- Digital Surfer
- Posts
- How We Finally Solved the Topical Hierarchy Problem
How We Finally Solved the Topical Hierarchy Problem
Wave 254; New Google Update; Readers Still Click AIO Links; Perplexity Wants to Pay Publishers; Gemini's Hot New Image Model; Newsjacking Strategy; and Much More!
FIRST …
We just released version 1.8.0 of Floyi. One of the biggest updates is something we’ve been quietly building and testing for months: a brand-new topical hierarchy methodology for topical maps.
On the surface, it sounds simple. Run embeddings, cluster the data, call it a hierarchy. But that is why most attempts fall flat. The clusters look clean, but too often they’re false positives. The similarity thresholds change from one topic to another. You end up endlessly tweaking numbers, and the result is the same: the topics don’t connect, the context breaks down, and all you’re left with is a giant list of keywords no brand or audience can use.
We took a different path. A hierarchy has to reflect your brand and your audience. If your audience doesn’t care about a topic, it does not belong. If the structure doesn’t match how your brand frames expertise, it will never build lasting authority.
Displaying that hierarchy is just as tough. Most platforms stop at two or three levels: a cluster here, a category there. Floyi is the only platform that organizes topics across four levels.
Think of it like college. Four years, four levels of courses. Only then do you master the subject. Authority works the same way.
And if you’ve taken my Topical Maps Unlocked course, you know how much work it takes to build a four-level map by hand. Now you can let Floyi 1.8.0 do the heavy lifting. You spend less time wrangling topics and more time building authority. Try Floyi out here.
With TopicalMap.com and hundreds of completed maps as our proving ground, we tested, refined, and rebuilt until we had a methodology that connects data to strategy, aligns brands with audiences, and mirrors how search engines reward topical authority today.
It took more trial and error than we expected. But the outcome is clear: a list of topics is noise. A hierarchy is a strategy.
If you want to see how it all comes together, I just uploaded a full walkthrough of Floyi - from Brand to Topical Maps to Briefs.

AD
The AI Wearable That Makes Your Life Unforgettable
Your greatest asset is your time. So stop wasting it jotting notes or chasing forgotten conversations.
The Limitless AI Pendant captures, transcribes, and organizes every meaningful interaction automatically. Get instantly searchable transcripts, smart summaries, and contextual reminders - all at your fingertips, all fully encrypted.
Tap into the future of productivity and free your mind to focus on what truly matters with Limitless.
SEO + GEO
Google released a Spam Update that will take a few weeks to complete.
Today we released the August 2025 spam update.
It may take a few weeks to complete, and we'll post on the Google Search Status Dashboard when the rollout is done:
status.search.google.com/incidents/a7Aa…— Google Search Central (@googlesearchc)
4:05 PM • Aug 26, 2025
Barry Schwartz reports a sharp crawl drop starting August 8 that hit select large sites across hosts like Vercel, WP Engine, and Fastly. John Mueller confirmed it was and issue on Google’s side, not just a reporting glitch, and rates are rebounding. Check Crawl Stats in GSC and server logs, verify robots.txt and IP blocks, resubmit sitemaps, and watch fetch time and 5xx errors. Compare Bing and platform status pages to rule out hosting issues. Document the anomaly for stakeholders and keep monitoring as crawl normalizes.
Charlotte Tobitt reports a Pollfish survey of 1,000 US adults, commissioned by NP Digital, showing most users still click through after an AI Overview. Only 4.4% never click, while 85.3% sometimes, often, or always do. Trust is mixed, 41.1% trust AI Overviews about the same as classic results, 31% more, 27.9% less, with 25.3% recalling major errors. Net satisfaction lands at 57.9%.
Andy Crestodina turns GA4’s messy acquisition soup into a playbook. Treat channels like oars, sails, and motors. Build GA4 Explorations to see spikes and winners. Tag every email and social link with UTMs. Many other takeaways, including:
Target keywords you can win based on authority
Publish one comprehensive page per topic
Refresh older pages near page-one rankings
Track AI chatbot traffic as referral
Expect GA4 to be about 20% inaccurate
Dark social hides roughly 80% of shares
Average bounce rate benchmark is 61%
Cloudflare rolls out AI Crawl Control, so publishers can negotiate instead of just block. The headline feature is customizable HTTP 402 Payment Required responses. Pick specific bots, return 402, and include a plain-English message like “Email [email protected] for access” or “Use our API at /pricing.” Paid plans can set this in the dashboard, creating a direct path from crawl to licensing. Use 402s to signal terms, protect archives, and route deals to a human. Cloudflare hints at richer 402 metadata next, while Pay Per Crawl advances in beta to enable usage-based monetization.
BrightEdge analyzed tens of thousands of identical prompts across ChatGPT, Google AI Overview, and Google AI Mode. The platforms disagreed on brand recommendations in nearly 62% of cases, with only 17% overlap across all three. Google AI Overview mentioned 2.5 times more brands than ChatGPT, while ChatGPT leaned heavily on brand mentions without citations. The takeaway from BrightEdge’s research is clear: visibility strategies cannot rely on one platform. To win, brands need to optimize for both trust-driven recommendations and source-backed citations across multiple AI search engines.
Dan Petrovic ran an experiment to see if schema markup improves how AI models like GPT-5 process web content. The verdict: schema didn’t make it through at all. OpenAI’s browsing tool only picked up visible text, ignoring JSON-LD and other structured data. That means schema isn’t a direct ranking or comprehension signal for AI today. The real benefit is indirect - schema boosts how Google and Bing display snippets, which then feeds the data AI pulls from search providers. Keep implementing schema for SEO, but don’t expect AI crawlers to use it natively just yet.
AI
The Perplexity team unveiled Comet Plus, a $5 subscription designed to fix the broken economics of the web in the AI age. Instead of relying on outdated pageview models, Comet Plus compensates publishers across three fronts: human visits, AI search citations, and agent-driven actions. For subscribers, it means direct access to premium journalism, smoother integration with AI assistants, and answers powered by trusted sources. For publishers, it offers a sustainable revenue stream and renewed brand trust.
My Take: These types of programs are always great in theory, but unless you’re a large publisher, I doubt there’ll be any significant income.
Google unveil a marketer’s Swiss Army image knife - nano-banana. Gemini 2.5 Flash Image blends multiple photos into one, keeps characters and products consistent across scenes, and performs pinpoint edits with plain English prompts. It taps world knowledge to follow complex instructions and power educational, catalog, and brand use cases. Every output carries SynthID watermarking for disclosure.
OpenAI shows off their new gpt-realtime model that improves natural speech, instruction following, and precise function calling. Connect phone workflows with SIP, ground conversations by adding screenshots with image input, and plug in tools instantly through remote MCP servers. Reuse prompts across sessions, keep chats fluid with asynchronous tool calls, and control spend with fine-grained token limits plus cached input. Pricing is $32 per 1M audio input tokens, $0.40 cached input, and $64 per 1M audio output tokens.
Michael Stelzner and Leslie Samuel shows marketers how to ship cinematic videos without studios or stress by using Google’s AI products. Start with intent, then storyboard and script with Gemini. Build “character DNA” for visuals and voice so faces and tones stay consistent across eight-second clips. Prototype every scene in Google Whisk to refine prompts cheaply, then generate in Veo 3 using the Fast model for most work and reserve Quality for hero shots. Stitch clips in Google Flow, upscale to 1080p, and add music and SFX in post.
Microsoft AI has unveiled two new in-house models. The first, MAI-Voice-1, is a lightning-fast speech generation system that delivers expressive, natural audio at scale and is already live in Copilot Daily, Podcasts, and Copilot Labs. The second, MAI-1-preview, is a mixture-of-experts foundation model trained on 15,000 H100 GPUs, now available for public testing on LMArena and rolling into Copilot text use cases.
Ben Collins shows how Google’s new AI function in Sheets can save hours of grunt work. Instead of exporting data to an external tool, you can now run prompts directly inside cells with =AI(). The function handles tasks like summarizing reviews, categorizing products, parsing messy addresses, generating marketing copy, and even unpivoting data. While it cannot replace core formulas like SUM or FILTER, it shines where language and context are needed.
Anthropic is testing a new Claude extension for Chrome that lets the AI interact directly with your browser, from drafting emails to scheduling meetings. They’re piloting with 1,000 Max plan users and have a waitlist. As useful as this sounds, the team is clear-eyed about risks. In red-team tests, Claude initially failed against prompt injection attacks 23.6 percent of the time, following hidden instructions to delete emails or take unsafe actions.
Thomas Germain reports that YouTube quietly applied AI cleanup to select Shorts, sharpening faces, smoothing skin, and sometimes warping details without telling creators. Musicians Rick Beato and Rhett Shull spotted subtle artifacts that made their videos feel synthetic. After weeks of speculation, YouTube’s Rene Ritchie confirmed an “experiment” using machine learning to unblur and denoise during processing while offering no clarity on opt outs.
On X, Brave warns that agentic browsers can be tricked by hidden on-page instructions, not just user prompts. In tests, Comet obeyed malicious commands buried in Reddit content, then tried to access authenticated data. Perplexity has shipped a fix that doesn’t fully work. Practical steps you can use now to protect yourself: segment logins by profile, keep finance in dedicated apps, disable password autofill on sensitive sites, restrict agent permissions, and force explicit consent for email, data export, or account actions. Treat every summarized page as untrusted input.
AI Ripples
Google Translate added AI-powered features designed for real-world use. The app now supports live conversations in over 70 languages with instant audio and on-screen translations, even in noisy environments. Google is also piloting a language practice tool that adapts to your goals and skill level, offering personalized speaking and listening exercises built with input from learning experts.
Anthropic has reached a proposed settlement with US authors who accused the company of training its Claude AI models on millions of pirated books. The class action alleged “Napster-style downloading” that could have cost Anthropic billions. A June ruling had allowed training on legally purchased books under fair use but left piracy claims open. With trial looming in December, Anthropic avoided a courtroom showdown.
LINK BUILDING
Emma Kessinger explains how brands can turn trending stories into strategic wins. Newsjacking is no longer a gimmick. Done right, it is one of the fastest ways to earn backlinks, media mentions, and authority. Timing is everything: aim to publish within hours of a breaking story using tools like Google Trends or X’s trending tab. Pick stories that align naturally with your brand and avoid tone-deaf jumps. Prepare formats in advance - press releases, SME commentary, or quick social posts - to move fast. Build a repeatable system with metrics, templates, and clear spokesperson roles.
SOCIAL MEDIA
Jennifer Ouellette covers a new survey of 813 scientists showing a clear migration from Twitter to Bluesky. Since 2022, respondents say Twitter’s utility for networking, learning, and outreach has dropped sharply amid spam, weaker moderation, and lower engagement. Nearly 40 percent deleted Twitter, many report using it far less, while Bluesky earns higher quality interactions despite smaller followings. Referral data backs it up, with Bluesky driving dramatically more page views for science blogs this year.
WAYS WE CAN WORK TOGETHER
Floyi - The only AI-powered tool that builds 4-level topical maps. Don’t just plan your content strategy - make it unstoppable.
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
Download our guide on AI-ready training data.
AI teams need more than big data—they need the right data. This guide breaks down what makes training datasets high-performing: real-world behavior signals, semantic scoring, clustering methods, and licensed assets. Learn to avoid scraped content, balance quality and diversity, and evaluate outputs using human-centric signals for scalable deployment.
What Did You Think of This Week's Wave? |
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