Zero-Click Hits 68%, Amazon’s Alexa Becomes an Ad Platform, and Anthropic Calls for a Brake Pedal
Estimated reading time: 17 minutes
Plenty to get through in this week’s search marketing news. The headline figure is a grim one: fresh data from SparkToro and Similarweb puts Google’s zero-click rate at 68%, with barely a quarter of searches now sending anyone to the open web. Elsewhere, Amazon turned Alexa into a shopping agent and an ad platform, OpenAI quietly made ChatGPT ads far easier to run, and Anthropic published a blog post asking the entire AI industry to fit a brake pedal. Publishers pushed back hard against AI crawlers, too. Let’s dig in.
Table of Contents
SEO Updates & Search Data
Zero-click searches climb to 68% as open-web traffic keeps slipping
May core update aftermath: Reddit now holds almost 5x more top-3 spots than YouTube
Schema.org now shows how many sites use each schema type
AI Search, GEO & Visibility
BrightEdge: AI engines assign different roles to the same sources
The panel verdict: brand is the new backlink for AI search
Claude is the fastest-growing AI referral source
AI, Publishers & the Open Web
US publishers hit Common Crawl with a cease-and-desist
AI bots are 80% of your crawl load. Should you keep paying for it?
PPC & Paid Media
OpenAI adds product-feed uploads to ChatGPT ads
Amazon turns Alexa into a shopping agent and an ad platform
Walmart Connect audiences arrive in Google DV360
Local & Business Profiles
Business Profiles come to Gemini, and WhatsApp numbers appear unannounced
Security & Tooling
UpdraftPlus flaw puts 3 million WordPress sites at risk, and Chrome flags WebMCP
Industry Moves
Anthropic asks the industry for a brake pedal
Wrap-Up
Strategic Direction: Where Search Marketing Is Heading
Zero-Click Searches Climb to 68% as Open-Web Traffic Keeps Slipping
Start with the number that should worry anyone who relies on organic clicks. The latest instalment of the SparkToro and Similarweb zero-click study, the one Rand Fishkin has been tracking for years, found that 68% of US Google searches now end without a single click to anywhere. That alone isn’t new. What stings is where the remaining clicks go.
For every 1,000 US Google searches, only 276 now land on the open web. That’s 27.6%, and the slope is getting steeper since AI Overviews became a fixture in the results. Of the roughly 32% of searches that do produce a click, the study breaks the destinations down like this:
- 66% to the open web (sites like yours)
- 27% to Google’s own properties: AI Mode, YouTube, Maps, Images
- 6% to paid ads
Read that middle line again. More than a quarter of all clicks now stay inside Google’s own walls, never reaching an independent publisher at all. Google is increasingly the destination, not the doorway.
For practitioners, the takeaway isn’t to panic over a single quarter’s figures, it’s to stop treating raw search visibility as the whole game. If a smaller share of impressions ever becomes a session, then the value of each click that does arrive goes up, and so does the case for measuring branded search, direct traffic and assisted conversions rather than rankings in isolation. I’ll be honest, I’ve spent years telling clients that position one is the prize. The prize is shrinking, and the honest response is to say so.
At a strategic level, this is the data point to put in front of a board that still equates SEO with blue links. The work is shifting toward being the brand people already trust before they search, and being citable inside the AI answers that increasingly replace the click.
May Core Update Aftermath: Reddit Now Holds Almost 5x More Top-3 Spots Than YouTube
The May 2026 core update wrapped up a couple of weeks ago, but the analysis is still landing, and SE Ranking’s read on the winners is the one that caught my eye. Across the niches they studied, Reddit now holds almost five times as many top-three positions as YouTube. User-generated forum content keeps eating space that brands assumed was theirs.
This continues a pattern we’ve covered before, but the gap is widening rather than settling. Reddit threads rank because they carry exactly what Google’s systems, and the AI answers layered on top of them, seem to reward right now: genuine opinion, messy debate, lived experience that reads as human.
So what do you actually do with that? Not much, if your instinct is to spam Reddit with thinly veiled promotion, which the platform’s moderators will bury in hours. The more durable play is to earn the kind of authentic discussion that happens organically, and to make sure your own content carries the first-hand signals that forum posts have in abundance. Original data, named authors, real product testing.
If you’ve watched a category page slide while a three-year-old Reddit thread sits above it, this is why. The fix isn’t a tweak to your title tag.
Schema.org Now Shows How Many Sites Use Each Schema Type
Here’s a small change that quietly helps anyone making structured-data decisions. Schema.org has added usage statistics to every schema type, so you can now see how widely each one is actually deployed before you spend dev time on it. Author schema, for instance, appears on more than 10 million domains; event schema sits under 1 million.
Why does that matter? Because it turns a guessing game into evidence. When you’re arguing for engineering resource to implement a markup type, being able to say how common it is across the web, and therefore how well understood it’s likely to be by search engines, is a useful card to play. Rare schema types are rare for a reason, and adoption data hints at which ones the big consumers genuinely parse.
It won’t change your roadmap on its own. But the next time someone proposes an obscure schema type because a blog post said it might help, you’ve got a quick sanity check to reach for.
BrightEdge: AI Engines Assign Different Roles to the Same Sources
BrightEdge published research this week that reframes how to think about AI citations, and it’s the most useful GEO finding I’ve read in a while. The core idea: AI engines don’t just decide whether to cite a source, they assign it a role, and the same source can play completely different roles depending on the engine and the query.
Reddit is the clearest example. In ChatGPT it behaves almost like an authority, appearing in around 36% of citations for some query types and sitting alongside medical sites for how-to questions. In Google’s AI Overviews, that same Reddit content shows up more like social proof, around 6%, framed as community debate for comparison queries. LinkedIn, meanwhile, is consistently the professional and B2B voice: cited roughly 33% in ChatGPT and 22% in AI Overviews for how-to content.
The practical lesson is to stop chasing mentions for their own sake. A citation only helps if it comes from the kind of source the engine associates with the question you want to win. That means matching the platform to the intent: professional, credential-led content where LinkedIn dominates, genuine consumer discussion where Reddit leads, and authoritative first-party pages where your own domain has a shot.
Generic link-building logic doesn’t map onto this. You’re not building authority in the abstract anymore, you’re building it in context.
The Panel Verdict: Brand Is the New Backlink for AI Search
A WordCamp panel of SEOs reached a conclusion that’s becoming hard to argue with: for AI search, brand is the new backlink. The reasoning is that large language models lean on brand mentions, entity recognition and structured data far more than on the link graph that dominated classic SEO.
It pairs neatly with the BrightEdge findings above. If AI systems decide who to cite partly on whether they recognise you as a known entity in a category, then unlinked brand mentions, consistent naming and co-occurrence with category leaders start to matter as much as any anchor text ever did. Search Engine Land made a similar point this week about co-mentions: you want your brand appearing in the same third-party content as the obvious leaders in your space.
None of this kills link-building. Authoritative links still help, and they remain a strong signal of trust. But the centre of gravity is moving toward reputation that exists independent of links, the kind a model can pick up from being talked about across the web. Brand-building and SEO were always cousins. They’re becoming the same job.
Claude Is the Fastest-Growing AI Referral Source
If you only track referral traffic from ChatGPT, you’re missing the fastest-growing source on the board. New data shows Claude referrals have grown nearly fourfold year on year, with one analysis putting the rise as high as 386%. It’s still a smaller pipe than ChatGPT in absolute terms, but the growth rate is the story.
For anyone setting up AI-referral tracking, the message is to widen the net now. Add Claude, Perplexity and Gemini to your referrer segments in GA4 before the traffic gets big enough that you wish you’d been measuring it all along. Retrofitting attribution after the fact is always messier than setting it up early.
There’s a strategic read too. The AI-assistant market isn’t consolidating into one winner the way search did around Google. Several assistants are growing fast at once, which means your visibility strategy can’t be built around optimising for a single engine. The sources each one trusts differ, as the BrightEdge research shows, so spreading your bets across assistants is starting to look less like hedging and more like basic coverage.
US Publishers Hit Common Crawl With a Cease-and-Desist
The fight over who gets to train on the open web escalated this week. Digital Content Next, the trade body for major US publishers, sent a cease-and-desist letter to the Common Crawl Foundation, demanding it stop collecting publisher content and delete what it has already archived. Common Crawl matters here because its archive has fed the training of major models; it made up a large share of the data behind GPT-3.
DCN’s argument is a pointed one about how copyright should work in the age of scrapers:
“This challenges a growing assumption that content created through substantial investment can be collected, stored, repurposed, and monetized simply because it is technically accessible.”
– Jason Kint, CEO, Digital Content Next
In plain terms, DCN says permission should be opt-in, not opt-out. Publishers shouldn’t have to chase down every scraper after the fact; the scraping shouldn’t happen without consent in the first place.
This didn’t happen in isolation. The same week, Reuters and Time both moved to block AI crawlers by default, allowing only approved bots through an allowlist. The drawbridge is going up across serious publishers at once.
For marketers, two things follow. First, if you publish content you want cited in AI answers, the rules of access are tightening, and you’ll need to decide deliberately which crawlers you allow. Second, the supply of freely scrapeable quality content is shrinking, which raises the value of licensing deals and of being a source the models are explicitly permitted to use.
AI Bots Are 80% of Your Crawl Load. Should You Keep Paying for It?
Tied to all that: a sharp piece this week asked a question site owners are increasingly muttering under their breath. If AI bots now account for the bulk of your crawl traffic, why are you paying to serve them? One report cited puts AI-related crawling at around 80% of bot activity, most of it for model training rather than anything that sends you a visitor. AI bot traffic is up roughly 300% over the past year.
The cost is real and uneven. Static pages cached at the edge are cheap to serve. But crawlers hitting cart URLs, checkout paths, search pages and parameter-heavy URLs bypass the cache and trigger database queries and PHP execution, which is exactly the expensive stuff. E-commerce sites feel this most.
The sensible response isn’t to block everything in a panic. It’s to get deliberate:
- Watch your logs for repeated requests, loops and hits on dynamic endpoints
- Protect high-cost paths like cart, checkout and internal search from training bots
- Treat search crawlers, AI training bots and unknown scrapers as three different policies, not one
That last point is the crux. Blocking Googlebot to save money would be self-harm; blocking an unknown training scraper that never sends a soul costs you nothing. Know the difference before you touch robots.txt.
OpenAI Adds Product-Feed Uploads to ChatGPT Ads
OpenAI keeps building ChatGPT into a real ad platform, and this week’s addition is the practical one. Advertisers can now upload a product feed to the ChatGPT Ads Manager, and it will automatically generate ads for each product in the feed. For retailers, that turns a fiddly manual setup into something close to the Shopping workflow they already know.
We flagged in a previous update that conversion-focused ChatGPT ads were coming. This is the plumbing that makes them usable at scale: no feed support, no serious e-commerce spend. With it, ChatGPT ads move from an experiment you read about to a channel a performance team can actually plan around.
It’s early, and the targeting and reporting are nowhere near Google’s maturity. I wouldn’t shift real budget yet. But I would get a product feed validated and ready, and I’d be watching for the first case studies on cost-per-acquisition, because the moment those look competitive, the pressure to test it will arrive fast. Retail and e-commerce teams especially should treat this as a get-ready signal, not a someday one.
Amazon Turns Alexa Into a Shopping Agent and an Ad Platform
The bigger paid-media story might be Amazon’s. It’s rebuilding Alexa into a shopping agent powered by Rufus, its AI assistant, so people can discover, compare and buy products through a back-and-forth conversation. And, naturally, ads will appear inside those conversations.
Think about what that means. A shopper asks Alexa to find a good cordless drill under a hundred pounds, and the assistant returns a curated, conversational answer with sponsored placements woven in. The product discovery journey collapses into a single exchange, and the brand that wins is the one the agent recommends, not the one with the best-ranked product page.
This is agentic commerce arriving in a living room near you, and it raises the same question GEO has been circling for months: how do you influence a recommendation made by an AI on the buyer’s behalf? Feed quality, review signals, price competitiveness and structured product data all point the same way, toward being the obvious machine-readable choice.
For now this is an Amazon-ecosystem play, most relevant if you sell on or advertise with Amazon. But the direction matters for everyone. When the assistant becomes the shop assistant, optimising for the algorithm and optimising for the agent stop being separate disciplines.
Walmart Connect Audiences Arrive in Google DV360
Retail media keeps bleeding into the open web. Walmart Connect audiences are now available in Google’s Display & Video 360, which means advertisers can target Walmart shopper segments across YouTube and the wider Google inventory, with Walmart’s sales data closing the measurement loop.
This is a meaningful development for anyone selling through Walmart. You can reach Walmart’s first-party audiences in environments well beyond Walmart’s own properties, and then tie exposure back to actual sales rather than proxy metrics. Retail media networks have spent two years trying to prove incrementality; partnerships like this one are how they do it.
The broader trend is retailers turning their purchase data into ad products that travel. Amazon, Walmart and the rest increasingly sit alongside Google and Meta as places to buy audience, not just placement. If your media plan still treats retail media as a walled-off line item, this is a nudge to rethink where it fits.
Business Profiles Come to Gemini, and WhatsApp Numbers Appear Unannounced
Two Google Business Profile changes landed this week, one welcome and one that needs watching. The welcome one: Google is bringing Business Profile management into the Gemini app, so you’ll be able to ask Gemini about your reviews, your local listing and your Maps and Search performance in natural language. It rolls out this month.
The one to watch is stranger. Google appears to have auto-added WhatsApp phone numbers to large numbers of Business Profiles over the past week, sometimes with the wrong number, sometimes with one that isn’t even reachable. Some owners got an email; many didn’t.
So here’s a five-minute job for your Monday: check the WhatsApp and contact details on every Business Profile you manage. An incorrect or dead contact number on a local listing is the kind of silent error that quietly leaks leads for weeks before anyone notices. Nobody complains to you; they just call the wrong number, or no number, and move on.
The Gemini integration is the headline. The WhatsApp surprise is the one that’ll actually cost someone money this month.
UpdraftPlus Flaw Puts 3 Million WordPress Sites at Risk, and Chrome Flags WebMCP
A reminder that the boring security basics still bite. A vulnerability in UpdraftPlus, one of the most widely installed WordPress backup plugins, has put as many as 3 million sites at risk, with the flaw potentially letting attackers compromise a site and inject malware. If you run UpdraftPlus anywhere, update it now, before you read the rest of this sentence twice.
Separately, Chrome’s team warned that WebMCP, an emerging standard for letting AI agents act inside web pages, can be abused to hijack those agents while they’re operating in a logged-in browser session. As more people hand browsing tasks to AI agents, the attack surface grows in ways most marketing teams haven’t begun to think about.
The thread connecting both: the tools we bolt onto our sites and browsers carry risk that has nothing to do with rankings, and everything to do with whether your site is online and trusted tomorrow. Patch hygiene isn’t glamorous. It’s also the difference between a quiet week and explaining to a client why their site is serving malware.
Anthropic Asks the Industry for a Brake Pedal
Finally, the strangest story of the week, and the one with the longest shadow. On 4 June, Anthropic, the company behind Claude, published a post titled “When AI Builds Itself” arguing that AI is advancing so fast that the industry needs a coordinated way to slow down, a “brake pedal” it can pull if certain thresholds are crossed.
The supporting numbers are striking. Anthropic says more than 80% of the code merged into its own codebase as of May was written by Claude rather than by human engineers, and that its engineers ship roughly eight times as much code per day as they did in 2024. An external evaluator, METR, found the length of tasks AI can handle on its own has been doubling about every seven months.
Not everyone took it at face value. Plenty of commentators pointed out the awkward timing of a company reportedly near a trillion-dollar IPO calling for a slowdown that would conveniently freeze the competitive order in its favour. The Wall Street Journal noted critics who read it as marketing more than policy.
Why include this in a search marketing roundup? Because the pace it describes is the same force reshaping our work. The tools we use, the way content gets made, the assistants that increasingly sit between us and our audiences, all of it is moving at this speed. Whether or not anyone fits a brake pedal, planning as if the ground will keep shifting is just realism now.
Strategic Direction: Where Search Marketing Is Heading
Step back from the individual stories and one theme dominates: the open web is being disintermediated, and search marketing is adapting to a world where the click is no longer the unit of value. Zero-click at 68%, a quarter of clicks staying inside Google, publishers slamming the door on crawlers, assistants like Claude growing fast, Amazon turning a voice assistant into a storefront. These aren’t separate trends. They’re the same shift seen from different angles.
The practical focus is moving in two directions at once. Toward brand, because being recognised and trusted before the search happens is what survives when the click disappears. And toward being machine-readable and citable, because a growing share of demand now passes through an AI layer that summarises, recommends and transacts on the user’s behalf.
For agencies and in-house teams, the planning question for the second half of 2026 is uncomfortable but clarifying. If half your organic value already arrives without a traditional click, what are you actually optimising, and how do you prove it worked? The teams that answer that with brand metrics, AI-citation tracking and assisted-conversion data will look a lot smarter in six months than the ones still reporting average position.
Key Takeaways
- Zero-click reached 68%, and only about 28% of US searches now send a click to the open web. Shift your reporting toward branded search, direct traffic and assisted conversions.
- AI engines assign different roles to the same source. Match the platform to the query: LinkedIn for professional and B2B, Reddit for consumer discussion, your own domain for first-party authority.
- Brand is becoming the strongest AI-search signal. Unlinked mentions, consistent naming and co-occurrence with category leaders now rival backlinks.
- Claude is the fastest-growing AI referral source, up nearly 4x year on year. Add Claude, Perplexity and Gemini to your GA4 referrer tracking now.
- Publishers are closing the gates: DCN’s cease-and-desist to Common Crawl, plus Reuters and Time blocking AI crawlers by default. Decide deliberately which bots you allow.
- OpenAI’s ChatGPT ads now support product feeds, and Amazon’s Alexa is becoming a shopping agent with ads. Get product feeds clean and ready even if you don’t spend yet.
- Patch UpdraftPlus immediately if you use it; up to 3 million WordPress sites are exposed.
Frequently Asked Questions
Is SEO still worth investing in if zero-click searches have hit 68%?
Yes, but the goal is shifting. With fewer searches producing a click, the value lies in being the brand people trust and the source AI answers cite, not just in ranking first. Measure branded search, direct traffic and assisted conversions alongside rankings.
What does it mean that AI engines assign “ranking roles” to sources?
BrightEdge found the same source can be treated as an authority in one AI engine and as social proof in another, depending on the query. Practically, you should earn citations from the platforms each engine associates with your topic, rather than chasing mentions anywhere.
Should I block AI crawlers from my website?
It depends on your goals. Blocking training bots that never send traffic can cut server costs with little downside, but blocking the assistant crawlers that cite sources could reduce your AI visibility. Treat search crawlers, AI training bots and unknown scrapers as separate decisions.
How should retailers prepare for ChatGPT and Alexa shopping ads?
Get your product feed clean, complete and validated now. Both OpenAI’s product-feed ads and Amazon’s Alexa shopping agent reward structured, high-quality feed data, so the groundwork pays off before you commit budget.
Why is Anthropic asking the AI industry to slow down?
Anthropic argues AI is improving fast enough to risk losing meaningful human oversight, and wants an agreed mechanism to pause if thresholds are crossed. Critics counter that the timing suits a market leader, so read it as one influential view rather than settled policy.
Conclusion
If there’s a single thread running through this week’s search marketing news, it’s that the click is quietly being demoted. Zero-click is at 68%, a quarter of clicks never leave Google, publishers are closing their doors, and AI assistants are growing into the place where discovery and even buying now happen. None of this means search is dying. It means the job is changing from chasing positions to building a brand that AI systems recognise, trust and recommend.
The teams that thrive through the rest of 2026 will be the ones who measure value beyond the click, keep their feeds and technical foundations clean, and treat every AI assistant as a channel worth understanding. The ground is moving. The smart response is to move with it, deliberately.
Need help adapting your search strategy for the AI era? Contact the Anicca team for expert SEO and PPC guidance.
