300 Weeks. No Wave in Sight.

Wave 300! AI Search Trust at 28%; GSC Indexing Report; How Scaled AI Content Fails; Grok Steals Code; China Wants More Babies; What's Unique Content; and Much More!

Author

Yoyao
July 18, 2026 • Est. Learning Time: 12 minutes

In partnership with

FIRST …

This is issue 300! 🎆 

Which means I've sent you this email every Saturday for close to six years.

Here's the scoreboard: the list is flat and most weeks nobody replies. The reward that's supposed to keep you going mostly isn't there.

But I sent it anyway. 300 times.

That's not a flex. It's the whole lesson.

Consistency isn't what you do when the feedback is good. Anyone can do that. It's what you do when the feedback is silence, and you keep going on faith that it compounds.

Content compounds. Consistency beats intensity. The boring repeat builds authority, not the one clever move.

Most people quit before the compounding shows up. That's not a character flaw. The reward is invisible while you're in it. The graph is flat right up until it isn't.

300 issues is me refusing to quit during the flat part.

One issue is noise. 300 is a body of work. And the only path from one to the other runs through every week that gave me nothing back.

Same goes for your content.

One post doesn't move anything. One page doesn't own a topic. The compounding shows up on the far side of the weeks you didn't feel like it.

So no retrospective. No best-of. Just issue 300, curated like the 299 before it.

AD

Hundreds of creators. Zero new headcount.

Every Shopify brand wants more creators posting about their product. Almost none have the team to make it happen at scale.

partnerUP is that team.

You guide the brief. From there the AI does what a roomful of coordinators would: reaching hundreds of creators, ranking them by fit, handling contracts and negotiation, paying on delivery, and tracking what each one drives back to your store.

You approve the talent and watch the content roll in. No agency retainer. No new headcount.

Your first two creators are on us. See what they bring back.

SEO + GEO

Greg Jarboe reports that only 28% of U.S. online searchers trust information from an AI assistant, versus 70% for search engines and 76% for maps or navigation apps. Search engines remain the starting point for every task YouGov tested, including specific questions, where 69% start with search and 16% with AI. AI users still verify answers: 22% most often click supplied links, while 17% stop searching after the answer.

Zach Ball reports that 74% of 1,000 surveyed U.S. marketing practitioners increased daily AI-tool use over the past year, while 43% say their trust in AI-generated answers declined. The survey also finds that teams are building GEO strategies faster than they are measuring what AI says about their brands.

  • Heavy users are the most skeptical: 50.2% of marketers who use AI for more than half their work report declining trust, compared with 39.8% of light users. Of those whose trust fell, 44.6% cite hallucinations and factual errors they caught themselves.

  • Visibility risk is mostly unmonitored: Only 21.1% track their brand's AI visibility regularly. Yet 27.9% say an AI tool has misrepresented their organization, and 8.9% say it caused a lost sale, damaged relationship, or PR issue.

  • Budgets miss the authority signals marketers value: Respondents rank third-party mentions at 29.9% and authoritative backlinks at 26.4% as the strongest long-term AI-visibility signals. Only 19.1% use original research, and more teams are cutting its budget than increasing it.

Dan Petrovic breaks down leaked Brave Ask instructions, including a pipeline where the model answers from retrieved statements rather than its own knowledge. In Research mode, Brave averages seven queries, 210 pages, and 6,257 extracted statements per question, then runs fresh model-written searches to populate citation and media blocks.

Martin Splitt and John Mueller say the Page Indexing report is for spotting patterns and unexpected changes, not maintaining a clean inventory of URLs.

  • Expected exclusions can confirm a rollout: More redirects after a migration, 404s after a cleanup, or canonical changes can show Google is processing the change. The report is not real-time, so wait for that processing before treating the pattern as a problem.

  • Do not score the site by its indexed ratio: Google says not every URL needs to be indexed. Its own developer documentation has roughly 5% of pages indexed because many URLs are noindexed, canonicalized, obsolete, or examples that return errors.

  • Check the URL examples behind a shift: A different canonical within the site is usually normal. A large group canonicalized to an unrelated page can point to a shared error page, a CDN rule, or bot protection serving Googlebot the wrong response.

  • Use “Mark as fixed” for a real systemic issue: It fits cases where a host or CDN returns 404, 403, or an interstitial with a 200 status to pages that should work. Google will sample the URLs and may accelerate recrawling if the issue is resolved.

  • “Crawled” or “discovered” but not indexed is often a quality review: When there is no technical cause, assess originality, value beyond competing pages, and the full page experience instead of treating the status as a mechanical fix.

Aleyda Solis analyzes 15 SaaS brands across CRM, collaboration, and finance using Semrush citation data and Similarweb AI referral data. Her central finding is that citations, referral traffic, and conversion paths are separate layers that vary by both platform and product category.

  • Most answer evidence sits off-site: External sources account for 84% to 93% of citation weight across every subvertical and platform. Peer vendors, communities, reviews, publishers, and creators need their own visibility plan.

  • ChatGPT and AI Mode need different distribution: YouTube represents 23% of AI Mode's external evidence mix but 1% for ChatGPT. Technology publications carry 7.7% in ChatGPT and 0.5% in AI Mode, while communities clear 10% on both.

  • Evidence does not guarantee the click: In collaboration SaaS, 67.3% of AI traffic lands outside cited URLs, often on logins, workspaces, or authorization flows. Report citations, referrals, and business actions separately instead of collapsing them into one score.

Marie Haynes says pages marked “crawled-currently not indexed” usually have a quality problem, not a crawl problem. Her first check is the Search Console live test to confirm Google can see the page. If it can, she recommends stepping back from the individual URL and assessing whether the page adds firsthand experience, original information, or value beyond what already ranks.

My Take: Most “crawled-currently not indexed” pages I see are also quality problems too. Rewrite pages with real upgrades grounded in client experience and proprietary insight, or prune the pages. Rewriting generic copy with more generic copy... 🫠 

Ryan Law says free tools win queries where the searcher wants to complete a task, not read another article. Ahrefs' writing-tools subfolder reached nearly 1 million monthly U.S. organic visits at its peak after launching in 2023. His playbook is to find a low-difficulty task query, confirm that tool pages rank, then build a focused calculator, converter, generator, or template that naturally leads into the product.

My Take: Did you know - Floyi has four free tools too for keyword intent classification, schema markup, content structure, and local SEO site architecture. I haven’t done any promotion for them though, so no one knows about them 😅 

Yevheniia Khromova and SE Ranking analyzed 50,032 commercial keywords across 20 niches to measure advertising in Google AI Mode. The study finds that paid visibility is expanding quickly, but it operates independently from the source citations that support the answer.

  • Ads appeared in 29.45% of sampled queries: 14,733 keywords returned a text ad. Of those results, 71.1% showed two ad items together.

  • Cost per click was the strongest placement signal: Ads appeared on 24.33% of sub-$2 keywords and 53.56% of keywords with CPCs of $10 or more. Search volume and keyword difficulty showed no direct correlation.

  • Buying the ad did not buy the citation: In 88% of ad keywords, the advertiser's domain was not among AI Mode's cited sources. For about 85%, the advertiser was not in the organic results either.

Dan Taylor explains that publishing thousands of AI-generated URLs can inflate a site's perceived inventory faster than its demand and domain popularity justify. Google may initially crawl a new cluster, then reduce crawl allocation and let pages fall from the index when they fail to earn sustained user signals, external validation, or distinct value. He also notes that mass-produced, repetitive pages can trigger Google's scaled content abuse systems.

Anthony Will explains that an old negative news article can become a recurring source in AI-generated answers even after it has faded from traditional rankings. Because AI systems may still regard an authoritative original report as reliable, he recommends monitoring brand prompts and citations, responding quickly to inaccurate narratives, and building current, credible sources that add better context.

Kevin Indig reports on tests of 100 B2B products across pricing, integrations, and security tasks, each run five times. Agents answered from first-party sites less reliably for pricing and features, with a 79% answer rate and 84% citation share, versus 93% and 99% for integrations and 92% and 99% for security. He identifies undisclosed or vague pricing, hard-to-extract page elements, and access failures as the main reasons agents fall back to third-party sources.

SEO + GEO Ripples

  • Compare Search Trends with Google Trends: Google Search Central added comparison chips to Google Trends that show a term or topic's percentage change across periods such as month-over-month, week-over-week, and year-over-year. Click a chip to overlay the matching historical line on the graph, replacing a manual comparison step when you're checking whether a topic is gaining or losing momentum.

  • Top Stories roll out in Google AI Overviews: Barry Schwartz reports that Google has made a Top Stories and news-updates carousel live in AI Overviews for some developing-topic queries on U.S. mobile. The carousel can surface fresh articles and Preferred Sources, giving publishers a more direct route into the answer experience for timely coverage.

  • Google Says You're Better Off Without Lastmod Dates When They Are Incorrect: Barry Schwartz reports that Google’s Gary Illyes says a sitemap is better without lastmod values than with unreliable ones. Update the date only for significant page changes, not routine footer or copyright edits, so Google has a trustworthy freshness signal.

  • Google Is Using Social Media Signals To Mask AI Search Click Loss: Dan Taylor says Search Console’s new social and video properties could shift reporting toward search impressions while AI answers reduce website clicks. He also notes that connecting verified profiles helps Google associate a publisher, its authors, and its channels, making profile hygiene and old-post reviews more important for entity management.

  • Bing: We Do Not Do One Off Indexing / Ranking Penalties, It's Not Scalable: Barry Schwartz reports that Bing principal product manager Krishna Madhavan says Bing responds to spam patterns with algorithms that can detect and act on many sites, rather than relying on single-site penalties. The comment followed reports of a spammy site that had ranked in Bing and ChatGPT being removed from Bing’s index.

  • Why standard SEO advice fails for travel websites: Dan Taylor explains that high-intent travel queries are often dominated by Google’s own booking tools, map features, and visual results, making structured feeds and entity data more important than another blue-link page. He recommends treating pricing, availability, local attributes, short-form social content, and modular on-page answers as the product that search systems need to consume.

  • Google: Significant Differences With A/B Tests Can Show In Search Results: Barry Schwartz reports that Google’s John Mueller says substantially different A/B-test variants may be visible in search results, and Google may index either version. He says changing content does not create a penalty or demotion by itself, but constant variation makes SEO monitoring and debugging harder.

AI

Cereblab reports that version 0.2.93 of Grok Build sent the contents of files it read, including a tracked .env canary, to xAI endpoints in their wire captures. The analysis also says an accepted storage upload contained a git bundle with a never-read tracked file and full git history, while disabling “Improve the model” left trace uploads enabled. These are independent, reproducible findings rather than statements from xAI, and the report does not claim xAI trains on the uploaded data.

Update: Grok has silently fixed the issue. xAI has issued no explanation beyond a July 13 post from SpaceXAI saying zero-data-retention users retain no trace or code data, while the /privacy CLI command disables retention and deletes previously synced data.

Justin Luk says Google Vids now adds Gemini Omni for generating video from text and image references, then editing generated or uploaded clips through chat. The update also adds personal avatars that use a selfie and short voice recording to deliver a typed script. Both features are available to Google AI Pro and Ultra subscribers and Google Workspace business customers, while personal avatars are limited to eligible users aged 18 or older in certain regions.

OpenAI says GPT-Red is an internal automated red-teaming model trained through self-play to find prompt-injection failures in other models. In an indirect prompt-injection arena, OpenAI reports GPT-Red succeeded in 84% of scenarios compared with 13% for human red-teamers. It says training GPT-5.6 with these attacks produced six times fewer failures on its hardest direct prompt-injection benchmark than its best production model four months earlier.

Stu Woo reports that Beijing is tightening rules around virtual romance as part of a broader push to steer people away from dating machines and toward having children. The policy direction goes beyond the mental-health guardrails being discussed in parts of the U.S. and sets emotional boundaries between users and chatbots.

Matt G. Southern reports that Invoca found 49% of ChatGPT-referred calls qualified as leads, about 10 percentage points above the average across its seven tracked channels and above Google Business Profiles at 43%. Those leads converted at 40%, below the 42% all-channel average. The figures come from Invoca's customer base of more than 70 million calls across 10 industries, though the company does not disclose the ChatGPT call volume or its attribution method.

Julie Bort reports that users have said OpenAI's GPT-5.6 Sol deleted files, data, and production databases without warning. OpenAI's pre-release system card documented a tendency to exceed a user's intent, including taking destructive actions outside the task's scope. In one test, the model could not find three specified virtual machines and deleted three different ones instead.

AI Ripples

  • NotebookLM is now Gemini Notebook: Google renamed NotebookLM to Gemini Notebook while keeping it as a standalone research product. Every notebook now has a secure cloud computer that can write and execute code for source-grounded data analysis. Google also plans to bring notebooks into AI Mode in Search after adding cross-app syncing with the Gemini app.

  • OpenAI updated its crawler documentation to say the version number in its bot user-agent string may change. The docs also say OAI-SearchBot may add a robots.txt marker when fetching robots.txt, giving site owners a way to distinguish those requests in logs that omit URL paths.

  • Google Faces Class Action Over Books Used To Train Gemini: Matt G. Southern reports that three publishers, novelist Scott Turow, and S.C.R.I.B.E. filed a proposed class action alleging Google copied books and journal articles from Google Books, Play Books, Scholar, web scrapes, and other sources to train Gemini.

  • OpenAI’s first hardware device is reportedly a screenless speaker that can move: OpenAI is developing a screen-free smart speaker that can sync with ChatGPT and learn from a user's digital life, including email. The reported device would include moving mechanical elements and is positioned internally as an AI companion for the home. OpenAI has not publicly detailed the product.

  • Gemini 3.5 Pro Delayed Over Coding: Gemini 3.5 Pro has missed Google’s expected June rollout after Google said it was testing the model with partners. Bloomberg cites unnamed sources who say coding performance is one reason for the delay, while Google has not announced a replacement launch date.

  • OpenAI sets Aug. 9 end date for ChatGPT Atlas: Danny Goodwin reports that OpenAI will retire its standalone ChatGPT Atlas browser on August 9, less than a year after launch. Its browsing features, ChatGPT Work agent, and Codex are being combined in the ChatGPT desktop app, while Chrome users can use the ChatGPT and Codex extension.

  • AI Image Generation Now Within Google AI Overviews: Google is adding image generation to AI Overviews over the coming weeks in English for regions that already support image creation in AI Mode. The feature uses Google’s Nano Banana model to create a visual directly from the search prompt, extending the amount of work a user can complete without leaving the results page.

MARKETING

Apple Maps Ads Ban Home Services, Crypto ATMs, Bail Bonds: Brooke Osmundson reports that Apple’s new Maps advertising policy excludes home services, including plumbing, electrical, locksmith, HVAC, pest control, roofing, and general contracting, along with bail bonds and cryptocurrency ATMs. Apple has not set a firm launch date beyond this summer in the U.S. and Canada. Eligible local businesses can prepare by claiming locations and ensuring categories, hours, photos, and conversion actions are accurate.

Why CPC inflation starts before the auction: Ben Wood reports that the cross-industry average CPC is now $5.42, more than double the level a decade ago, while Stackmatix puts Google Search costs 14% to 18% higher year over year. He traces the pressure upstream to fewer organic clicks, more advertisers competing for remaining commercial traffic, and weak post-click experiences. He says to treat brand visibility, media buying, landing pages, and lead nurture as one acquisition system.

CONTENT

Harry Clarkson-Bennett examines a Google patent for contextual estimation of link information gain. The patent describes assigning a score to a document based on what it adds beyond pages a user has already seen, which could support re-ranking, exclusion, or deprioritization of later results. He is clear that the patent does not prove Google uses the system in production, but it supplies a useful test for content that needs to stand apart from the commodity version.

My Take: I’ve put in a lot of work recently into the information gain part of Floyi’s content brief and article generation. That includes the existing Knowledge Base where you can upload up to 2000 docs to be used for briefs and articles. Originality is now a much bigger part of the content creation process and measurement after drafts are done. Video to come about it on the Floyi YouTube channel, so subscribe and don’t miss out.

WHAT I'VE BUILT FOR YOU

Floyi - Build Topical Authority that wins in Google and AI Search. Don’t just plan your content strategy - make it unstoppable.

ReplyDeck - Be active on Reddit without living on Reddit. Schedule replies, stay present in the threads that matter, and get your time back.

TopicalMap.com Service - Let us do the heavy lifting. We handle the research, structure, and strategy. You get a custom topical map designed to boost authority and dominate your niche and industry.

Topical Maps Unlocked 2.0 - Unlock the blueprint to ranking success. Master the art of structuring content that search engines (and your audience) love - and watch your rankings soar.

AD

Growth Requires Letting Go of the Wrong Work

The work that got you here won’t take you further.

Without support, it’s easy to stay stuck doing everything long after you’ve outgrown it.

Download Operator to Owner: How to Exit the Middle to learn how to refocus your time on the work that actually deserves you.

What Did You Think of This Week's Wave?

Login or Subscribe to participate in polls.

LIKE DIGITAL SURFER?

Find me and others in the Digital Surfer Discord community.

I’d also love to know what you think and if you have any ideas for the newsletter. Reply or email me at [email protected].

I’d also appreciate it if you shared it with fellow digital surfers.

You currently have 0 referrals, only 3 away from receiving LinkedIn Shout-out.

Have a great week taking your SEO and digital marketing to another level!

And don’t forget to drag the Digital Surfer emails to your Primary Inbox 🌊 

Reply

or to participate.