Lily Ray, the search consultant who founded Algorythmic, published an essay on her Substack on November 19, 2026, arguing that every shortcut the search industry has chased over two decades eventually stopped working, and that the one investment to survive each cycle is putting real, identifiable experts at the center of marketing.
The essay, titled Why Building Trust Through Real People Stands the Test of Time in SEO & AI Search, runs to roughly 6,000 words. Its subtitle sets out the thesis in two sentences: "For 20+ years, Google has warned against scaled, automated content, and AI slop is just its latest form. I believe focusing marketing efforts on real people with lived experience is the best antidote." What follows mixes Google documentation going back to 2003, Ray's own client history, third-party studies on AI search citations and a five-part description of how brands can put their experts in front of audiences. It is one practitioner's argument, and Ray says so throughout, framing most claims with "I believe" or "in my experience."
In Short
Search consultant Lily Ray published an essay on November 19, 2026, arguing that SEO shortcuts from bought links to AI-written listicles all expired, while putting real, named experts at the center of marketing kept working. She backs the case with Google guidance dating to 2003, her own client recoveries and outside data showing AI search leaning heavily on YouTube, Reddit and individual LinkedIn members. For brands, the essay frames a public track record held by real people as the one signal cheap AI content cannot copy, though much of its evidence comes from Ray's own work and selected studies.
A loop of loopholes
Ray opens with a blunt line. "When it comes to organic search marketing, a loophole is not a strategy," she writes. The essay then lists the fads that have come and gone: exact-match domains, article directories, parasite SEO, guest posting at scale, private blog networks, programmatic pages, and now "the AI-generated listicles, scaled AI content, and 'inauthentic mentions' of the GEO era." Most of them, according to Ray, "work for a while... until they don't."
The cost of that pattern, in her telling, is not only lost traffic. Sites that leaned on these approaches often end up worse off than if they had never tried, sometimes spending years cleaning up while visibility and leads stay suppressed. Ray counts herself among those with scars. Her first real SEO wins, she writes, came from buying links in 2010 and 2011, and Google's Penguin update erased them in a single morning in 2012. Six years later, the 2018 "Medic" update hit a client of hers that had been using AI and automation to scale medical content, which she describes as the hardest hit from an algorithm update she had experienced firsthand since Penguin.
She frames those scars as part of the job, describing the lesson as thinking one knows better than Google and "learning the hard way that you almost always don't."
PPC Land has covered Ray's version of this argument before. In a December 2025 interview, she said sites were losing rankings because of past over-optimization such as scaled content and aggressive link buying, and predicted a crackdown on AI spam in 2026. In May 2026, she warned that self-promotional "best of" listicles were being treated as spam.
Twenty-three years of the same warning
The core of the essay is historical. Ray argues that Google's guidance has been remarkably consistent "if you read between the lines (or read the guidelines at all)." Google's Webmaster Guidelines from 2003 already told site owners to "make pages for users, not for search engines," warned against "cookie cutter" approaches with "little or no original content," and said that site owners following the spirit of the guidelines would do better than "those who spend their time looking for loopholes they can exploit."
Automation, according to Ray, has been in Google's sights for well over a decade. In August 2010, John Mueller told a site owner that publishing raw Google Translate output "could be seen as creating auto-generated content, which would be against our Webmaster Guidelines," and recommended having a human correct it first. By 2012, Google had a dedicated help page on automatically generated content. Its examples included text "translated by an automated tool without human review or curation," Markov chains, automated synonymizing, scraped feeds and "combining content from different web pages without adding sufficient value."
Ray notes that the same three techniques - automated translation, synonymizing and stitching - remain on Google's list 14 years later, and that marketers are now using large language models to experiment with all three. She also says she still sees sites hurt by scaled AI translations that get little human review. That 2012 list, she writes, "reads a lot like the sales pitch" for AI content tools that have captured the attention of senior marketers, many of which work in ways that come very close to, if not directly violate, Google's anti-spam policies.
The thread continued after ChatGPT arrived. In February 2023, Google wrote that it has "many years of experience dealing with automation being used in an attempt to game search results." In March 2024, it introduced its scaled content abuse spam policy, which applies whether content is made "through automation, human efforts, or some combination of human and automated processes." The main difference now, according to Ray, is that generative AI sits at the top of the list of examples.
Her summary of two decades of documentation is short: "the common thread is fairly simple: content made for search engines at scale, with no human reviewing it, adding to it, or standing behind it, eventually becomes a liability."
Google guidance cited in the essay
| Date | Document | Wording or change cited by Ray |
|---|---|---|
| 2003 | Google Webmaster Guidelines | "make pages for users, not for search engines"; warns against "cookie cutter" approaches with "little or no original content" |
| August 2010 | John Mueller forum reply | Unedited Google Translate output "could be seen as creating auto-generated content, which would be against our Webmaster Guidelines" |
| 2012 | Help page on automatically generated content | Lists text "translated by an automated tool without human review or curation," Markov chains, synonymizing, scraped feeds, stitching |
| 2014 | Search Quality Rater Guidelines | Google starts using the acronym "E-A-T" |
| December 2022 | Search Quality Rater Guidelines | "Experience" added to the acronym; raters asked to weigh the "extent to which a human being actively worked to create satisfying content" |
| February 2023 | Google Search Central blog | "many years of experience dealing with automation being used in an attempt to game search results" |
| March 2024 | Scaled content abuse spam policy | Applies to content made "through automation, human efforts, or some combination of human and automated processes" |
| October 2026 | Helpful content documentation | "Fabricating creator profiles (such as by using AI-generated headshots, made-up names, or false credentials to make content appear as if it was written by human experts) is a form of deception." |
The October 2026 change is one PPC Land has already documented: Google added the passage on October 2, and PPC Land reported on October 7 that the documentation now treats fake author profiles as a sign of a low-quality page. Google also confirmed in May 2026 that its spam policies, including scaled content abuse and inauthentic mentions, apply to AI Overviews and AI Mode, the same week it published a guide to generative AI features in Search with a section debunking common optimization myths.
E-E-A-T as a way out
When Google started using the acronym E-A-T in 2014, Ray writes, it finally gave practitioners a framework for doing SEO work without getting caught in the loopholes. She concedes that E-E-A-T has since been misunderstood and misused, often treated as a checklist or a single ranking factor that works the same way for every site, which she says it does not. She wrote an article debunking those misconceptions in 2021.
She still calls it the best framework available. "There's a reason E-E-A-T is mentioned 120 times in the current version of Google's Search Quality Rater Guidelines," she writes, adding that Bing uses similar language about "quality and credibility" in its own guidelines. The work is harder than loophole-focused SEO, according to Ray, because it needs time and resources from the experts at a company, not just the SEO team operating behind the scenes.
Ray built her consultancy, Algorythmic, around this approach and says it is the core philosophy of the business. That is relevant context: the essay argues for a method Ray sells.
The Medic client
The essay explains how Ray arrived there. The 2018 Medic client had been scaling a fast-growing library of health articles with no visible credentials and "no sign of a real human behind any of it." When the update hit, Google's Danny Sullivan pointed site owners to the Search Quality Rater Guidelines. Ray spent months reading them, and even printed and laminated a copy for her desk.
Her diagnosis turned on how the guidelines treat YMYL topics such as health and finance, which face the strictest standards: "the site was heavily YMYL (Your Money or Your Life) and showed nowhere near enough E-A-T for the health topics it covered." The client disagreed at first and the two parted ways. After almost two years without recovering, the client came back. About six months after a renewed strategy began, the site relaunched, rebuilt around real doctors, medical reviewers and their original insights, and climbed back to its previous levels. "From the day the update hit, the full recovery process took more than 3 years," Ray writes.
Health publishers have kept feeling this pressure. Ray's own visibility chart, which PPC Land reported in January 2026, showed WebMD down roughly 43% from its mid-2024 peak after Google's December 2025 core update.
A 2019 study, with caveats
Ray's case also rests on a 2019 analysis. She examined 64 sites that had clearly won or lost across the 2018 and 2019 core updates, checking each for 30 on-page E-A-T signals and researching the credentials of the people writing their content. "The winning sites were 258% more likely than the losing sites to use real experts on their content," she writes, adding the caveat herself: the sample was small, and correlation is not causation.
Other milestones in her account include a January 2019 talk at SMX West, where she told an audience to "cultivate your authors' profiles, even if it means starting from scratch," and Google's August 2019 guidance on core updates, which linked to Ray's BrightonSEO deck on E-A-T as a recommended resource alongside work by other SEOs. In 2020 she studied Google patents with the late Bill Slawski to understand how Google's systems might recognize individual experts.
What Ray means by "real humans"
Ray is careful to define the term, since "humanize your brand" and "keep a human in the loop" have become marketing clichés. By real humans she means practitioners with genuine expertise or experience: the doctor who has treated the condition, the engineer who built the system, the plumber who owns a local business, or the patient who lived through the illness. Such people, in her description, earn audiences across platforms by sharing insights that are honest, original and often opinionated, and any sale tends to follow from the trust they have built.
Her reason for thinking this survives whatever comes next is compact: "Nearly every popular shortcut in the vicious cycle of SEO could be scaled. A real expert's reputation can't." As AI makes everything else cheaper to produce, Ray expects the value of earned reputation to keep climbing. She repeats a line from an April 2024 Authority Hacker podcast appearance: "I've been talking about this for years: People like following people."
The essay lists five things real people bring that AI cannot: expertise earned over years, feelings and lived experience, something new to the conversation, humor and personality, and flaws and vulnerability. Each comes with a supporting reference. Ray cites a study in which, across nine experiments with more than 6,000 people, the same supportive message was rated as more empathic and more genuine when people believed a human wrote it. She points to a 2026 study in PNAS Nexus that found leading AI models score about as well as people on creativity tests, but give answers far more similar to each other than people's are. She notes that Cambridge Dictionary named "parasocial" its 2025 Word of the Year, and invokes the pratfall effect, in which a highly competent person became more likable after spilling a cup of coffee.
On LinkedIn, she adds, personal profiles outperform company pages by roughly 2x to 8x on impressions, "depending on whose data you believe." That attention has limits as an indicator: an Originality.ai analysis PPC Land covered in January 2026 classed 53.7% of long-form LinkedIn posts from 99 influential profiles as likely AI-written.
The trail real people leave
Ray argues that signs of a real human may be the clearest way to separate spam from the content that offers the most value. "A real human expert leaves evidence everywhere, and that footprint is easily detectable," she writes: conference photos, video of talks, a personal website, podcast interviews, years of social posts, and journalists quoting them by name.
She believes Google has systems to identify and likely reward those signals, from Knowledge Panels to its newer Search profiles. She also points to the leaked Google API documentation of May 2024, which, according to Ray, showed that Google stores the author of a document and has an attribute called isAuthor marking whether an entity mentioned on a page is also its author.
The typical fake author, by contrast, has a stock-looking AI-generated headshot, a two-sentence bio and no other evidence of existing anywhere online. Ray cites Sports Illustrated's 2023 product reviews under authors who did not appear to exist, including one whose headshot was for sale on a site selling AI-generated portraits. Fake authors were also one of the most common patterns she found among sites hit by the 2023 Helpful Content Update.
She expects this trail to matter more as the web fills up with homogeneous vibe-coded sites, which anyone can now spin up in an afternoon complete with logos, a hundred blog posts and team pages full of AI-generated headshots. Many share "the same purple gradient" and the same "Trusted by" logo strip, according to Ray. "When a convincing-looking brand costs next to nothing and takes little time to build," she writes, "polish alone doesn't tell anyone very much anymore."
Search rebuilt around voices
Ray reads the past eight years of Google product changes as a steady build-out of systems that reward that kind of trail. Starting in 2021, Product Reviews updates rewarded evidence that reviewers had actually used products, such as original photos, videos or measurements. Garage Gym Reviews, whose equipment is tested by certified trainers, coaches and athletes, and Travel + Leisure, which runs a testing lab in New York, are her examples.
In May 2023, Google said it wanted to surface more "hidden gems," such as forum comments and posts on small blogs, and later confirmed the change had been part of core ranking for months. By November 2023, Ray was posting on LinkedIn that Google seemed to be "way overindexing on 'authentic content' and 'real experience'." Social media, user-generated content and forums were the biggest winning category in her 2023 winners and losers analysis. An Amsive analysis by Silvia Gituto, using Sistrix data, found Reddit moved from the 68th to the 5th most visible domain in Google's US results within a single year. At one point, a five-year-old Reddit post from a deleted account outranked the Mayo Clinic for "how to potty train a boy." Google has since denied favoring the site: in August 2026, a Google spokesperson said Reddit "gets no special preference".
The most recent phase, in Ray's account, is a focus on individual creators. Google announced on September 17, 2025, that Discover would let users follow publishers and creators and would add social posts. On June 4, 2026, Search profiles launched in the US, giving creators a claimable page whose followers are more likely to see their content in Discover. The thresholds Google published are 100,000 subscribers or followers on YouTube, Instagram or X, or 300,000 on TikTok.
In October 2026, Damien Andell of 1492.vision flagged a Google job posting about turning Discover into a "proactive, hyper-personalized" experience, which Ray says explicitly calls for keeping human voices and sources at the center. PPC Land reported on October 12 that the role, a principal UX position for Search Discover, carries a US base salary of $275,000 to $382,000.
Where AI search finds its sources
Ray's argument extends to AI answers. "Real experts and real lived experiences appear to drive a big share of AI search citations," she writes, and the evidence she cites is which sites get cited most.
| Metric | Figure | Source as cited by Ray |
|---|---|---|
| YouTube share of AI Overview citations among the top 50 cited domains | 24.5% | Ahrefs, November 2026 |
| Reddit share of AI Overview citations among the top 50 cited domains | 18.1% | Ahrefs, November 2026 |
| LinkedIn rank among cited domains in ChatGPT Search, Google AI Mode and Perplexity | 2nd | Semrush, 325,000 prompts |
| Share of cited LinkedIn content in ChatGPT Search and AI Mode from individual members | 59% | Semrush |
| LinkedIn personal profile impressions versus company pages | 2x to 8x | Various studies |
| Reddit visibility rank in Google US results | 68th to 5th in one year | Sistrix, via Amsive |
| US adults who trust AI-generated content | 35% | Emplifi, 2026 |
| Time for a LinkedIn Pulse article to top an AI Overview and earn an AI Mode citation | 7 hours | Ray's own test, March 25, 2025 |
The Semrush study is one PPC Land has reported on twice. Its March 2026 release found LinkedIn in 11% of AI responses on average, drawn from 325,000 prompts in January and February 2026. A June follow-up showed the split by platform: ChatGPT Search and Google AI Mode cited individual creators 59% of the time, while Perplexity cited Company Pages 59% of the time. Ray's figure covers the first two platforms only.
Her own test is the most personal data point. On March 25, 2025, Ray posted on X: "SEO test confirmed. I wrote this 7 hours ago. LinkedIn Pulse for the win. (Also, this is so dumb, I can't believe it actually works)." A second post that evening showed the same article cited in AI Mode.
The point she draws is that much of what AI systems use comes from places where people share what they know: a YouTube creator walking through a fix, a Reddit thread full of firsthand experience, a practitioner's LinkedIn post. Those same platforms, she acknowledges, are also being flooded with AI slop.
YouTube and Reddit push back
Both platforms have said what they want. In July 2025, YouTube renamed its "repetitious content" monetization rule to "inauthentic content," covering AI-generated videos made from templates "without adding the creator's original, authentic insights or perspective." PPC Land reported the change on July 15, 2025. Ray adds that YouTube CEO Neal Mohan's 2026 letter called out "AI slop" by name.
Reddit CEO Steve Huffman wrote in February 2026 that "if you can't easily distinguish a real person's thoughts or recommendations from a bot, that trust erodes." Ray says Reddit started labeling automated accounts a month later. PPC Land's March 30 report described Huffman's March 25 post introducing an [App] label for permitted bots, alongside the removal of about 100,000 bot accounts a day, though it gave no timeline for a broad rollout.
What one person did for a company
The essay's most specific numbers come from Ray's own career. When she joined the agency Amsive, she had never given an SEO talk or done anything public-facing. In almost nine years on its SEO team since, she has published research under her own name, alongside podcasts, conference talks and webinars. A recent look at the agency's backlinks in Ahrefs produced the figures below, all as reported by Ray and not independently checked.
| Organization | Figure | Detail as described in the essay |
|---|---|---|
| Amsive | About 7,800 linking sites | Roughly 1 in 13 mention Ray by name in or next to the link |
| Amsive | 9 of 10 most-linked articles | Original data studies from the SEO team, 7 written or co-written by Ray |
| Amsive | 120 linking sites | Ray's author page, linked from more sites than nearly all of the company's articles |
| Wix | 2022 | Crystal Carter joins; Ray says Wix also hired Mordy Oberstein to represent the brand within the SEO community |
| Rick Steves | 1980 | First self-published travel guide |
| Rick Steves | More than 100 employees | Company built on one person's travel experience |
| Rick Steves | Since 2000 | PBS series on air |
| Rick Steves | More than 30,000 travelers a year | Tour program to Europe |
The links all point to the company, Ray stresses. Wix is her industry example: a platform the Ahrefs blog once described as having something of a bad reputation for SEO, which hired respected SEOs to represent it. Ray says SEO publications now often point to Carter by name when assessing Wix, and that the same Ahrefs article credits Carter and Oberstein with pushing for "more SEO features and better communication." Outside search, she points to travel writer Rick Steves, who has spent about four months a year in Europe for decades and whose content is free from affiliate influence because the business earns money from books and tours. "His honesty is the product," she writes, noting that Google's AI Overviews often cite his website, YouTube and TikTok pages.
The five parts Ray describes
The essay closes its practical section with a description of how a company can build what Ray calls a human-first strategy with brand evangelists. According to Ray, it has five parts.
- Identify the experts. The starting point is the people who actually know things: a product lead, the head of support, the engineer who answers hard questions in Slack, or a founder who still takes customer calls.
- Build their presence on the company site. Ray describes robust About pages and author and expert pages that work as a personal portfolio, connect to the expert's social profiles and give them a voice in company communications, from the blog to LinkedIn.
- Build a footprint outside the site. This covers conference talks, podcast interviews, Reddit posts or management of an official branded subreddit, LinkedIn and other communities where buyers spend time, plus earning and claiming a Knowledge Panel and, where eligible, a Search profile.
- Drop the sales pitch. "Keep the pitch out of it," Ray writes, arguing that "effective marketing is increasingly becoming authentic, not salesy." She lists sharing data, mistakes and "I don't know yet," and crediting other people's work.
- Make people easy to verify. Ray cites an enterprise client whose About page, a detailed company history, was its most cited page in AI Overviews and AI Mode. In an October 22, 2026 post, she wrote: "The most cited page on the company's website in AI Overviews & AI Mode, with 5x more impressions than the homepage, is its 'About' page."
That About page figure came from the generative AI report in Google Search Console, which Google introduced for a subset of sites on June 3, 2026 and extended to all sites on August 31. The report counts impressions, not clicks or queries.
Ray offers two counterpoints of her own. Some industries have compliance teams that make personal posting genuinely hard, and people may leave a company and take their personal brands with them. Her response is to build several voices rather than rely on one, citing Viola Eva of Flow Agency, and to note that links and coverage earned while someone represented a company remain assets after they leave.
Why Ray thinks it lasts
The final argument is about audiences. In July 2023, Ray posted on LinkedIn that even if Google could not tell AI content from human content, users could. She says consumer research has since caught up: a Gartner survey found that a majority of US consumers say AI content has made them less trusting of brand messaging, and Emplifi's 2026 research found that only 35% of Americans trust AI-generated content at all. Ray calls this slop fatigue. PPC Land has reported similar findings, including a Raptive study in July 2025 in which suspected AI content cut reader trust by nearly 50%.
Every tactic that blew up in the industry, from links bought in 2011 to listicles and paid mentions in 2026, was scaled for the sole purpose of driving search visibility, according to Ray. She says that list is almost exactly what consumers say they dislike most about SEO.
The second reason is reputation. Ray quotes a rater guideline passage she put on a slide in 2019: "When there is disagreement between what the website says about itself and what reputable independent sources say about the website, we'll trust the independent sources." AI systems, in her view, are getting better at judging reputation from what others say, including spotting self-promotion. She cites an example, credited to Gagan Ghotra's post on November 19, of Gemini's visible reasoning steps assessing whether "Best AI SEO agency" lists were self-promotional, with agencies ranking themselves highly. "Being genuinely recommended by other people is the most sustainable way forward," she writes.
That claim connects to Ray's earlier research. In October, PPC Land reported her finding that AI Overviews cited 38% fewer vendor-written "best of" pages than in June, and that brands ranking themselves first were left out of 83% of answers that cited their pages. In June, she described an AI slop loop in which AI systems recycled fabricated SEO claims into confident answers. And in September, Google's AI Overview affirmed Ray's peer-voted SEO World Champion title, with her own site rather than the host's posts shown as the visible source.
The essay ends on the line it started with. "People like following people. That was true before Google," Ray writes, adding that "the brands that figure that out won't need the next loophole."
Why this matters for marketers
The essay was published nine days into Google's November 2026 core update, which started on November 10 and may take up to two weeks. Ray does not mention the update.
For marketers, the more durable shift is that the signals Ray describes now sit in Google's own documentation and products rather than only in practitioner theory. The fabricated-creator passage in the helpful content guidance, Search profiles with follower thresholds, Discover follows and the Search Console generative AI report all give companies concrete surfaces where named people show up, or fail to. For work on AI search visibility, the citation data Ray assembles points at platforms where individuals, not brands, do most of the talking. The figures also show limits: Search profile thresholds of 100,000 followers or more put that product out of reach for experts without large social audiences, and Perplexity's preference for LinkedIn company pages cuts against the essay's main claim.
The essay's evidence is uneven. Much of it comes from Ray's own clients, posts and agency backlinks, none of which can be independently checked; the 2019 study is small and correlational, as she notes; outside studies are chosen to support the case; and Ray sells consulting built on this approach. She also gives no cost or timeline for a company building expert voices, beyond the three-plus years one client needed to recover.
Key facts
- Title: "Why Building Trust Through Real People Stands the Test of Time in SEO & AI Search"
- Author: Lily Ray, founder of Algorythmic; almost nine years on Amsive's SEO team, according to the essay
- Publication: Lily Ray's Substack, November 19, 2026
- URL: https://lilyraynyc.substack.com/p/why-building-trust-through-real-people
- Length: roughly 6,000 words
- Google documents quoted: Webmaster Guidelines (2003); help page on automatically generated content (2012); Search Central blog on AI content (February 2023); scaled content abuse spam policy (March 2024); helpful content documentation, fabricated creator profiles passage (October 2, 2026)
- Rater guidelines: E-E-A-T mentioned 120 times in the current version, according to Ray's count; Experience added December 2022
- Ray's 2019 study: 64 sites, 30 on-page E-A-T signals, winners 258% more likely to use real experts
- Medic client recovery: more than three years from the August 2018 update to full recovery
- AI citation figures cited: YouTube 24.5% and Reddit 18.1% of AI Overview citations among the top 50 domains (Ahrefs, November 2026); LinkedIn second most cited domain, 59% from individual members in ChatGPT Search and AI Mode (Semrush, 325,000 prompts)
- Platform rules cited: YouTube "inauthentic content" rule (July 2025); Reddit bot labeling (announced March 25, 2026)
Timeline
- 1980: Rick Steves self-publishes his first travel guide
- 2000: Rick Steves' PBS series begins
- 2003: Google's Webmaster Guidelines tell site owners to "make pages for users, not for search engines"
- August 2010: John Mueller says unedited Google Translate output could count as auto-generated content
- 2010-2011: Ray's first SEO wins come from bought links
- 2012: Penguin update erases those wins; Google publishes a help page on automatically generated content
- 2014: Google starts using the E-A-T acronym
- August 2018: Medic update hits Ray's health client; recovery takes more than three years
- January 2019: Ray's first conference talk, on E-A-T, at SMX West
- 2019: Google's core update guidance links to Ray's BrightonSEO deck; Ray analyzes 64 sites
- 2021: Product Reviews updates begin rewarding evidence of hands-on testing
- 2022: Crystal Carter joins Wix
- December 2022: Google adds Experience to E-A-T
- February 2023: Google says it has years of experience with automation used to game results
- May 2023: Google says it wants to surface more "hidden gems"
- 2023: Sports Illustrated found publishing reviews under authors who did not appear to exist
- March 2024: Google introduces the scaled content abuse spam policy
- April 2024: Ray tells the Authority Hacker podcast "People like following people"
- May 28, 2024: Leaked Google API documentation draws scrutiny
- March 25, 2025: Ray's LinkedIn Pulse article tops an AI Overview within seven hours
- July 15, 2025: YouTube renames "repetitious content" to "inauthentic content"
- July 2025: Raptive study finds suspected AI content cuts reader trust by nearly 50%
- September 17, 2025: Google adds creator follows and social posts to Discover
- December 2025: Ray says sites lose rankings over past over-optimization
- January 2026: WebMD down roughly 43% from its peak after the December 2025 core update
- February 2026: Steve Huffman warns bots erode trust on Reddit
- March 2026: Semrush finds LinkedIn second among domains cited by AI search tools
- March 25, 2026: Huffman announces an [App] label for permitted bots
- May 14, 2026: Ray warns self-promotional listicles are treated as spam
- May 15, 2026: Google says its spam policies cover AI Overviews and AI Mode
- June 3, 2026: Search Console gains generative AI reports for a subset of sites
- June 4, 2026: Search profiles launch in the US
- June 2026: Semrush shows ChatGPT Search and AI Mode cite individual LinkedIn creators 59% of the time
- June 21, 2026: PPC Land covers Ray's AI slop loop findings
- August 31, 2026: Search Console generative AI report reaches all sites
- September 2026: Google's AI Overview affirms Ray's SEO World Champion title
- October 2, 2026: Google adds fabricated creator profiles to its helpful content documentation
- October 4, 2026: Ray finds AI Overviews cite 38% fewer self-ranking listicles
- October 11, 2026: Damien Andell flags Google's Discover concierge job posting
- October 22, 2026: Ray posts that a client's About page drew 5x the homepage's AI impressions
- November 10, 2026: Google starts its November 2026 core update
- November 2026: Ahrefs puts YouTube at 24.5% and Reddit at 18.1% of top-50 AI Overview citations
- November 19, 2026: Ray publishes the essay
Related PPC Land coverage
- Google says fake author profiles signal a low-quality page - The October 2026 helpful content documentation change Ray quotes.
- Google AI Overviews cut self-ranking listicle citations 38%, Lily Ray finds - Ray's September measurements of vendor-written "best of" pages in AI Overviews.
- The AI slop loop: how fake SEO advice is gaming search results - Ray's analysis of AI systems repeating fabricated SEO claims.
- Lily Ray: what the SEO industry is getting dangerously wrong about AI search - Ray's May 2026 podcast remarks on listicles and scaled comparison pages.
- SEO expert says websites lose rankings because they're doing too much SEO - Ray's December 2025 interview on over-optimization and long recoveries.
- Google's AI answer on Lily Ray's SEO world title cites one source: her own - How an AI Overview sourced Ray's SEOktoberfest title.
- Health sites that once dominated SEO just got crushed by Google - WebMD and Healthline visibility after the December 2025 core update.
- Google Search profiles give publishers a dedicated space in Discover - The June 2026 US launch of claimable creator profiles.
- Google Search profiles go live with a follower wall - but some already got in - The follower thresholds for Search profiles.
- Google Discover adds social posts and creator follows - The September 2025 Discover redesign Ray cites.
- Google offers up to $382,000 for a lead to make Discover an AI concierge - The Discover job posting flagged by Damien Andell.
- LinkedIn ranks #2 in AI citations - what 89K URLs reveal about B2B visibility - The Semrush study of 325,000 prompts.
- Semrush maps how LinkedIn content earns citations in AI search tools - The platform-by-platform split between individual and company content.
- YouTube clarifies "inauthentic content" policy changes - The July 2025 monetization rule rename.
- Reddit labels good bots and removes 100,000 bad ones daily - Huffman's March 2026 bot labeling announcement.
- Google spam policies now officially cover AI Overviews and AI Mode in Search - The May 2026 clarification covering scaled content and inauthentic mentions.
- Google finally gives Search Console its own generative AI visibility reports - The report behind Ray's About page observation.
- Raptive study shows AI content cuts reader trust by half - Survey of 3,000 US adults on suspected AI content.
- Google's core update shifts rankings for up to 2 weeks before Black Friday - The November 2026 core update underway when the essay was published.
Summary
Who: Lily Ray, founder of the consultancy Algorythmic, who has spent almost nine years on the SEO team at the agency Amsive. The essay discusses Google, Bing, YouTube, Reddit, LinkedIn, Wix and travel writer Rick Steves.
What: An essay arguing that SEO shortcuts, from bought links to AI-written listicles and scaled AI content, eventually stop working, and that real, identifiable experts are the approach that has held up. It cites Google guidance from 2003 to October 2026, a 2019 study of 64 sites, a client recovery that took more than three years, and data showing YouTube, Reddit and individual LinkedIn members among the most cited sources in AI search.
When: Published November 19, 2026, nine days into Google's November 2026 core update.
Where: On Lily Ray's Substack, with examples drawn from Google Search, Google Discover, AI Overviews, AI Mode, ChatGPT Search, Perplexity, YouTube, Reddit and LinkedIn.
Why: Ray argues that as AI makes generic content nearly free to produce, a reputation built by real people over years becomes the hardest signal to fake, and that Google's documentation and products have been moving toward rewarding it. The evidence is largely Ray's own, and she sells consulting built on the approach.
Discussion