New Google materials for SEO: search traffic, CDN and “things to do”

Between 9 and 12 November 2026, Google published three materials on Search Central that matter for SEO. The first explains how to recognise search traffic by the HTTP Referer header. The second explains why Googlebot activity rises after moving to a CDN and how to avoid blocking it with security settings. The third extends the documentation of aggregator and supplier units in the European Economic Area to “things to do” queries. Below I describe each change and what it means in practice.

How do you recognise traffic from Google Search?

Traffic from Google Search is recognised by the HTTP Referer header, which the browser sends after a result is clicked. This is described on a new Google documentation page from 12 November 2026.

Google uses the origin referrer policy. The header therefore contains only the scheme and domain, without the path or URL parameters. The user's query cannot be read from the header alone.

Traffic from Discover has the same values as the surface on which the user saw it. Traffic from other services on these domains, such as Google Images, is labelled the same way.

The documentation does not single out traffic from AI Overviews or AI Mode. This means it cannot be separated from ordinary clicks based on the header alone.

The header can also be empty. This happens, among other cases, in private browsing, with some tracking blockers, after opening a result in a new tab in some browsers and when moving from a search widget to the browser.

Most analytics tools classify this traffic as organic Google traffic on their own. When it shows up as referral or direct traffic, Google advises checking your own channel rules, referral exclusions and filters. The free GA4 audit checks whether the GA4 tag on your site sends correct data.

Analytics data usually does not match clicks in Google Search Console 1:1. Google lists four main reasons:

• privacy settings, ad blockers and lack of tracking consent,
• client-side redirects that replace the Referer or strip parameters, e.g. gclid,
• different measurement methods: Search Console counts a click, while analytics counts a visit after the page loads,
• paid traffic, which Search Console does not include.

Why does Googlebot activity rise after moving to a CDN?

After moving to a CDN, Googlebot activity rises because the server responds faster, and Google reads this as a signal that it can crawl more. This is explained in a Google Search Central blog post from 11 November 2026 by Cherry Prommawin and Gary Illyes.

Google sets the so-called crawl capacity limit based on the health of the server. Slower responses lower the crawl rate. Fast, stable responses gradually raise it, as long as Google has demand for crawling the site.

What is the cold cache problem?

A new CDN starts with an empty cache. On the first request for each URL, the edge server has no copy of it and must fetch it from the origin server. Only subsequent requests for the same URL are served by the CDN itself.

Users usually visit a small group of popular pages, which quickly land in the cache. Crawlers do the opposite and systematically go through every URL on the site. Each URL not yet served therefore hits the origin server.

Google notes that crawlers do not speed up immediately, and many CDNs smooth out sudden traffic spikes. Even so, until the cache warms up, the origin server still carries the main load.

Why is blocking all bots a bad idea?

When they see a spike in automated traffic, administrators often switch on general bot protection in the CDN. Such a rule without exceptions treats Googlebot like any other bot. Googlebot cannot solve a CAPTCHA, so it lands on an error page.

To Google, such a site looks deliberately blocked or broken. If the block stays in place for longer, crawlers will reduce crawling, and pages and product offers will start disappearing from the results.

How do you move to a CDN safely?

Google proposes three steps:

• Allow verified bots. Most CDNs and WAF firewalls have a ready-made rule for known bots. Google's crawlers can be verified by their published IP ranges. Unwanted bots are better blocked selectively, e.g. in robots.txt.
• Warm up the cache before switching traffic. Fetch your most popular pages, pages for planned promotions and the most frequently crawled URLs through the CDN, at a controlled pace.
• Use 429 or 503 codes as an emergency brake. If the origin server still cannot cope, return a 429 or 503 code for a short time. Google recommends a few hours at most, 1–2 days maximum.

However, you should not leave Google's crawlers facing 403 error pages or CAPTCHA challenges. Google also covered the topic of the post in episode 5 of the video series “The Non-Canonical Cabinet”.

How does Google handle “things to do” queries in the EEA?

In the European Economic Area, queries about attractions and activities, i.e. “things to do”, can now trigger aggregator and supplier units. Google added this category to the documentation of both units on 9 November 2026.

Both units work only for users in the EEA. They now cover queries about hotels, flights, trains and long-distance buses, products, local businesses, and attractions and activities.

In the aggregator unit, results from the highest-rated supplier are expanded by default. A click leads to the aggregator's site. For “things to do” queries, the aggregator supplies data through the Local and things to do Point of Interest Feed.

In that feed Google recommends, among other things, accurate descriptions, photos without watermarks, genuine ratings, and up-to-date prices and availability. It advises against writing names in capital letters, emoji and promotional slogans in names.

For travel businesses in Poland, the change may mean new competition for a place in the results. Portals with tour and ticket offers can apply to the aggregator unit, and attractions themselves to the supplier unit.

What should you do after these changes?

Several concrete tasks for SEO teams and administrators follow from the three Google materials:

• check whether Google traffic lands in referral or direct traffic in your analytics because of your own channel rules,
• when analysing server logs, count only requests for landing pages, not images, CSS and scripts,
• before moving to a CDN, warm up the cache for the most important and most frequently crawled URLs,
• in the CDN and WAF, allow verified bots instead of switching on a block for all bots,
• after the migration, watch the crawl stats in Search Console,
• in the travel industry, check whether the business can join the aggregator or supplier unit in the EEA.

Frequently asked questions

What sources is this article based on?

This article is based solely on Google Search Central materials:

• “Identify Google Search traffic in your web analytics or server logs”, updated 12 November 2026,
• “Moving to a CDN without overwhelming your origin (or blocking Googlebot)”, a blog post from 11 November 2026,
• “Aggregator unit in Google Search”, updated 9 November 2026,
• “Supplier unit in Google Search”, updated 9 November 2026.