AI Crawlers, Vanishing Traffic & What Still Works in SEO | HighDegree* – Issue 15

Subscribe to HighDegree* Marketing Newsletter!

International SEO changed. Google traffic crashed. Here’s what matters now.


Hi friends,

Welcome to the new issue of HighDegree*: Cutting Through the Noise in SEO & Digital Marketing

The SEO landscape shifted hard in 2025. Publisher traffic from Google dropped by a third. AI bots now crawl more sites than ever while training bots get blocked left and right. Meanwhile, the old international SEO playbook stopped working the moment AI-mediated search became the norm. This week’s HighDegree* Marketing Newsletter breaks down what’s actually happening, what still moves the needle, and how to think about SEO when the rules keep changing.

Let’s jump in.


➞ In this Week

  • International SEO Collapsed in 2026, Here’s What Survived
  • Google’s Court Filing Confirms User Data Drives Rankings
  • AI Bots Are Crawling Half the Web While Training Bots Get Blocked
  • Publisher Traffic from Google Dropped 33% Year Over Year
  • Good SEO vs. Bad SEO: A Product-Minded Framework
  • Articles that will Help You Refine Your Marketing Knowledge
  • Everything from Social Media and Artificial Intelligence World
  • Everything from Google Search this Week

➞ International SEO Collapsed in 2026, Here’s What Survived

The traditional international SEO playbook, dedicated URLs, localized content, hreflang tags, isn’t enough anymore. AI-driven search systems now retrieve and collapse multilingual content into shared semantic representations before ranking even happens. Translation-only pages rarely get picked up because AI models favor the “most confident version” of a concept, often defaulting to English.

What still works: market-scoped URLs with real differences like pricing, legal disclosures, shipping rules, or availability. Hreflang still prevents duplication in traditional search, but it can’t influence which version AI systems choose during synthesis. The new priority is entity clarity, making it unmistakably clear which organization, brand, product, and market context applies.

This means stable naming, predictable URL patterns, consistent internal linking, and structured data that reflects actual business relationships. Local authority also matters more than ever. AI systems evaluate trust within a market context, checking for in-country experts, local certifications, and regional citations.

Global brand authority alone won’t cut it. Bottom line: stop optimizing page by page. Start governing content as a system that proves which version of your business should be trusted in each market.

Read the full report from searchengineland.com ➞

Now let’s look at what Google’s own court documents reveal about how ranking actually works.


➞ Google’s Court Filing Confirms User Data Drives Rankings

A recent declaration from Google’s Liz Reid in the DOJ antitrust case confirms what many SEO professionals suspected: user behavior data is central to Google’s ranking systems. Google’s GLUE system collects massive amounts of data on every query, what users searched, what appeared in results, what they clicked, how long they stayed on pages, and whether they returned to search.

This data trains RankEmbed BERT, one of the deep learning systems that reranks traditional search results. The system continuously learns from what users click and stay on, then fine-tunes based on quality rater feedback. Google also marks every indexed page with proprietary annotations including spam scores and freshness signals. If spammers got hold of those spam signals, Google argues, it would become much harder to fight manipulation.

The filing also reveals that only a fraction of the web makes it into Google’s index, and Google organizes that index based on how frequently content needs to be accessed and how fresh it needs to be.

The takeaway for site owners is clear: user satisfaction matters far more than traditional ranking factors. If people click your result and stay, you win. If they bounce back to search, you lose.

Read the full report from mariehaynes.com ➞

Meanwhile, the broader web crawling landscape is changing fast.


➞ AI Bots Are Crawling Half the Web While Training Bots Get Blocked

An analysis of 66.7 billion bot requests across 5 million websites shows a clear split in how AI crawlers behave. Assistant-facing bots that power ChatGPT, Apple’s Siri, TikTok Search, and Petal Search are rapidly expanding their reach, now accessing over half of all monitored websites. These bots fetch content on demand to answer user queries, not to build training datasets.

At the same time, LLM training bots like OpenAI’s GPTBot and Meta’s ExternalAgent are getting blocked at record rates. GPTBot’s coverage crashed from 84% to just 12% of sites as content creators refuse to let their work train AI models for free. Traditional search engine bots remain stable and predictable, with Google’s main crawler still reaching 72% of sites on average.

SEO and monitoring tools like Ahrefs and SEMrush are shrinking in coverage as they focus on actively optimized sites.

The emerging strategy: allow assistant bots that drive traffic while blocking training bots that extract value without attribution. Website owners now face a choice, optimize for AI-driven discovery or protect proprietary content. The middle path appears to be gaining traction.

Read the full report from hostinger.com ➞

This crawler shift is happening as publisher traffic from Google collapses.


➞ Publisher Traffic from Google Dropped 33% Year Over Year

New Chartbeat data covering 2,500 publisher websites shows Google search referrals declined 33% globally in the year ending November 2025. In the US alone, organic search traffic was down 38%. Google Discover referrals dropped 21% globally and 29% in the US.

The decline hit hardest for publishers focused on lifestyle and utility content like weather, TV guides, and horoscopes, exactly the content Google now summarizes in AI Overviews at the top of search results. Publishers expect traffic to decline by another 43% on average over the next three years due to AI summaries. Most now plan to put less effort into traditional Google search optimization. Social referrals remained flat or slightly up, with X up 15% globally and Facebook up 9% after it began resurfacing news content in early 2025.

ChatGPT referrals are rising rapidly but still represent just 0.02% of total traffic. To adapt, publishers are shifting focus to original investigations, on-the-ground reporting, and contextual analysis rather than service journalism or evergreen content.

The strategy is clear: create what AI can’t easily replicate.

Read the full report from pressgazette.co.uk ➞

Given all this disruption, how should SEO teams actually operate?


➞ Good SEO vs. Bad SEO: A Product-Minded Framework

Modern SEO is a product discipline, not a marketing channel. Good SEO aligns strategy with business goals like revenue, leads, and retention, not just traffic or rankings. It deeply understands users, their problems, and search intent, then designs pages around jobs to be done rather than chasing keyword volume.

Björn Darko writes, Good SEO frames problems clearly, separates symptoms from root causes, and prioritizes work using impact versus effort frameworks. It treats every change as a hypothesis, measures results, and iterates based on learning. Most importantly, good SEO works embedded with product, engineering, and marketing teams, influencing roadmaps instead of reacting to them.

The shift is fundamental: good SEOs think in products and systems, creating reusable playbooks that enable teams to do SEO without constant hand-holding. They build trust through clarity and focus on high-leverage changes that create measurable business value.

Read the full blog from Björn Darko ➞


➞ From Google

Everything from Google search this week —

Google Local Hotel Photos “Good To Know” AI Labels (seroundtable.com)

Google Hit Self-Promotional Listicles In Recent Unconfirmed Updates? (seroundtable.com)

Google Search Ranking Volatility Heats Up Big Time Starting On February 2nd (seroundtable.com)

Google’s February 2026 Discover Core Update (developers.google.com)

Google Personal Intelligence Small App Icons In Responses (seroundtable.com)

Google: Search Algorithms, Spam Detections & Policies Don’t Fundamentally Change With AI Search (seroundtable.com)

Googlebot File Limit Is 15MB But 64MB For PDF & 2MB For Other File Types (seroundtable.com)

Google On Serving Markdown Pages To LLM Crawlers (seroundtable.com)

Google Ads tightens access control with multi-party approval (searchengineland.com)

Google AdSense Reports Gain Browser, Hosting App & Operating System (seroundtable.com)

Google Search Adds Preferred Sources Help Docs (seroundtable.com)

Google On Recent Google Search Ranking Volatility – No Insights To Share (seroundtable.com)

Google: Don’t Spend Too Much Time On Redirects Analysis For SEO (seroundtable.com)

Google: Search Algorithms, Spam Detections & Policies Don’t Fundamentally Change With AI Search (seroundtable.com)

Google Search Console Adding AI Visibility Reporting? (seroundtable.com)

Google Vastly Increases Local Pack Ads Placement (seroundtable.com)

Google tests third-party endorsements in search ads (searchengineland.com)

Social Channel Insights In Search Console: What It Means For Social & Search (searchengineland.com)

Google may give sites a way to opt out of AI search generative features (searchengineland.com)

The new era of browsing: Putting Gemini to work in Chrome (blog.google)

Google’s approach to website controls for Search AI features (blog.google)

Google launching the Ads Decoded Podcast to connect advertisers with the people building Google Ads (blog.google)

Personal Intelligence in AI Mode in Search: Help that’s uniquely yours (blog.google)

Google Search Team Does Not Endorse LLMs.txt Files (seroundtable.com)


➞ AI + Social

Find out what’s happening in the social media and artificial intelligence world —

Microsoft Publisher Content Marketplace – Pay For AI To License Content (seroundtable.com)

Meta Tests a Standalone AI Video App (socialmediatoday.com)

Meta Is Working on a Dedicated AI Chatbot for Group Chats (socialmediatoday.com)

X Tests Collaborative AI-Powered Community Notes (socialmediatoday.com)

X Adds Images to In-Stream Polls (socialmediatoday.com)

TikTok Offers Enhanced Shops Promotion Program (socialmediatoday.com)

Perplexity Model Council for all Perplexity Max users on web (linkedin.com)

Anthropic says Claude will remain ad-free as ChatGPT tests ads (searchengineland.com)

Introducing GPT-5.3-Codex—the most powerful, interactive, and productive Codex yet (openai.com)

ChatGPT With Top Stories & More Visual Knowledge Panels (seroundtable.com)

ChatGPT and Google Gemini increasingly cite Elon Musk’s Grokipedia (thenews.com.pk)

EXCLUSIVE: OpenAI Confirms $200,000 Minimum Commitment for ChatGPT Ads (adweek.com)

OpenAI Prepares ChatGPT Code To Allow Ads? (seroundtable.com)

Bing Multi-Turn Search Rolls Out Worldwide (seroundtable.com)

Elon Musk’s xAI launches Grok Imagine 1.0 amid sexualised images controversy (indiatimes.com)

With Moltbook, AI agents build their own civilisation (indiatimes.com)

Meta Outlines AI-Powered System Improvements (socialmediatoday.com)

Meta Announces Discovery Improvements for Creator Marketplace (socialmediatoday.com)

TikTok Shares Insight Into Effective Ad Measurement Approaches (socialmediatoday.com)

LinkedIn Continues To Add Users, Revenue (socialmediatoday.com)

Introducing Moltworker: a self-hosted personal AI agent, minus the minis (blog.cloudflare.com)

OpenAI launches standalone Codex app for Apple computers (cnbc.com)

Bing Webmaster Tools testing new AI Performance report (searchengineland.com)

Clawdbot Security Risks: What You Need to Know Before Giving Access to Your Email (getinboxzero.com)

Retiring GPT-4o and other ChatGPT models (openai.com)

Grok Imagine API (x.com)

Check out what YouTube has planned for 2026. (blog.google)

From Cool Blue to Persimmon: Meet the 2026 Pinterest Palette™ (pinterest.com)


➞ Worth Reading

These are the articles that will help you refine your marketing knowledge —

Framework for AI-Generated Content That Doesn’t Suck (sample.com)

Year in Email 2025.
Fun edition
(omnisend.com)

Are your pop-ups tanking your rankings? A guide to interstitials and dialogs (searchengineland.com)

Visual Fan-Out: Make Your Products and Destinations Discoverable in AI Mode (wordlift.io)

NEW Research: AIs are highly inconsistent when recommending brands or products; marketers should take care when tracking AI visibility (sparktoro.com)

Unlearning Old SEO Content Strategies to Improve Brand Visibility | Himani Kankaria (advancedwebranking.com)

Your SEO backlog is costing you revenue. Here’s why marketing needs to own it. (salt.agency)

The future of search visibility: What 6 SEO leaders predict for 2026 (searchengineland.com)

State of Search Q4 2025: Behaviors, Trends, and Clicks Across the US & Europe (datos.live)


➞ What This All Means for Your Strategy

The common thread across all five pieces: SEO is no longer about manipulating rankings or gaming algorithms. It’s about creating genuine value, building trust at scale, and understanding how AI systems retrieve and synthesize information.

Whether you’re running an international site, a publisher property, or a small business blog, the fundamentals remain the same; solve real problems for real users, then make it unmistakably clear to both humans and machines why your content deserves attention.

If you found this useful, forward it to a colleague who’s navigating these changes. And as always, hit reply with questions or topics you’d like us to cover next.

Until next week,

Nishat from HighDegree*

Thanks for reading HighDegree*! This post is public, so feel free to share it.


P.S. Have a question about implementing these strategies? Hit reply – we read every email and often feature reader questions in future issues.


➞ Who is Nishat Shahriyar?

I am a Digital Marketing Strategist, having worked in this field since 2007. Now working as a Product Marketing Lead at Fluent Forms (The best lead generation tool for WordPress), previously at Fluent Support (The best WordPress Helpdesk Plugin).

Connect with me, if you are not connected through my LinkedIn or follow me on X/Twitter – @rednishat

Subscribe to HighDegree* Marketing Newsletter!


Posted

in

by

Tags:

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *