Analytics

Search Console's November Changes: New API Filters, a Google-Agent Guide, and a Tool Shutdown

Google Search Console changelog logged four changes in November 2026, and they point in different directions. On November 16, the Search Analytics API gained a branded-queries filter, a web multimodal type, and multi-country and multi-device filters. On November 17, a new guide on allowing Google-Agent traffic joined the crawling documentation. On November 18, the Data Feed validation tool was turned down without a named replacement. And on November 12, Google added documentation on identifying Google Search traffic by the Referer header. Separately, a logging error cut Generative AI report impressions for November 9 to 11.

Four changes, one changelog

November was a busy month for Search Console documentation, even though none of the four changes arrived with a major announcement. The Search Analytics API update came through a Google Search Central blog post. The Google-Agent guide appeared in the crawling documentation. The Data Feed validation tool shutdown surfaced through a rewritten help page rather than a blog post. The Referer header documentation was added quietly to the existing pages. Together they form a picture of a product that is still being actively developed, but in small increments rather than big releases.

The mix matters because the changes pull in opposite directions. The API additions give teams more ways to slice their search data, which is an expansion of what Search Console can do. The Data Feed validation tool shutdown takes something away, and it does so without naming a replacement. The Google-Agent guide and the Referer documentation are clarifications: they do not add features, they explain how Google systems already behave. A month that contains all three kinds of change is worth reading as a whole, not as four separate notes.

The Search Analytics API gets branded-query filters

On November 16, 2026, the Search Analytics API gained a branded-queries filter, announced in a Google Search Central blog post by Daniel Waisberg, Nati Yosephian, and Karin Gadasi. The filter lets API users separate queries that mention a brand from queries that do not, which is one of the most common splits in search reporting. Until now, teams that wanted that split had to build it themselves, usually by matching query strings against a list of brand terms and hoping the list was complete.

The value of a branded filter is not in the filter itself but in what it makes possible. Brand and non-brand traffic behave differently, respond to different work, and matter to different stakeholders. A site can grow its non-brand traffic while its brand traffic stays flat, and that is a very different story from a site whose total traffic grew because more people searched for its name. Separating the two is the first step in any honest search report, and having Google do the separation in the API removes a whole layer of guesswork from the process.

There is also a practical benefit for anyone who builds reports on the API. A branded-queries filter defined by Google is consistent across properties and across time, which means two teams pulling the same data get the same split. A hand-built brand list is never consistent: one analyst includes product names, another does not, and the numbers drift. The new filter does not solve every reporting argument, but it removes one of the most common sources of disagreement.

Multi-country and multi-device filters

The same November 16 update added multi-country and multi-device filters to the Search Analytics API. Previously, pulling data for several countries or several devices meant making separate requests and combining the results, which was slow and error-prone. The new filters let a single request cover more than one country and more than one device, which simplifies the code that sits on top of the API and reduces the number of calls a reporting system has to make.

For teams that report across markets, the multi-country filter is the more useful of the two. A site that operates in several countries can now pull a single view of its performance across all of them, and then break it down by country in the same dataset, rather than stitching together separate exports. The multi-device filter does the same for desktop, mobile, and tablet, which matters because search behavior differs sharply by device, and because a change that looks like a traffic problem is sometimes just a shift between devices.

The filters also make automated reporting more reliable. Every separate API request is a chance for something to go wrong: a timeout, a partial response, or a mismatch in the date range that only shows up when the numbers are combined. Fewer requests means fewer failure points, and it means the data in a dashboard was pulled in one pass rather than assembled from several. That is a small improvement in engineering terms, but it is the kind of improvement that shows up as fewer broken reports and fewer late-night fixes.

A web multimodal type joins the API

The November 16 update also added a web multimodal type to the Search Analytics API. The type covers search experiences that combine more than one mode, such as text and images together, and its arrival in the API reflects the way Google search results have been changing. Search is no longer a list of blue links with the occasional image result. It is a mix of formats, and the API is catching up to that reality by letting teams query the mixed formats as a category of their own.

For most teams, the web multimodal type will not change daily reporting right away. It is a new dimension to explore, not a replacement for the dimensions that already exist. Its value is in the questions it lets a team ask: how much of our traffic comes through multimodal results, and is that share growing? Those are the questions that matter when a search feature is new, because the teams that measure it early are the ones that notice when it starts to matter.

There is a longer-term point here as well. Every time Google adds a new search type to the API, it is signaling that the type is established enough to be measured, and that it expects teams to start tracking it. The web multimodal type is the latest example of that pattern. It does not require any action today, but it is worth adding to the list of dimensions a team checks when it reviews its API setup, so that the data is there when the questions about multimodal traffic become urgent rather than speculative.

A guide for Google-Agent traffic

On November 17, 2026, a new guide called Allow Google-Agent traffic was added to the crawling documentation. The guide explains how site owners should handle traffic from Google-Agent, which is the crawler Google uses for its AI features. The addition of a dedicated guide is a sign that Google-Agent traffic has become common enough, and confusing enough, that it needs its own documentation rather than a mention inside a broader page.

The guide matters because Google-Agent is not the same as Googlebot, and treating the two as interchangeable causes problems. Googlebot crawls for search indexing. Google-Agent crawls for AI features, and the two crawlers have different purposes, different behaviors, and different implications for a site. A site owner who blocks Google-Agent thinking it is a scraper may also be blocking the traffic that feeds AI answers about their content. A site owner who allows it without understanding it may be opening the door to more crawling than they expected. The guide exists to make that choice an informed one.

The timing is also worth noting. The guide arrived on November 17, one day after the API update and one day before the Data Feed tool shutdown, which makes it part of a cluster of documentation changes rather than a standalone event. November has been a month of small, steady updates to the Search Central documentation, and the Google-Agent guide is the one that most directly affects how sites interact with Google AI systems.

The Data Feed validation tool is gone

On November 18, 2026, the Data Feed validation tool was turned down and is no longer available. The tool let site owners test a structured data feed, most commonly a book feed, before submitting it, and its disappearance was not announced in a blog post. It surfaced through a rewritten help page, where the section that used to say Test your feed with the Data Feed validation tool was replaced with a section called Check your feed for common errors.

The quietness of the shutdown is the story. A tool that people relied on for validation is gone, and the only public record of its departure is the absence of its name from a help page. Google did not say when the tool closed, and it did not name a replacement. For the sites that used it, the practical effect is that a step in their publishing workflow has disappeared, and they have to find another way to check their feeds before they go live.

What replaces the tool: three checks

The replacement section, Check your feed for common errors, lists three checks that site owners should run on their feeds. The first is the spelling of the @type field. The second is the @context value. The third is setting the @context for both ReadAction and BorrowAction. These are the most common errors in book feeds, and the help page now presents them as a manual checklist rather than something a tool verifies for you.

A checklist is not a validator. A tool catches errors a person misses, and it catches them before the feed goes live. A checklist relies on the person running it, and it relies on them running it carefully every time. The three checks are useful, and they cover the errors that cause the most failures, but they shift the work from Google systems to the site owner. For a small publisher with one feed, that shift is manageable. For a publisher with many feeds, it is a new source of risk.

The open question is whether a replacement is coming. Google did not say when the tool closed, and it did not name a replacement, which leaves two possibilities: the tool is gone for good, and the checklist is the new workflow, or a replacement is being built and has not been announced yet. Until Google says something, the safe assumption is the first one, and the practical move is to build the three checks into the publishing process rather than waiting for a tool that may never arrive.

The Referer header, explained

On November 12, 2026, Google added new documentation on identifying Google Search traffic by Referer header. The documentation explains that Google Search uses the origin referrer policy, so the Referer header contains only the scheme and host of the referring page, without the URL path or query parameters. In plain terms, a site can see that a visitor came from Google Search, but it cannot see which search results page they came from, or what they searched for, from the header alone.

The documentation is a clarification, not a change in behavior. Google Search has used the origin referrer policy for some time, and the new pages explain what that means for anyone trying to identify search traffic from server logs. The practical consequence is that log-based analysis of Google traffic is limited by design: the header tells you the visit came from Google, and nothing more. Teams that need query-level detail have to get it from Search Console, not from their own logs, which is exactly the boundary Google intends.

The addition is also useful for the teams that maintain analytics pipelines, because it settles a question that comes up often in log analysis: why does the Referer header from Google contain so little? The answer is the referrer policy, and the documentation now states it plainly. That clarity saves time for anyone debugging a traffic source report, and it prevents the common mistake of treating an empty path as a logging error when it is actually the intended behavior.

The Generative AI report logging error

Separately from the changelog entries, a logging error cut Generative AI report impressions for November 9 to 11, 2026. The error meant that impressions in the Generative AI report were undercounted for those three days, and anyone looking at the report for that period would see a dip that did not reflect a real change in traffic. Google confirmed the error, and the affected dates are November 9, 10, and 11.

The error matters because it overlaps with the start of the November core update rollout, which began on November 10. A team looking at its Generative AI report for the week of November 9 would see a dip that coincides with the update, and the natural conclusion would be that the update caused it. The real cause was the logging error. That overlap is a reminder to check for known data issues before attributing a change to an algorithm update, especially when the dates line up neatly.

Why a quiet month matters

None of the four November changes is dramatic on its own. An API filter, a documentation guide, a retired tool, and a referrer policy note do not make headlines. But together they show the direction Search Console is moving: more granular data through the API, more documentation for AI-related crawling, and a willingness to retire tools that no longer fit. The direction is more important than any single change, because it tells teams where to invest their attention.

The retirement is the change that deserves the most attention, because it is the only one that removes capability. API filters can be adopted gradually, and documentation can be read when it is needed, but a retired validation tool changes a workflow immediately. The teams that used the Data Feed validation tool have already had to adapt, whether they noticed the shutdown or not. The teams that did not use it should still note the pattern: Google retires Search Console features without a blog post, and the first sign is often a rewritten help page.

The pattern is worth watching because it is likely to repeat. Search Console has a long history of retiring features quietly, from old reports to old tools, and the changelog is not always the first place the retirement appears. The practical habit is to check the tools you depend on, not just the changelog, and to treat a rewritten help page as news in its own right. A team that notices a retirement on the day it happens has time to adapt. A team that notices it months later, when a feed fails, does not.

What to check in your own account

The practical takeaway from November is a short checklist. If you build reports on the Search Analytics API, look at the branded-queries filter and the multi-country and multi-device filters, and decide whether they replace any of the manual splitting your reporting does today. If you have questions about Google-Agent traffic, read the new guide before changing any crawler rules. If you publish book feeds, build the three manual checks into your process now, because the validation tool is not coming back on the evidence available.

And if you looked at your Generative AI report for November 9 to 11 and saw a dip, check it again with the logging error in mind. The impressions for those three days were undercounted, and the dip is not a real change in how your pages performed in AI results. The November changelog is a small set of changes, but the logging error is the one that most easily leads to a wrong conclusion, and it is the one most worth remembering when the month is reviewed.

Sources

Sources: Google Search Central (November 16-18, 2026).

Related reading on this site: Google Confirms a Logging Error Behind the November 9–11 Generative AI Report Dip, Google Tag and Tag Manager Unification Begins Rolling Out, New Robots.txt Guide Separates AI Training Crawlers From Search and Human Visits, Who Blocks AI Crawlers? A Robots.txt Study of 1,000 Sites Has Answers, 32 AI Search Studies Reviewed: Citations Without Brand Mentions Are the Norm, and November Core Update Volatility Stays Muted as Rollout Nears Its Second Week.

More from Linkbaits

Turn this playbook into links

Linkbaits helps you create content that earns citations and backlinks. Start with the toolkit and build your first linkable asset this week.

Start free trial