Week 38/2026: Google pays publishers for AI, AI Overviews take clicks, and Cloudflare lets you block AI training
Google has started paying publishers for content that genuinely shapes AI answers, but won't say how it judges that. The same week, another study showed that AI Overviews take roughly one in four clicks from heavily exposed sites.
The week in three sentences
For the first time, Google is paying publishers because their content helps build AI answers. At the same time, new data shows AI Overviews keep taking clicks, and a strong Google ranking does not automatically mean visibility in AI chats. Site owners also got better tools to block AI training without dropping out of search.
Google will pay for content that builds AI answers
Signal: 3 independent sources this week, 6 in 30 days; tracked since 2024-07-24.
Google has launched a pilot programme called AI Contribution in Search Console. Selected publishers, mostly small and mid-sized ones, are paid for content that clearly helps generate answers in Gemini, AI Overviews and AI Mode.
The catch is that a link under the answer does not count. The content has to “significantly contribute” to the answer, and the criteria are not public (Search Engine Journal).
What to do: check Search Console to see whether you have access to the programme, and write pieces with your own data that AI cannot find elsewhere.
Sources:
- Search Engine Journal, 2026-09-20: Google's AI Payment Pilot Vs. Cloudflare & Microsoft Models
- Search Engine Journal, 2026-09-17: Google Search Profile Badges Expose The Publisher Traffic Crisis
- Search Engine Journal, 2026-09-17: Beyond Content Parity: Building A Validated Content Workflow For AI Search
AI Overviews take one in four clicks from the most visible sites
Signal: 3 independent sources this week, 7 in 30 days; tracked since 2026-07-08.
Another study has measured how many clicks AI Overviews take away. Compared with a control group, domains that are heavily featured in them lose roughly one click in four (Ahrefs).
Other data points the same way: when an AI summary appears, people click a regular result almost half as often (Pew Research). In ads, this pushes down CTR (the share of people who click) on informational queries.
What to do: give your page something AI cannot repeat in two sentences, such as your own research, data or a tool.
Sources:
- Ahrefs Blog, 2026-09-17: AI Overviews Cut CTR by 23.1% in France
- Search Engine Land, 2026-09-17: The follow-up query: Rethinking SEO for conversational search
- Search Engine Land, 2026-09-17: AI Overviews and paid search: When to fight, influence, or generate demand
Brands that lead on Google often disappear in AI chats
Signal: 3 independent sources this week, 5 in 30 days; tracked since 2025-11-17.
A strong Google ranking does not guarantee a place in AI answers. In a study of 12,000 AI prompts, more than 40% of brands that dominate Google did not appear in chats at all (State of AI Search 2026, Delante).
It works the other way too: some brands that are highly visible in AI have only average Google rankings. Brands that tell their story through images lose out the most, because AI bots only read text.
What to do: check your brand in several AI chats separately, and do not assume your Google results carry over.
Sources:
- Delante, 2026-09-20: Jak zbudować benchmark konkurencji w AI Search?
- Ahrefs Blog, 2026-09-17: 8 AI Marketing Trends I've Seen Firsthand in 2026 (Backed by Data)
AI-written text can rank first
Signal: 3 independent sources this week, 11 in 30 days; tracked since 2025-11-17.
Ahrefs analysed a million pages from search results. AI-generated content appears in every position, including first place.
Pages that rely more heavily on AI rank slightly lower, but the researchers see this as a quality effect, not a Google penalty. Using AI is not the problem; text that adds nothing is.
What to do: use AI for writing, but add knowledge, examples and the perspective of someone who knows the subject to every piece.
Sources:
- Ahrefs Blog, 2026-09-17: 8 AI Marketing Trends I've Seen Firsthand in 2026 (Backed by Data)
Block AI training in Cloudflare and stay in Google
New signal: one source so far.
Cloudflare has separated blocking AI training from blocking search engines. The “Disallow AI Training” setting cuts off bots that collect content to train models, but still lets Googlebot in for search.
Watch out for a trap: the “Block” option stops Googlebot entirely, including for regular results. Presence in AI Overviews and AI Mode is controlled separately, in Search Console (Search Engine Journal).
What to do: if you block AI training in Cloudflare, check that you picked “Disallow AI Training”, not “Block”.
Sources:
- Search Engine Journal, 2026-09-20: Google's AI Payment Pilot Vs. Cloudflare & Microsoft Models
- Search Engine Journal, 2026-09-18: Cloudflare Will Write Your Robots.txt, And It Has A Point
- Search Engine Journal, 2026-09-16: Cloudflare Lets Sites Disallow AI Training Without Blocking Googlebot
Google will measure how ads hurt page experience
Signal: 2 independent sources this week, 2 in 30 days; tracked since 2022-10-06.
Google has added four experimental ad metrics to CrUX, a public dataset on how real Chrome users experience websites.
The new metrics show how ads affect a page's usability. For sites that rely on ad revenue, this is the first hard measure of whether ads get in readers' way (Search Engine Land).
What to do: if you run ads on your site, look at the new CrUX data and compare pages with many ad slots to pages with few.
Sources:
- Search Engine Journal, 2026-09-17: Google Adds New Ad Experience Metrics To CrUX Report
- Search Engine Land, 2026-09-17: Google Chrome launches new metrics to measure ad-heavy websites
Local Services Ads are moving into Google Ads
Signal: 2 independent sources this week, 3 in 30 days; tracked since 2026-07-24.
Local Services Ads campaigns are moving into Google Ads. They will become a specialised version of Performance Max where you pay for a lead - a contact from a customer.
Weekly budgets become daily, and manual bids per service category give way to automated bidding towards a target cost per acquisition. The ads are expected to stay where they were, in Search and Maps, rather than spreading to every Performance Max placement (Search Engine Journal).
What to do: before the migration, write down your current lead costs so you have something to compare the new results with.
Sources:
- Search Engine Journal, 2026-09-17: Local Services Ads To Performance Max: What To Audit Before Migration
Companies want to invest in AI search before they see results
New signal: one source so far.
A survey of marketers shows that SEO is not losing budget. More than three quarters of companies plan to keep spending at the same level or increase it.
The interesting part is the gap: only one in seven respondents named optimising for AI answers as a strong driver of results. Even so, almost half plan to invest in it within a year (Search Engine Journal).
What to do: before you shift budget to AI, decide how you will measure the results, because competitors are often investing on faith.
Sources:
- Search Engine Journal, 2026-09-18: State Of Search 2027: What To Stop, Measure & Fund