After nearly 30 years at Microsoft, I am seeing one of Bing’s most influential search leaders close a remarkable chapter. Fabrice Canel announced that he is retiring from Microsoft, writing on LinkedIn, “I am retiring from Microsoft, effective today July 1st.” He also reflected, “Today marks nearly 30 years with Microsoft. Thirty years…”
When I think about Fabrice Canel’s impact, I think first about the foundation of Microsoft Bing Search. He was responsible for indexing at Bing, including crawling, URL discovery, content selection, and content processing. Those areas are core to how search engines understand the web, and Fabrice helped shape them at massive scale.
He was also the person behind the IndexNow initiative, and he played a major role in creating and powering Bing Webmaster Tools. For anyone working in SEO, publishing, or technical search, those contributions matter because they helped make discovery, indexing, and webmaster communication faster and more practical.
I have watched Fabrice contribute far beyond product work. He has spoken at countless industry events, including SMX, and has written extensively about how search works, how sites can perform better in Bing, and how search is evolving with generative AI. He helped run one of the world’s most important search engines, while also giving the SEO community tools, education, and direct insight.
In his retirement message, Fabrice addressed fellow Microsoftees, engineers, attorneys, marketers, webmasters, publishers, SEO champions, product leaders, journalists, people across search and AI, and even friends at Google. His note was warm, personal, and full of gratitude for the people who shaped his Microsoft journey.
He described his three decades at Microsoft as a wonderful adventure, from solving real business problems with IndexNow to helping webmasters and publishers thrive in the constantly changing world of SEO and AI. He thanked colleagues, partners, publishers, and the people he trained and mentored, saying they are ready to carry the mission forward.
Fabrice also shared that, after many conversations with family and friends, he decided to take advantage of Microsoft’s Voluntary Retirement Program. His message ended with the same sense of warmth and storybook style that many in the industry have come to associate with him: gratitude for Microsoft, confidence in the Bing team’s future, and a final wish that everyone stay curious, keep innovating, and make content easier to find.
Why do I care so much about this? Because Fabrice has been a true friend to the search industry. His work will live on through the products, systems, and initiatives he helped create, and his willingness to share knowledge has made a lasting difference for SEOs, publishers, developers, and search professionals.
I know Fabrice has trained a team to continue the work, and I believe Bing remains in good hands. Still, I would be lying if I said I am not sad to see him retire. It has been an honor to work with him and learn from him over the years, and his legacy at Microsoft Bing will be felt for a long time.
I’m seeing Google make recipe results in AI Mode more publisher friendly with a new visual treatment that gives recipe creators more visibility. For some recipe responses, Google is now showing details such as the creator name, recipe ratings, and the number of ingredients.
What is new. Google’s Robby Stein said AI Mode now includes “prominent links at the top of responses with useful details and images,” including creator names, ratings, and ingredient counts. From my view, the key shift is that Google is trying to make recipe sources easier to recognize and visit directly from AI Mode.
What it looks like. The new treatment places recipe links, images, and useful recipe details more prominently in the AI Mode experience, giving users a clearer path from the AI-generated response back to the original recipe page.
Previously. Back in March, Robby Stein announced earlier changes to recipe results in AI Mode. At the time, he said Google had heard feedback and was making updates to better connect people with recipe creators across the web.
I see this latest update as part of Google’s effort to address concerns around AI recipe slop and to make original recipe content more visible when people search for cooking ideas through AI-powered results.
Why I care. Recipe bloggers, and content creators more broadly, have been frustrated that Google’s AI experiences often send less traffic than traditional search results. This update suggests Google is trying to encourage more searchers to click through from AI Mode to the publishers and creators behind the recipes.
If Google continues adding more clickable link units into AI search experiences, I think it could help ease some of the tension between publishers and Google. The bigger question is whether these changes will drive enough meaningful traffic back to recipe sites and other content creators.
I see generative AI and automation creating both excitement and anxiety across the SEO industry. With 87% of Americans reading AI summaries, I believe any SEO team that is not adapting its toolset is already starting to fall behind.
When I move away from rigid enterprise tools and toward agile, AI-driven workflows, I can work faster, spot new search signals earlier, and show clients or internal stakeholders that I understand where search is heading.
In this guide, I’ll walk through what the old SEO stack looked like, what I now add to it, and how I combine both approaches without abandoning the fundamentals that still matter.
Here’s what an old SEO stack looks like
I still believe traditional SEO practices matter because generative AI search experiences continue to depend on core search ranking systems, quality systems, and the broader signals search engines have used for years.
That said, the classic SEO stack was built for a simpler search environment. It usually centered on rank tracking, keyword research, and technical site audits.
Rank trackers
For a long time, I treated keyword rankings as the heartbeat of an SEO campaign. I would add target keywords, monitor SERP positions, and expect higher rankings to translate into more search traffic. But rankings have become far more fragmented.
Now I need to pay attention to AI Overviews, local packs, shopping carousels, and many other search features that can change the value of a ranking completely.
A third-place local pack ranking, for example, may drive two or three times more traffic than a number one ranking in an AI Overview. That makes old-school rank tracking useful, but incomplete.
Keyword tools
Keyword tools still help me understand what people search for, how competitive a topic might be, and which queries match specific user intent. In the past, that information often felt close to a crystal ball.
I would choose keywords based on difficulty, search volume, intent, and other factors. The better the data, the easier it was to shape a campaign around the right opportunities.
The problem is that search volume has always looked backward. A keyword may have shown 10,000 monthly searches last month, but that does not mean it will perform the same way this month. Demand can rise, fall, or shift quickly.
Today, the bigger issue is opportunity loss. A keyword that generated tens of thousands of clicks in 2022 may now be answered directly inside an AI Overview. Even when search volume has not dropped, zero-click behavior can reduce the traffic I can realistically capture.
Site audit tools
I still rely on site audit tools because crawlers still crawl websites, interpret content, and surface technical issues. I need to know whether search engines can access, understand, and navigate the pages I care about.
Audit tools help me find broken links, redirect problems, missing metadata, slow pages, thin content, and other technical issues that can hold a site back.
But I do not expect crawl audits alone to tell me whether my content will appear in AI-driven search experiences. Technical health is necessary, but it is no longer the full picture.
Signals such as brand mentions can influence whether a site is included in LLM outputs from tools like ChatGPT, Claude, and Gemini. Many older site audit tools were not built to track those signals.
That is why I still keep parts of the old stack, but I now add tools and workflows that help me understand AI visibility, brand presence, and faster data-driven decision-making.
Here’s what a new SEO stack looks like
If I am optimizing only for Google’s traditional results, I am missing where search behavior is moving. Between the first and second half of 2025, LLM referral traffic grew by 80%. Conversion rates reached 18%, even though LLM referrals still represented 2% or less of total traffic in the dataset.
That tells me the channel is still small, but meaningful. Now is the time to build a stack that helps me understand, measure, and improve performance across AI-driven discovery.
LLMs
I want my site to appear in LLM responses, but I also use LLMs to strengthen my SEO process. These tools can support analysis, content review, competitor research, metadata refinement, and structured data work.
For example, I can connect ChatGPT with Google Search Console to automate SEO analysis, use Claude to refine copy and conduct content audits, or use Gemini to generate schema markup and compare competitor pages against my own.
I use the LLM that best fits the task, but I keep human oversight in place. These tools help me improve speed and performance; they do not replace judgment, strategy, or editorial review.
The biggest shift is speed. Large datasets that once took hours, days, or weeks to review can now be explored in minutes when I use LLMs carefully and integrate them into a repeatable workflow.
APIs
The old workflow often meant logging into dashboards, exporting CSV files, and cleaning everything in Excel. I still do that when needed, but APIs let me pull data directly from platforms like Google Search Console and Google Analytics.
APIs can sound intimidating, but LLMs make the learning curve easier. I can use them to help with authentication, JSON parsing, and the basic structure of repeatable data workflows.
Once I can connect to APIs, I can stop waiting on manual exports and start building faster reporting, monitoring, and analysis systems around the data I already use.
Lightweight scripts
Python scripts are now within reach for many SEOs, especially with tools like Claude Code and similar coding support inside ChatGPT or Gemini. I do not need to be a full-time developer to automate repetitive SEO work.
I can create scripts that pull top pages from Google Search Console, compare title tags against character limits, flag 30-day performance changes, or generate a clean CSV output for review.
Instead of waiting for a vendor to add the exact feature I need, I can build a small script that removes a bottleneck. A hundred-line script can replace hours of manual work without requiring another SaaS license.
I also like that scripts make the logic visible. If I hand the workflow to another teammate, they can inspect what the script does and understand how the output was created.
Notebooks and local workflows
SEO teams usually have data scattered across shared folders, Google Sheets, Notion docs, monthly CSV dumps, and long-running audit trackers. I have seen how quickly that fragmentation slows decisions down.
Notebooks and local workflows help me turn scattered files into a working system. A script can pull the data, an API can surface the signal, and an LLM can help interpret the results before the output lands in a notebook or spreadsheet.
The value is consistency. I get cleaner data formats, shared access, and documented logic instead of rebuilding the same process every time someone needs a report or audit update.
As search optimization becomes more connected to generative AI, I need workflows that scale. Local workflows help me keep data consistent while giving the team a faster way to act on what we find.
Creating hybrid workflows that mix old and new SEO stacks
I do not think the old SEO stack is obsolete. I also do not think the new tools replace everything. The strongest approach is a hybrid workflow that keeps proven SEO fundamentals while adding AI, APIs, scripts, and notebooks where they create real leverage.
Tool + custom script + AI layer
To build a practical hybrid workflow, I would start with a familiar audit tool such as Screaming Frog, then run a Python script that joins the crawl data with Google Search Console data.
From there, I could flag pages with high impressions and low clicks, send those pages to an LLM for title and intent analysis, place the output into a notebook or spreadsheet for editors, and turn approved recommendations into change logs.
Work like this used to take weeks, so many teams pushed it aside. At enterprise scale, the amount of data could easily become overwhelming. With a hybrid SEO stack, I can complete larger projects in a fraction of the time.
For me, the goal is not to chase every new tool. The goal is to build a more agile SEO stack that can handle today’s massive datasets, identify AI search signals, and help teams move faster without losing the core SEO basics.
When one person is responsible for paid campaigns, landing pages, reporting, email, social posts, sales requests, and last-minute website updates, I know exactly what usually happens to SEO: it waits.
I have seen this play out on small marketing teams over and over. Everyone knows SEO can bring in qualified demand, reduce dependence on paid media, and support buyers long before they fill out a form. The problem is that SEO rarely feels urgent until traffic drops, rankings slide, or something breaks.
That is why I like a simple 120-minute weekly SEO workflow. It gives me a practical way to protect visibility, find opportunities, improve high-value pages, and turn search data into business impact without pretending I have unlimited time.
Why I keep SEO simple on lean teams
When SEO falls behind, I rarely see effort as the real problem. The bigger issue is usually competing priorities and a lack of clear prioritization.
On a lean team, SEO is one tab among 20. The person responsible for organic growth may also be sending newsletters, briefing designers, updating landing pages, and pulling the report leadership wants by Friday.
Then the advice starts piling up: fix technical issues, publish more, build topical authority, refresh old posts, add schema, improve Core Web Vitals, build links, optimize for AI search, and keep going. Most of that advice may be valid, but no small team can do all of it in one week.
The question I come back to is not, “What could I do?” It is, “What is the highest-leverage thing I can actually finish this week?”
I also try to avoid the reporting trap. It is easy to spend an entire SEO block looking at rankings, traffic, impressions, clicks, CTR, conversions, competitor movement, and keyword shifts. Then the hour ends and nothing ships.
For a small team, reporting has to be short enough to leave room for action. The goal is to decide what to fix next, not to build another dashboard.
Why 120 minutes can be enough
I do not try to run a lean team like an enterprise SEO department. If I audit everything, track everything, collect endless keywords, and ship nothing, I have not improved organic growth.
The point of time-boxing is to force a decision. Every weekly session should end with one or two changes that improve visibility, traffic quality, or conversion potential.
In my 120-minute workflow, I focus on four outcomes: finding what is already working, fixing what is blocking performance, improving the pages closest to revenue, and turning search data into next week’s actions.
I am not trying to “do SEO” for two hours. I am using two focused hours to make decisions and ship work that has a realistic chance of moving the business forward.
My 120-minute weekly SEO workflow
0-15 minutes: Check organic data
I start with a pulse check so I can catch problems before they turn into bigger performance drops.
I look at Google Search Console clicks, impressions, CTR, and average position. I also check organic conversions or assisted conversions in GA4, top landing pages gaining or losing traffic, branded versus non-branded movement, and any indexing, crawling, or manual action warnings.
What I do not do is turn this into a full reporting session. This is not a board deck. I only want to answer one question: is organic visibility moving in a direction that needs action?
My output is a short weekly note: the biggest organic win, the biggest organic concern, one page or query to investigate, and one action to take this week.
15-35 minutes: Find query opportunities
Next, I look for the easiest opportunities in Google Search Console. The richest ones are often queries ranking in positions 4-15 with real impressions. Those pages are already close, and a focused improvement can help them move.
I also watch for pages with strong impressions but weak CTR, queries climbing week over week, and rankings where the current page only partially matches search intent.
I resist the urge to build a long keyword list. Instead, I pick three things: one page to improve, one query to answer better, and one title or meta description to test.
For example, when I reviewed search data for a local accounting client, several queries kept appearing around tax help for freelancers, small-business tax mistakes, and the difference between an accountant and a bookkeeper.
The obvious reaction would have been to write three new articles. Instead, I rewrote one service page around freelancers, added a short FAQ based on those queries, and linked it to an existing bookkeeping article. One page served three search intents, which was far more useful than three unfinished drafts.
35-60 minutes: Improve one money page
This is the most important part of the workflow. I define a money page as any page close to revenue, pipeline, bookings, sales, demos, or consultations.
Money pages can include product pages, service pages, category pages, comparison pages, demo pages, consultation pages, pricing pages, and high-intent landing pages.
My weekly goal is not to optimize the entire website. It is to improve one important page in one meaningful way.
I ask what the buyer needs to believe before converting, what objection is missing, what proof would reduce hesitation, what comparison the buyer already has in mind, and what query the page almost satisfies but does not fully answer.
A meaningful update might be adding three FAQs based on real queries, improving the H1 and introduction, adding comparison language, including proof points, linking to a case study, clarifying who the offer is for, improving the CTA, or adding a short “how it works” section.
That is SEO work, but it is also conversion work. The best page improvements usually help both search engines and buyers understand the value faster.
60-80 minutes: Fix one technical or indexing issue
Technical SEO can take over the full two hours if I let it, so I stay focused on impact.
The question I ask is simple: what could stop an important page from being discovered, understood, indexed, or trusted?
That usually points me toward issues like priority pages not being indexed, broken internal links, redirect chains, duplicate or missing titles on key pages, incorrect canonicals, schema errors on important templates, or valuable pages buried too deep in the site.
I want one of three outcomes from this block: a fix shipped, an issue assigned, or a clear developer brief.
For example, if I find that ecommerce collection pages are not indexed because of incorrect canonical tags, documenting the affected URLs and writing a clear developer brief may be more valuable than publishing another generic article.
80-100 minutes: Improve internal links
Internal linking is one of the fastest SEO wins I can create because it does not require new content.
It helps search engines understand which pages matter, helps users continue their journey, and helps informational content support commercial outcomes.
Each week, I look for links from high-traffic articles to money pages, links from product or service pages to supporting guides, links from older articles to newer strategic content, and opportunities to use clearer anchor text.
If an article ranks for “how to choose accounting software,” I do not want it to be a dead end. I want it to guide readers toward a comparison guide, a relevant case study, and a demo or pricing page. The traffic is already there, so I try to make it more useful.
100-115 minutes: Turn one search insight into messaging
I do not want search data to stay trapped in an SEO silo. The best query I find each week is often a useful signal for the rest of marketing because it shows the language buyers actually use.
A query like “best CRM for small agencies” can become a comparison section on a landing page, a LinkedIn post, a sales email angle, and a paid search ad group.
A query like “is [product] worth it” can become a proof section, a pricing explainer, a “who this is not for” paragraph, or a ready-made answer to a sales objection.
When I share one search insight each week, SEO becomes more than a channel. It becomes a source of customer intelligence.
115-120 minutes: Choose next week’s priority
I end with a decision, not a long list. I choose one clear priority for next week based on business impact, search demand, ease of execution, current performance gap, and proximity to revenue.
The template I use is: “Next week, my highest-leverage SEO action is [X] because [Y].”
For example: “Next week, my highest-leverage SEO action is updating the pricing page because it gets non-branded traffic, supports demo requests, and does not answer implementation cost questions.”
That is how I make SEO operational. The work becomes specific, owned, and easier to repeat.
A sample month for the workflow
To keep the workflow balanced, I like rotating the emphasis each week.
In week one, I focus on a revenue page. I update copy, add FAQs, improve internal links, check indexing and schema, and sharpen the CTA.
In week two, I refresh existing content. I choose one article with impressions but weak clicks or rankings, improve the title, add missing sections, update examples, link to money pages, and better match search intent.
In week three, I handle technical cleanup. I focus on one crawl, indexing, or template issue, such as broken links, duplicate titles, sitemap problems, or a developer brief for a higher-impact fix.
In week four, I turn SEO data into broader marketing assets. That may mean one landing page insight, one sales objection, one content brief, one paid or social angle, or one FAQ or comparison section.
This rotation keeps me from spending every week in dashboards, technical audits, or new content production while ignoring the pages that already have potential.
What I stop doing
Most small teams do not have a doing problem. They have a stopping problem.
I stop chasing every low-impact technical warning. I stop creating content just because a tool found a keyword. I stop publishing AI-assisted articles at scale without a strategy. I stop rewriting pages without a hypothesis. I stop optimizing low-value pages before revenue pages. And I stop treating rankings as the only score that matters.
Before I create new content, I review the pages I already have. The highest returns often come from pages that already rank on Page 2, already get impressions, sit close to revenue, and are one focused update away from doing more.
My test for any task is simple: if I cannot connect it to qualified traffic, conversions, discoverability, buyer education, or trust, it does not belong in the 120 minutes.
How I make it work without a dedicated SEO person
This workflow does not require a full SEO department. It requires one owner, a weekly rhythm, and a bias toward shipping.
A marketing manager can own prioritization and the weekly SEO note. A content marketer can update copy, FAQs, and page sections. A developer or web support partner can handle technical fixes. A paid search manager can share query and conversion insights. A founder or sales team can contribute objections and buyer language.
The owner matters most. Someone has to protect the 120 minutes, choose the priority, and make sure the session ends with an action.
Without ownership, SEO becomes everyone’s job and nobody’s job.
How I use AI to save time
I use AI to shorten repetitive SEO work, not to hand over strategy.
That might mean using a focused workflow to identify queries in positions 4-15, pages with high impressions and low CTR, search queries that should become FAQs, internal linking opportunities, or technical issues that should become developer briefs.
For agencies, client-specific assistants can reduce context switching by remembering each client’s services, priority pages, competitors, and customer objections.
The most useful AI workflows are narrow: a GSC opportunity analyzer, a money page refresh assistant, an internal linking assistant, a technical SEO brief generator, or an SEO reporting summarizer.
I do not want one generic SEO assistant trying to do everything. I want small workflows that help me move faster from data to decisions.
Consistency is the advantage
Small teams win SEO by doing the highest-leverage things repeatedly.
A 120-minute weekly SEO workflow will not replace a full strategy. It will not solve every technical issue, build every content asset, or uncover every opportunity.
But it gives me a practical way to protect visibility, learn from search data, improve revenue pages, and keep organic growth moving.
The mindset is simple: less auditing, more shipping, more buyer intent, less busywork, and more business impact.
I’m deeply saddened to share that Bruce Clay, widely known as the Father of SEO, passed away in late May. Bruce was one of the true founding figures of the SEO industry, having launched a professional SEO agency back in 1996, long before search marketing became the discipline we know today.
For me, Bruce’s impact is hard to overstate. He was the first sponsor of the first-ever SEO conference, and he gave an extraordinary amount of his time, resources and money to help build the search community. Few people have supported this industry for as long, or with as much generosity, as Bruce did.
Tribute video. The Bruce Clay, Inc. team prepared a tribute video honoring Bruce’s life and legacy. It describes him as a pioneer who devoted much of his life to helping the SEO industry grow. During his three decades as CEO of Bruce Clay Inc., he wrote three books, built tools, spoke at conferences, hosted training events and helped the company expand internationally.
Because of Bruce’s founding principles, hundreds of employees around the world have contributed to SEO, and thousands of students have benefited from his experience and teaching.
The Bruce Clay team told me, “We are absolutely heartbroken, but we find strength in the vibrant community and lasting values that Bruce built. Our teams in the U.S. and around the world remain dedicated to carrying forward the mission Bruce loved so dearly.”
Kyle Pouliot, Sr. Video Production Manager at Third Door Media, also shared a personal reflection with me.
“I’ve gotten to know Bruce on a more personal level over these past few years and interacted with him frequently for our online conferences. What I’ve learned about Bruce in that time is that he was genuinely thoughtful and caring about the search community. Never short of an honest opinion, Bruce shared some really practical ideas for Search Engine Land and SMX. He loved sharing his deep experienced knowledge to everyone, it didn’t matter if you were a beginner or 20+ year industry veteran, he treated everyone the same. We talked about the hundreds of golf balls that would find their way into his property every day, food, raising kids and how incredible the weather was in Simi Valley. He will be greatly missed.”
On a personal note, I’ve known Bruce Clay since I entered the SEO industry more than 20 years ago. He was a role model to me, often a mentor, and always someone who was approachable, professional and deeply caring. In many rooms, he was likely the most generous and thoughtful person there.
I loved his SEO talks. I loved meeting him at industry events. And I especially valued the personal emails he sent about shaping the future of our industry. Those moments showed me how much Bruce cared, not just about search, but about the people building it.
With Profound’s Agent Template Marketplace, I can start from pre-built AI agent workflows instead of building every process from scratch.
It gives me ready-to-clone templates designed for marketing, SEO, and AEO teams, so I can move from idea to live workflow in minutes.
For me, the biggest advantage is speed: I can choose a proven workflow, clone it, customize it for my team, and start using AI agents faster with less setup.
I structure content for AI search by making every page clear, credible, and easy for answer engines to understand. That means I do not rely on keywords alone. I combine strong SEO fundamentals with topical authority, earned media, and answer-first formatting so AI systems can recognize what my brand knows, where it is trusted, and why it should be surfaced in relevant responses.
When I think about AI visibility, I focus on discovery from the start. I want my content to answer real questions directly, connect related topics naturally, and support each claim with signals that build confidence. This approach helps improve how my brand appears across AI search experiences, traditional search results, and emerging discovery platforms.
For me, the goal is simple: create content that is useful for people and understandable for machines. By organizing information around intent, authority, and clarity, I make it easier for AI tools to cite, summarize, and recommend my brand when users are looking for trusted answers.
The best and worst part of the web, in my view, is that I can share an opinion freely even when that opinion is not technically accurate.
But I keep wondering what happens when that freedom comes with real accountability, not only for what I say online, but also for whether the words came from me or from AI.
A recent report makes that question feel a lot less theoretical. A German court held Google accountable for AI Overview content, treating those AI-generated summaries as Google’s own content and rejecting the idea that users alone were responsible for fact-checking the results.
I want to unpack what that could mean for businesses, SEOs, and individuals who are leaning harder on AI every day.
The ‘disclaimer’ defense is cracking
For the last few years, I have seen nearly every AI platform rely on some version of the same warning: AI can make mistakes, so users should verify important information.
Most of us accepted that as the price of using these tools.
But the German court essentially said that a warning about possible errors does not automatically erase responsibility when those errors cause harm. If a system creates new claims that were never in the source material, those claims are no longer just someone else’s words. They become the platform’s words.
Why? Because the conversation moves away from whether AI is useful and toward who owns the consequences when AI gets something wrong.
What this means for businesses
I see many companies rapidly adopting AI across content creation, customer service, product descriptions, reporting, legal reviews, hiring, and internal communications. In many cases, they are blindly trusting the output because the efficiency gains are so tempting.
Most of the conversation still centers on speed and cost. Can we create content faster? Can we answer support tickets more cheaply? Can we automate this process?
Those are fair questions. I ask them too.
But this ruling adds a more important question: Who is responsible when the output is wrong?
What happens if an AI-generated support response gives a customer inaccurate guidance? What happens if an AI-written article damages a competitor’s reputation? What happens if an AI-generated report includes fabricated information that influences a business decision?
The more we position AI as a trusted source of information, the harder it becomes to argue that we should not be accountable for what it says.
The situation is kinda funny…
The irony is that most AI vendors already know this.
That is why nearly every platform includes warnings, disclaimers, and usage policies.
At the same time, those same companies market AI as smarter, faster, more capable, and increasingly reliable.
I do not think you can tell users to trust the answer while also arguing that nobody should trust the answer.
At some point, those positions collide. We are already starting to see Google’s solution: an option to opt out of AI.
Germany may simply be one of the first courts willing to force Google, or any other LLM business, to take clearer responsibility for the systems it puts in front of users.
What SEOs should be paying attention to
Ironically, I think this ruling could end up benefiting everyone.
Right now, the debate is focused on whether AI companies should be responsible for the content their systems generate. But I can see accountability expanding well beyond AI.
The internet has spent decades creating distance between actions and consequences. Anonymous accounts, fake profiles, throwaway emails, and now AI-generated content all make it easier for people to say things without owning them.
That is why I find this ruling so interesting.
It is not just about Google. It is about the idea that “I did not write it” may no longer be enough.
The image below shows a real email that Russell and Nina Westbrook received. A real person sat behind a keyboard and sent a message hoping they would die in a car crash.
That is not free speech. It is hate speech.
The internet, especially now that AI is layered into it, needs more confidence that content is accurate and that the people and companies creating it can be held accountable.
I do not believe we get to claim the productivity gains when AI is right and then blame the algorithm when it is wrong.
This post first appeared on the author’s website and is republished here with permission.
What Google said. Google wrote, “Released the June 2026 spam update, which applies globally and to all languages. The rollout may take a few days to complete.”
Timing. I expect this update to move fairly quickly, since Google said the rollout may take only a few days to finish.
Why I care. Google releases search ranking updates several times each year, and spam updates are meant to target sites that use manipulative tactics to abuse the ranking system. If a site is not relying on those kinds of practices, I would not expect it to be the main target of this update.
More on spam updates. Google’s documentation explains that its automated systems are always working to detect search spam, but the company occasionally makes notable improvements to those systems and labels them as spam updates.
Google also points to SpamBrain, its AI-based spam-prevention system, as one example of how it improves its ability to identify spam and catch new types of abuse.
If I saw a ranking change after a spam update, my first step would be to review Google’s spam policies and make sure the site is complying with them. Sites that violate those policies may rank lower or disappear from results, while improvements can help over time if Google’s automated systems recognize that the site is now compliant.
For link spam updates specifically, Google says recovery can work differently. If Google removes the value of spammy links, any ranking benefit those links once created is lost, and that benefit cannot be regained simply by cleaning up the links later.
I see Google Discover’s “Tailor Your Feed,” now showing up as “Add topics to your feed,” as a meaningful shift in how people can shape what appears in their feed. Instead of relying only on Google’s inferred signals, such as clicks, dwell time, follows, and engagement history, I can now type what I want to see in natural language and let Google translate that request into feed instructions.
That matters because it creates a third visibility path for small and niche publishers. Until now, a smaller site usually needed either strong implicit affinity from a user or an explicit follow. With prompt-based tuning, a user can simply ask for a topic, creator, source, or type of content, and Google can retrieve matching material even when that content has barely appeared in Discover before.
In my tracking, the feature turns prompts into actions such as SEE_MORE and SEE_LESS. Those actions are applied after the user refreshes or updates the feed. The experience feels conversational, but underneath it appears to create persistent instructions that can affect both the current feed and future Discover sessions.
I also see signs of an LLM-style system behind the workflow. A user prompt is interpreted, converted into a readable assistant response, and returned with a structured result. In one observed example, the prompt “show me more content on seroundtable.com” produced an actionable SEE_MORE response and a persistent thread key, suggesting that feed tuning is treated as an ongoing conversation rather than a single isolated command.
The feature first appeared in Search Labs for US English accounts in December 2025. At that stage, the impact was subtle: after several refreshes, I could see a few on-topic cards, but the feed did not radically transform. By early 2026, Google started adding attribution, including labels such as “resulting from natural language tuning” and later “You asked to see,” making it easier to identify which cards were influenced by a prompt.
By spring 2026, “Tailor Your Feed” had effectively become “Add topics to your feed.” The interface moved toward a chat-style entry point with prompt starters such as “Show me content from…,” “I want videos about…,” and “Keep me updated…”. The same underlying verbs remained, but Google made them easier for everyday users to trigger.
The most important technical clue is the pipeline behind the feature. Discover cards influenced by these prompts can be associated with naturallanguagetuningcontent.f for current tuning and historicalnaturallanguagetuningcontent.f for older prompts that continue shaping the feed. I read that “historical” pipeline as evidence that these preferences are meant to last over time, not disappear after one refresh.
From the observed cards, I see two ways this content is selected. The first and dominant mode is entity or interest expansion. A prompt is mapped to related people, topics, publishers, or concepts, and Discover expands around that meaning. This is why asking for one source or creator may also surface related sources, related subjects, or nearby entities rather than only the exact name typed into the prompt box.
The second and more interesting mode is query-intent fan-out. In this mode, a prompt is decomposed into natural-language retrieval queries. A broad request about SEO, for example, can become query intents such as “SEO strategies algorithm changes,” “Google ranking system updates,” or “tips for getting content into google discover.” Those query intents then retrieve articles based on semantic relevance.
This is where the connection to Generative Engine Optimization becomes clear to me. The Discover fan-out behaves like the retrieval pattern we see in generative search: one user prompt becomes several more specific sub-queries, and content is selected because it answers one of those sub-queries well. Popularity can still matter in some cases, but it is not the only gatekeeper.
That distinction is what gives niche publishers a real opening. In the observed data, prompts surfaced examples such as vegan recipe creators, Mississippi Today, a LinkedIn post, niche Japanese-property blogs, and a gardening site tied to a seed-starting query. Some mainstream publishers still appeared, including Reuters and VentureBeat in certain contexts, but the pattern was not limited to the usual high-volume Discover winners.
In the most striking cases, the pipeline surfaced articles with no detectable prior Discover distribution in the tracking dataset. I am not using “distribution” here as an audience number or a Search Console metric. I mean that the article did not appear to have circulated previously in the Discover tracking data available for analysis.
That makes this pipeline different from classic Discover distribution. Traditional Discover systems often re-serve articles that already have engagement momentum. Prompt-based tuning can retrieve content because it matches what a user explicitly asked for, even if the article has not already built a Discover track record.
I would not treat this as a mass traffic channel yet. Google appears to promote these cards cautiously, and the pipeline does not seem to snowball the way broader Discover pipelines can. It serves the user who asked. It does not automatically broadcast the content to a much larger audience.
I would also be careful about false positives. In one Japanese-property cluster, relevant results such as guides to buying a home in Japan appeared alongside a video-game article about in-game home locations. That kind of loose match helps explain why Google may rank and distribute these cards conservatively.
For publishers, the practical implication is straightforward: I would optimize for both topical clarity and query-intent vocabulary. The entity-expansion mode rewards sites that are unmistakably about a topic users can name. The fan-out mode rewards titles, headings, and introductions that align with the natural-language questions and information needs Google derives from prompts.
That does not mean stuffing pages with raw keywords. The better move is to describe the content clearly in the language a real person would use when asking Discover for more of it. If a user might ask for “buying Japanese property guide,” “starting seeds indoors guide,” or “tips for getting content into google discover,” I want the page’s title, H1, and opening section to make that relevance obvious.
The strategic shift is that selection power moves closer to the user. In the classic feed, Google infers demand. In this model, the user declares it. Google then turns that declaration into entities, interests, and query intents that drive retrieval.
For small publishers, that is the opportunity. If the feature graduates from Search Labs and users adopt it at scale, a focused site with clear topical authority could appear because it directly satisfies declared demand, not because it already won the popularity contest inside Discover.
There are still real limits. The feature has been US English and Search Labs focused, with French feeds showing essentially no presence in the observed data. Adoption also appears early. A powerful prompt-based personalization system changes little if users do not actually use it.
What I am watching next is whether Google expands this beyond Search Labs, whether the current and historical tuning pipelines become more visible, and whether this behavior converges with broader generative retrieval systems. A nascent generativeretrieval.f pipeline has already appeared in tracking data, but that broader connection still needs confirmation.
My read is that Discover is moving from observed personalization toward declared personalization. Google still infers plenty, but users are beginning to write part of their own interest profile. If that model becomes mainstream, niche publishers with clear focus, strong entity signals, and natural-language relevance may gain a new route into Discover visibility.
Notes: In this analysis, a Discover pipeline means the selection circuit that chooses and serves cards. The .f suffix in identifiers such as historicalnaturallanguagetuningcontent.f is an observed internal marker attached to Discover card metadata. “Fan-out” refers to a mechanism where one prompt is broken into several retrieval sub-queries. “GEO” means Generative Engine Optimization, or the practice of optimizing content for visibility in generative search and answer systems. “AIO” refers to AI Overviews, and “AI Mode” refers to Google Search’s conversational interface.
Field tracking referenced here covers Google app Search Labs US English accounts from December 2025 through June 2026. Pipeline behavior is based on close observation of Discover feed cards and 1492.vision tracking data. The internal mechanisms described are my interpretation of observed data and public research, and approximate dates are treated as approximate.
