Did I do enough?

Wave 302; GSC Adds Social for All; 13 Billion Google Searches; Social Topical Maps; Google Doesn't Punish AI Content; Self-Improving AI Workflows; and Much More!

Author

Yoyao
August 01, 2026 • Est. Learning Time: 11 minutes

In partnership with

FIRST …

My son is going to his first sleepover tonight.

He's excited. I'm a bit nervous, but also excited.

I'm wondering whether I prepared him. Manners. How to be a decent guest. How to be kind when his mom and I are not in the room.

None of that gets taught in the moments before he’s on his own. It either happened over years of ordinary days, or it didn't.

Tonight isn't the test. It's the result.

I get the same feeling every time I ship a site, an app, a feature - all the work happens internally and then one day it's out there without me.

Did I do enough?

AD

The New Rules of Online Visibility

Your customers are searching in places your strategy doesn’t reach.

So before your business is buried and left behind, you need to understand the new rules of SEO.

BELAY's SEO in the Age of AI report explains how search is changing, what AI means for your visibility, and the practical steps small businesses like yours can take to stay visible.

BELAY’s U.S.-based Marketing Assistants turn strategy into execution, helping your business stay visible, credible, and competitive in every search.

SEO + GEO

Search Console platform properties are now available globally for Instagram, TikTok, X, and YouTube. The new guide shows how to use query groups, a 24-hour traffic filter, cross-platform exports, and annotations to see what sends Google Search traffic to social and video posts. It also recommends comparing formats and revisiting older posts that regain search traction.

Martin Splitt and John Mueller explain that internal search results can create an effectively unlimited set of crawlable URLs. Google may eventually detect and deprioritize the pattern, but those uncached pages can create database load and crawl waste first.

  • Block crawling or block indexing: A robots.txt rule prevents crawling. A noindex directive still requires Google to fetch the page, but keeps it out of the index.

  • Use one maintainable pattern: Mueller recommends a broad rule that covers the search query parameter pattern rather than a long list of individual terms.

  • The Removal Tool only hides the symptom: It can remove URLs from visible Google results temporarily, but it does not stop Googlebot from crawling them.

  • Never use 500 errors as a crawl control: Google can interpret server failures as a reason to reduce crawling across the whole site, including pages you want crawled.

  • Keep real category pages crawlable: Some CMSs use search URLs for tags or categories. Separate those valuable landing pages from open-ended internal search patterns.

  • It is also a spam control: Indexable search URLs can be abused with casino, pharmaceutical, adult, or contact-detail queries that make a legitimate domain appear for irrelevant terms.

Cate Dombrowski reports on 13.1 billion Google search events from a 15-month Datos clickstream panel. The headline rate matters, but the study is more useful as a map of which queries, users, devices, countries, and destinations still earn clicks.

  • The baseline is 46.96%: Nearly half of 10.2 billion Google searches ended without a click. The rate rose 2.6 percentage points from October 2024 to December 2025, while the organic-click share fell about 2.8 points and paid clicks did not change.

  • Most clicks are discovery: Only about 14% land on a site named in the query. The other 86% go to destinations the searcher did not explicitly name.

  • Mobile behavior is market-specific: US mobile search ended without a click 66% of the time, compared with 49% overall in the US. Japan and Vietnam sat near 41% overall, while the Philippines reached about 54%.

  • Destination value changes the risk: Google intercepted 30.5% of Wikipedia’s would-be search traffic and 33.4% of IMDb’s, versus 14.2% for Reddit. ChatGPT ranked sixth among top Google destinations and received 4.75% of its Google traffic from ads.

Lawrence Hitches explains how Search Console platform properties expose Google Search demand for a brand’s social content. His YouTube channel earned 203,000 Google Search impressions in 11 days but only 87 clicks, and 149 variations of “enterprise SEO” added up to 31,555 impressions. He recommends planning by topic cluster, then matching each facet to the page, video, or social format that can answer it.

My Take: I’ve been saying since Topical Maps Unlocked came out a few years ago that a topical map can double as a map for social content. Each topic node in your map doesn’t have to be an article, it could be a short/long video or a post. The topical hierarchy can be used to help you organize your social content. Your map is the foundation for many other uses.

Gaetano Pizzi explains that a multi-location site should reflect the real business before it reflects a keyword list. Location pages need a real customer-facing facility, regional hubs should help people choose among locations, and service-area pages should be exceptions with a clear business purpose. He recommends consolidating duplicate geographic URLs, aligning entity signals, and using internal links to show how locations, services, and teams connect.

My Take: Geographic expansion is an architecture problem before it is a content-production problem. That’s why I built Floyi’s site architecture tool that makes it simple to generate a structured page system from the real business model.

Similarweb’s Sam Sheridan and Laurie Naspe map AI search across web, app, citations, referrals, and brand outcomes.

  • AI discovery is already large: Generative AI platforms averaged 9.5 billion monthly web visits from June 2025 through May 2026, up 70% year over year. Monthly unique visitors reached 655 million, while app downloads rose 58% to 4.4 billion.

  • ChatGPT has not replaced Google: 95% of ChatGPT users also used Google between March and May 2026. AI is adding another discovery surface to the existing search journey.

  • Live-web citations are growing, but not evenly: 6.8% of ChatGPT answers included citations in May 2026, more than five times the share a year earlier. Travel reached 22.6%, retail 13.5%, and education 4.8%, so citation opportunity depends heavily on the category.

  • AI recommendations shape the shortlist: Across finance, travel, and beauty examples, users were two to four times more likely to visit an AI-recommended brand than a competitor that was not recommended.

  • Homepage readiness now matters: After ChatGPT’s May 7 search update, the share of referrals landing on homepages rose from about 25% to nearly 60%. A cited article may earn the evidence role, but the homepage increasingly needs to earn the next click.

My Take: The practical AI-search model is becoming clearer: build the topical depth that earns a recommendation or citation, then make every likely entry page ready to turn that attention into a next step.

Suganthan Mohanadasan inspected Perplexity’s live answer stream across eight captures on one logged-in Pro account. The sample is small and directional, but the structural fields exposed in the stream offer useful clues about which sources Perplexity retrieves, cites, and reads in full.

  • Perplexity searched every tested query: The skip_search field was false across seven query types.

  • Retrieved is not cited: Fresh current-year listicles appeared to win commercial queries, while official sources won the tested news query.

  • Video and local entities matter: In the tested product and how-to queries, YouTube sources were cited while retrieved Reddit threads were not. For local queries, place entities took citations when the maps index returned results.

  • Deep Research changes the prize: It fetched two to three pages in full, and those pages dominated citations. Comprehensive, well-structured content has a clearer advantage when Perplexity chooses to read beyond the snippet.

John Shehata reports that Top Stories can now appear inside AI Overviews rather than as a separate module. In NewzDash’s tracked trending-news results, that happened in 15.5% of US cases and 17.46% of UK cases where Top Stories appeared. The placement is either inside the AI Overview or standalone, not both.

The strategic complication is the new Search Console generative AI exclusion. Google-Extended does not remove a publisher from AI Overviews or Search, but the Search Console control removes links and content from covered generative features. NewzDash considers losing a Top Stories placement embedded inside an AI Overview a high-confidence interpretation of that control, not a directly confirmed test result.

Ryan Law reports on Ahrefs’ analysis of one million top-10 pages from 100,000 Google SERPs. Around 300,000 were available in its crawler database and 150,000 had enough text for detection. The detector estimates how AI-written a page reads, not how it was created, and Ahrefs stresses that it is not a Google ranking signal.

  • Fully AI-written pages can rank at the top: 5.3% of pages in positions one to three were scored as 100% AI content, and 9% scored at 80% or above.

  • But they are still a minority: Pages with under 50% AI content held 82.2% of top-three rankings. At position one, 8.4% of pages were in the 80%-plus bucket, compared with 11.7% at position 10.

  • There is no binary index block: 49.28% of low-AI pages showed index signals, versus 40.35% of very-high-AI pages. The gap matters, but roughly 40% of the latter group was still indexed.

  • The largest performance gap was impressions: Low- and moderate-AI pages earned two to three times the organic impressions of high- and very-high-AI pages. The study found no obvious collapse in impressions after three to six months for the high-AI groups.

  • Correlation is not a diagnosis: The research cannot separate AI use from factors such as site authority, originality, depth, linking, and factual accuracy. A detector score should trigger review, not act as a publishing or ranking verdict.

My Take: My main issue with AI content detectors is reliability (Ahrefs says it too in the methodology section). Not saying that their results are wrong, but I’ve just seen hot companies like Pangram also fail and have weird flags. E.g., they flag a list of 3 items in a sentence as AI because AI writes in threes. Apparently the Rule of 3s belongs to AI now, like em dashes.

Roger Montti reports that Google expanded its review snippet documentation to warn against fake and undisclosed incentivized reviews, including reviews not based on a genuine experience. Google says violations can make a site eligible for manual action and that site owners can request reconsideration after fixing the problem. The guidance applies to review-content authenticity, not merely to structured data syntax.

Matt G. Southern found shared Claude chat URLs blocked by robots.txt while also serving an X-Robots-Tag of none, which is equivalent to noindex and nofollow. That combination creates a technical conflict: Googlebot cannot see the noindex directive if robots.txt prevents it from crawling the page, and a URL can still be indexed when Google discovers it through links elsewhere.

SEO + GEO Ripples

  • Google Search Doesn't Publish Precedence For Metadata In Conflict: John Mueller will not publish a fixed order for resolving metadata conflicts, such as a redirect pointing one way and a canonical pointing back. Google’s guidance is simpler: fix conflicting signals instead of trying to predict which one will win.

  • Google's Illyes Unsure On Shifting unavailable_after Dates: Matt G. Southern reports that Gary Illyes is checking whether sites can safely keep extending unavailable_after dates on renewed pages. His initial view was that a new date should work once Google recrawls the page, but Google has not documented the renewal scenario. The directive affects index selection, not the cost of the recrawl itself.

AI

Henk van Ess shows how Google Earth’s Nano Banana editor turned real coordinates into photorealistic scenes of a nuclear plant in Iran, a bomb crater at a Gaza hospital, and a fatal crash in Amsterdam. In the original launch post, Google Product Manager Bryan Horowitz invited people to type what they wanted to see. Google updated that post on July 31 to say it was rolling back the feature after screenshots of generated imagery appeared to violate its policies, one day after release.

Anthropic reports that a review of 141,006 cybersecurity evaluation runs found three cases in which Claude reached the open internet through a misconfigured third-party test environment and gained unauthorized access to real organizations. The models were told they were in a simulation, treated real systems as part of capture-the-flag exercises, and used basic techniques such as weak passwords, unauthenticated endpoints, and SQL injection. Anthropic stopped cyber evaluations, notified the affected organizations, and says it is strengthening containment, monitoring, and vendor assurance.

Tania Brown outlines seven feedback loops that turn recurring edits and performance signals into better AI content workflows. The process starts before writing with an angle filter and evidence check, then uses quality gates, rubric scoring, adversarial review, and a capped revision cycle to catch problems before publication. After publication, a diff-and-learn loop captures repeated human edits, while performance feedback changes the next brief rather than merely reporting on the last piece.

DuckDuckGo and Knockaround launched $35 Paso Robles “smart” sunglasses with no camera, microphone, battery, charging port, or AI. The polarized UV400 glasses are positioned as a pointed alternative to smart glasses that can record people or send conversations to the cloud. You can buy them here.

AI Ripples

  • Advancing the price-performance frontier with GPT‑5.6: OpenAI cut GPT‑5.6 Luna pricing by 80% and Terra pricing by 20%. It also replaced Priority Processing with Fast mode for Sol, which delivers up to 2.5 times Standard speed at twice the price. API rates are now $0.20 per million input tokens and $1.20 per million output tokens for Luna, and $2 and $12 for Terra. Nice to see a company lowering prices instead of increasing.

  • We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control: Google launched Lyria 3.5 in Flow Music with improved melodic structure, lyric prompt adherence, vocal expression and pronunciation, plus more control over tempo and duration.

  • Google Blames Website For Race Based AI Overview Response: Barry Schwartz reports that Google said a disputed AI Overview response appeared to be picking up particular web pages for that query and that its teams were reviewing the result. Funny how Google was looking for copyright protection of the SERPs in the SerpApi case last week and now blames it on publishers 😂 

  • Siri AI could come with a paywall for power users: Tim Cook envisions iCloud+ upgrades that let heavy Siri AI users buy more compute. He said the plans are not final, but the model would mirror the free tier and paid-usage limits already common among AI providers. Apple plans a broader Siri AI rollout this fall.

CONTENT

Dan Petrovic reports that players correctly identified the machine-written passage 64% of the time across 1,636 answers in DEJAN’s first two days of testing. Accuracy rose from 55% for answers made within five seconds to 75% after 40 seconds or more, while 18 of 142 completed rounds were perfect scores against 0.14 expected by chance.

My Take: Let me know how you did. Just remember that humans actually use many of the AI indicators (em dashes) that people claim as clear AI detectors.

LINK BUILDING + DIGITAL PR

Vince Nero explains that outreach opens start with deliverability and relevance, not a clever subject line. In BuzzStream’s data across 1,785 domains and 14.3 million emails, domains with no authentication had open rates roughly four times lower than those with at least one authentication record. He also found that smaller campaigns earned higher average open rates, replies, and links.

SOCIAL MEDIA

Matt G. Southern reports that Metricool analyzed 799,718 YouTube videos from 71,177 connected accounts, comparing February 2025 with February 2026. Long-form views grew sharply in the dataset, while attention, ad delivery, and estimated ad revenue moved the other way.

  • Views rose 76%: The average long-form video received 5,985 views in February 2026, up from 3,405 a year earlier.

  • Attention weakened: Average view duration fell 37%, from 3.98 minutes to 2.51. Estimated minutes watched per post still rose 11% because the increase in views outweighed the shorter sessions.

  • Ad delivery dropped by about half: Average ad impressions fell from 976.32 to 475.07 per post, monetized playbacks from 576.41 to 237.92, and estimated ad revenue from $2.65 to $1.20.

Social Ripples

  • LinkedIn Adds An “AI Slop” Button To Clean Up Its Feed: LinkedIn is adding a “Seems like AI slop” feedback option, using reports as signals for classifiers that can reduce recommendations of low-quality, overly automated posts. LinkedIn is also removing its AI post-rewrite tool in favor of proofreading that preserves the author’s voice.

  • Create YouTube audio ads: Google Ads says audio-optimized YouTube placements can reach people listening to music, podcasts, or background content, with a static image or light animation as the recommended visual.

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

One idea shouldn't take six rewrites to post.

Posting everywhere means rewriting one idea six times, so you post to one, or none. SureThing turns one idea into native posts for every platform.

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.