web apps, and tells platforms to keep link-shared content out of the index by default (changelog).
• On 23 October the URL structure page clarified that hyphens are preferred over underscores, to match Google's video series. Google describes it as a clarification, not a new rule.
• On 16 October the Indexing API quota page was updated to show 200 publish requests per day per project.
• On 9 November the aggregator and supplier unit pages added support for "things to do" queries, which applies to Vertical Search Services in the EEA only.
• Documentation changes are not ranking updates. The core update that began on 10 November is separate (see my core update post).

The changelog, in one table

Dates come from Google's Search Central updates page, except 30 October, which comes from Google's blog post and Search Engine Roundtable's coverage. The updates page is long and I read its newest entries (October and November 2026) only, so treat this as a complete list of what I found, not a guarantee that nothing else changed.

The Referer page, 12 November

What it says

The page explains how to spot Google Search visitors in analytics or server logs. The browser usually sends a Referer header, and Google uses the origin referrer policy, so it contains only the scheme and host, with no path or search terms. For queries you still need the Search Console Performance report.

Values it lists:

• Google Search on the web: https://www.google.com/ or a local domain such as google.co.uk or google.co.jp.
• Google app on Android: either android-app://com.google.android.googlequicksearchbox/https/www.google.com or the web values.
• Google app on iOS: the web values.
• Discover: the same values as the surface where the user saw it, such as the Google app or Google Search.
• Other Google services on those origins, such as Google Images, send identical values, so they are included in matches.

Caveats Google lists

The header can be blank because of private browsing, privacy extensions or strict browser settings. Client-side redirects can replace the referrer with your own page, and redirects that drop query strings lose parameters such as gclid. Analytics counts visits after page load, while Search Console counts clicks and excludes Google Ads clicks, so numbers will differ.

What the page does not say

I could not find any mention of AI Mode or AI Overviews on the page as fetched. So I cannot tell you what referrer value AI Mode clicks send, and Google has not documented it there. Do not assume AI Mode traffic is separable from classic Search in your analytics. My post on measuring AI search traffic covers workarounds. Also note that Discover traffic sits inside these same values, so it cannot be split out using this page alone.

The shared AI training control, 30 October

Google said it will voluntarily adopt a proposal under discussion in the IETF's AIPREF working group. A site would add one shared policy token to robots.txt to state AI training preferences across participating providers, instead of listing each company. If your file uses the token to opt out of training and does not mention Google-Extended, Google will treat it as if Google-Extended were disallowed. An explicit Google-Extended setting takes precedence, and sites that allow Google-Extended can be eligible to appear in Gemini Apps responses and receive grounding clicks. Google plans support in 2027.

Search Engine Roundtable adds that the proposal is a work in progress, that one IETF participant said significant areas of dispute remain, and that a critic says the current draft exempts user-triggered actions. It also reports that Search Console's separate generative AI opt-out still exists (Roundtable). I explained that opt-out in Google's AI opt-out control.

My read: nothing to deploy yet, because Google is not honoring the token until 2027 and the standard is unsettled. Do review your robots.txt now so you know what you currently allow for training crawlers.

The user-generated spam update, 26 October

The prevention guide now covers "tools that let users publish shared pages, generated content, or web apps". It adds a section telling platforms to decide whether user-created pages are meant for Search. Pages made for sharing, collaboration or utility should default to noindex or a robots.txt block. It also advises verifying the site or subdomain in Search Console for manual action and security alerts, checking Page Indexing and Performance reports for spikes, and alerting on sudden jumps in newly published public pages or small groups of users producing many indexable URLs (Roundtable summary).

This follows a spam-heavy autumn: see the September spam update wrap-up. If you run a product where customers create public pages, this is the change I would read first.

Smaller changes

• Hyphens in URLs (23 October). Google's page recommends hyphens over underscores to separate words and warns against running words together. The changelog says the edit was made to synchronize with a video series. If your URLs already use hyphens, do nothing. Do not rewrite working URLs for this, because you would be creating redirects for no stated gain.
• Indexing API quota (16 October). The page shows 200 publish requests per day, 180 metadata requests per minute and 380 requests per minute combined, per project, with the daily quota resetting at midnight Pacific. It does not describe a process to request more, and does not say which content types the API is for.
• Control what you share (21 October). A tidy-up that lists the existing tools: robots meta tag, data-nosnippet, X-Robots-Tag and the Search generative AI control.
• Things to do (9 November). The aggregator unit is an EEA feature that shows results from vertical search services. It matters only if you are an approved service feeding Google data.

Changes from just before this window

Two earlier items are worth knowing about. On 6 October Search Console's Performance report added multi-select country filters (Roundtable), which I wrote up in the multi-country filter guide. On 1 October the generative AI content guide was updated with information from the rater guidelines, covered in the fact-check rule post.

What this means for marketers and founders

Do this week

1. Check your analytics channel rules against the new Referer page. If Google Search visits show as referral or direct in GA4, look at referral exclusions and custom channel groups first.
2. If you run logs, filter on landing-page requests only. Google notes that image, CSS and script requests carry your own site as referrer.
3. Open your robots.txt and list every AI crawler rule. Decide what you want in 2027 before the token exists.
4. If your product lets users publish pages, audit what is indexable. Compare newly indexed URLs in Page Indexing with newly created public pages.

Leave alone

• Underscore URLs that already rank. Google framed the hyphen page as a clarification.
• Any plan to "opt in" to the AI training token today. It is not supported yet.

What we do not know yet

• What referrer AI Mode clicks send, since the new page is silent on it.
• When in 2027 the shared AI training control becomes active, and whether IETF finalizes the draft as Google described it.
• Whether the Search Console AI performance reports will get click data. A Google representative said on 15 October they are working with site owners on which metrics to add.

FAQ

What did Google add to its documentation on 12 November 2026?

A new page on identifying traffic from Google Search using HTTP Referer headers. It lists the Referer values for web Search, the Android and iOS Google apps and Discover, and explains why analytics numbers differ from Search Console.

Does the new Referer page cover AI Mode traffic?

Not in the version I read. It does not mention AI Mode, so I cannot tell you what referrer it sends. Treat any claim that AI Mode traffic is cleanly separable in analytics as unverified until Google documents it.

What is Google's new AI training control?

It is a shared robots.txt policy token, based on an IETF draft, that lets a site state AI training preferences once for participating providers. Google announced on 30 October that it plans to support it in 2027, and Google-Extended keeps working alongside it.

Do I need to change robots.txt now for the AI training token?

No. Google is not honoring it until 2027 and the IETF draft is still disputed. Review your current AI crawler rules, but wait to add a token.

Did Google change how it treats hyphens versus underscores?

Google updated the page on 23 October to clarify that hyphens are preferred, and said the edit was to match its video series. I read no sign of a ranking change, so do not rewrite existing URLs for this.

What changed in Google's user-generated spam guidance?

On 26 October it was extended to tools that let users publish shared pages, generated content or web apps. It advises defaulting link-shared pages to noindex, verifying the site in Search Console and monitoring for spikes in newly published public pages.

Is the Indexing API quota page a new rule?

It is an update to show current limits: 200 publish requests per day per project, plus per-minute limits for metadata requests. It also removed an outdated form.

Want help with this?

I'm Younus Fardeen, an organic growth, SEO and AEO strategist with 4+ years of marketing experience across edtech and startup teams. If you want help auditing your analytics channel setup, robots.txt and indexation against Google's current documentation, see my work and reach me through the contact form at younusfardeen.com.