Only Google's Own AI Still Respects Page One

AD

Get more meetings by removing the LinkedIn limits.

When you need more leads, you want to do more, right?

• Adding mailboxes to cold emails.

• Placing more cold calls

• Running more ads.

But what do you do if your leads are on LinkedIn? LinkedIn outreach still depends on the limited number of profiles already on your team.

Aimfox Avatars lets you rent dedicated, customizable LinkedIn profiles that start conversations and hand interested replies to your existing sellers.

Add outreach capacity without hiring another SDR or turning every employee account into a prospecting channel..

Book more calls with Aimfox Avatars

SEO + GEO

The Ghost Citation: When AI Knows Your Brand Better Than It Knows Your Content

The Foundation team ran 20 buyer questions across six AI surfaces for n8n and found the brand appeared in 91 of 96 unbranded answers while owning just 15.7 percent of citation share. Reddit threads earned 95 citations against 33 for the n8n blog, and one enterprise thread with 177 upvotes kept resurfacing across prompts, platforms and both test rounds. ChatGPT leaned on n8n docs for technical questions while Google AI Mode and Gemini leaned on outside sources, and no single n8n comparison page became the consistent reference for n8n versus Zapier queries.

How Much Does AEO Cost? Pricing by Agency, Tools and Software

The HubSpot team breaks down what answer engine optimization actually costs, from self-serve monitoring near $30 a month to full-service agency programs above $15,000 a month.

• HubSpot AEO costs $50 a month and covers three engines, or comes included with Marketing Hub Professional and Enterprise.

• Profound starts at $99 a month for ChatGPT-only tracking, with multi-engine coverage from $399 a month on Growth.

• Peec AI starts from $95 a month for self-serve visibility tracking.

• Otterly.ai self-serve tiers run from about $29 a month on Lite to roughly $489 a month on Premium.

• Ahrefs Brand Radar costs $199 a month per answer engine.

• Entry agency plans start near $249 a month at NetWebMedia, while Nugentive's done-for-you plan runs $1,497 a month for schema and three to five pages.

• Ongoing agency programs typically run $9,000 to $15,000 a month, with RevenueZen's top package at $15,000.

• Single sprints start near $3,000 at Digital Elevator, with top engagements above $24,000.

• AI referral traffic roughly tripled from January to December 2025 yet still sits below 2 percent of referral traffic in the cited dataset.

My Take: If you’re in the AEO/GEO space with a tool or agency, this is interesting.

Google Ads Data Shows Query Length Shift Post-AI Mode

Jason Tabeling reports that longer queries have taken over Google Ads impressions and conversions through August 2026. One and two word queries fell from 42% to 24% of impressions while three and four word queries rose from 33% to 48%. Conversion share moved the same way, with one and two word queries dropping from 62% to 52% and three and four word queries jumping from 20% to 46%. Five and six word queries tripled from 3% to 9% of conversions and queries of seven or more words quadrupled from 1% to 4%. He recommends shifting budget toward long tail coverage, matching ad copy to the specific language searchers use and auditing performance by query length.

What Wikipedia Reveals About AI Overviews and Web Traffic

Matt G. Southern breaks down a University of Washington working paper that estimates default AI Overviews cut monthly search referrals to English Wikipedia by about 5%. The authors compared 499,927 English-German and 530,873 English-French article pairs around the May 2024 US rollout and calculate roughly 100.27 million fewer search referrals per month, or about 1.20 billion per year. Google disputes the metric because Wikimedia data groups all external search engines together. Matt notes Wikimedia separately reported an 8% drop in overall human pageviews and says nearly 90% of Wikipedia visitors have traditionally come from Google search.

When SEO Best Practices Become SEO Dogma

Stephanie Wallace explains how familiar SEO checklist items turn into dogma once the original reasoning gets lost. She says a slow page deserves engineering time when it hurts behavior on a high value template, not when a low traffic resource misses a benchmark by a fraction of a second. She walks through eight questions for prioritizing work, including what evidence calls for change, what problem a fix solves, how much of the site it touches and what happens if nothing changes. She warns against competitive cloning where every result starts to look the same and says visibility now depends on the broader footprint across publications, communities, video and AI answers.

Blogging Statistics 2026: What 1,042 Content Marketers Told Us About What Works

Andy Crestodina breaks down Orbit Media's 2026 survey of 1,042 content marketers and finds strong results at an all-time low.

• 13.9% report strong results, an all-time low and six points below the prior 12-year low.

• 18% say they do not know if their blog delivers results, up from a 9% to 14% range over the past decade.

• Average post length is 1,312 words, down from a peak of 1,427 words in 2023.

• Average time to write a post is 3 hours and 20 minutes, down from a peak of 4 hours and 10 minutes in 2022.

• 92.4% now use AI for blogging, yet AI users and non-users report strong results at the same 14% rate.

• Influencer collaboration beats the benchmark by 2.6 times yet only 7% do it, down from 25% in 2017.

• Publishing original research lifts the odds of strong results by 50%, yet fewer marketers do it than in prior years.

• Marketers using six or more of the eight effective strategies report strong results at 39% versus the 14% benchmark.

Google Ranking Factors Expert Survey 2026

Cyrus Shepard and Dawn Shepard break down ratings from 131 SEOs across more than 100 potential ranking factors, totaling 13,665 data points.

• 57.1% named relevance as a top three factor, with search intent match rated the single strongest signal.

• 54.8% named backlinks, with links from trusted and topically relevant pages rated highest.

• 47.6% named content quality, led by original research and first-party data.

• 36.5% named authority and trust in the site, creator or brand.

• 29.4% named behavior and click signals, with satisfaction and task completion rated highest.

• 27.0% named brand signals such as branded search volume and online reputation.

• 19.8% named user satisfaction as the ultimate outcome.

• 17.5% named technical SEO health around crawling and indexing as the foundation.

• 14.3% named topical authority and 11.1% named internal links.

ChatGPT, Gemini and Claude Lead AI Visibility, Is It Time To Stop Tracking Perplexity?

Greg Jarboe pushes back on calls to drop Perplexity from AI visibility trackers and makes the case for weighting instead of removing. He cites StatCounter showing Perplexity falling from 7.91% to 4.31% of AI chatbot referral share from June to August 2026 while Gemini rose from 7.94% to 10.9%. He cites Similarweb May 2026 shares of 53.9% for ChatGPT, 27.9% for Gemini, 9.2% for Claude and 1.3% each for Perplexity and Copilot. He notes AI Overviews reach more than 2.5 billion users per month and appeared in 43% of US Google searches by May 2026, up from 15% a year earlier. He proposes three tiers with ChatGPT, Gemini and Claude tracked separately, Google AI Overviews and AI Mode as its own search discovery layer and Perplexity monitored with lower weight.

My Take: In Floyi, we track AI overviews, AI Mode, ChatGPT, and Gemini. I never gave serious consideration to Perplexity because it’s never held a big enough share among most consumers. The people who talk about Perplexity the most are the ones reading this right now. But our AIRS Analyzer tool does fetch AI responses from 9 different AI engines for one-off checks. That’s still worth it. Just not for tracking regularly.

Almost 100% of People Also Ask Results Are Answered by Google AI Overviews

Barry Schwartz reports that nearly all Google People Also Ask answers are now AI Overviews, up from about 12% fourteen months ago. Allintitle's Also Ask Miner data puts the share at 100% starting in August 2026. A separate AlsoAsked sample of 19.2M English queries found 97% of PAA answers were AI Overviews in the first week of September, up from 86% in August. Mark Williams-Cook says the trend points to 100% shortly.

Google Says Old Low-Value Pages May Affect Site Recovery

Matt G. Southern reports that John Mueller says old low-value programmatic pages can still color how Google judges a site after changes go live. The comment came in a Bluesky reply about a site spinning pages from combinations of domain names, technologies and attributes, where Mueller said systems have possibly lost faith in the site providing good value. He says fixing it takes time and significant effort to show value again, and he points to spam and core update recovery work as a rough comparison for effort and timing. He names no ranking system, threshold or timeline, and Matt notes the reply changes no policy since scaled content abuse already centers on purpose and value, not scale alone.

My Take: I treat cleanup and trust rebuilding as two separate jobs. Deleting thin pages stops the bleed but it doesn't rebuild confidence on its own. I would prune first, then ship a run of genuinely useful pages and give it months before judging.

Threads Search and Indexing: How Meta's Platform Feeds Google

Tamara Biljman shows how public Threads posts get indexed in Google and keep earning search visits long after they leave the feed. She reports Threads drove 85.2 percent of Sendible's organic social engagement in July 2026, ahead of Facebook at 6.5 percent and TikTok at 4.1 percent, and she ties indexing odds to reply depth in the same way Google favors engaged Reddit threads. She walks through site searches like site:threads.net with a brand name or handle plus URL Inspection checks, and she notes Threads still is not a supported Platform Property in Search Console.

Before Scaling Programmatic SEO, Prove the Pattern Works

Bharath Ravishankar explains how to test a programmatic page pattern on a small pilot before building thousands of URLs. He says to define what changes per URL, from workflows and setup steps to inventory and pricing, then sketch about 10 candidates and cut the project if most pages repeat the same copy with swapped fields. He recommends a representative pilot that mixes large, mid-sized and smaller markets plus strong and thin data cases, with a recorded primary query and intended action for each page. He lays out four checks before scaling: consistent discovery and indexing, queries matching intent, traits shared by winners and whether visits drive trials, purchases or inquiries. If only part of the pilot works, he says scale the proven slice, fix or hold the rest and expand in stages rather than launching the full library at once.

SEO + GEO Ripples

• Search Console indexing report lost a chunk of June data across properties, with charts showing a gap where June history should sit. John Mueller says it traces to a June delay with no backfill for indexing data. Do not rebuild based on the gap and check again after the team confirms.

• Google brought post view counts back to Business Profiles for posts from the past 18 months, combining Search and Maps views on each post card. The metric returns more than three years after the February 2023 retirement. It is dashboard only for now with no API support, so expect manual checks until documentation catches up.

• Google says old position 1 to 10 mapping no longer works for modern results with AI Overviews, featured snippets, People Also Ask and other blocks. John Mueller says Search Console tracks these placements as blocks and does not split them out in the Gen-AI performance report.

• Interactive character counter inside AI Overviews that answers character counter queries directly in the result, a change spotted by Damir B. Utility sites like wordcounter.net face the same zero-click squeeze seen with weather, dictionary and calculator answers. Wait until they add a Rewrite button

• Apple clarified Applebot-Extended rules do not affect Search with a new help doc line stating those site rules are not considered in ranking for Search. The split mirrors Googlebot-Extended and lands ahead of iOS 27 and the new Siri push, so AI training blocks stay separate from search crawling.

AI

Introducing Muse: The World’s First Personal AI Agent Built for Everyone

Meta introduced Muse, a personal AI agent that can send email, book travel, fill out forms, and negotiate on a user’s behalf. It runs in a dedicated cloud computer called Muse Secure VM, where Meta says credentials stay in secure storage and a separate Sentinel agent must approve internet actions. Users choose each connected app’s permissions, approve sensitive actions, and can revoke access or opt out of training their interactions.

Introducing ChatGPT Images 2.5

OpenAI released ChatGPT Images 2.5 with sharper detail, faster generation, and more precise edits across repeated prompts. The company says generation latency is up to 50% lower than Images 2.0, while the model better preserves subjects in reference photos and keeps earlier changes intact. ChatGPT also adds Sketch, templates, image comments, and prompt sharing. Developers get GPT-Image-2.5 Flare for most API work and GPT-Image-2.5 Sunburst for more detailed creative workflows.

My Take: I’ve already added GPT-Image-2.5 Flare and Sunburst to Floyi for articles. They’re faster and have better quality output too compared to GPT-Image 2.

Detecting and countering misuse of AI: September 2026

Anthropic reports disrupting malicious Claude use between December 2025 and August 2026 across seven harm areas, including cyber operations, influence campaigns, surveillance, fraud, biological misuse, weapons development, and illicit distillation. It says most cited cyber operations used AI to execute or orchestrate reconnaissance, exploitation, and data exfiltration through multi-agent systems, rather than simply assist humans in a chat. One influence network generated 8,913 articles in roughly 20 languages, though Anthropic found little evidence that the content reached audiences beyond its own sites and social accounts.

AI Chatbots Talked 57.5% of AI Users out of Buying

Carlos Silva reports on an Exploding Topics survey of more than 2,300 US consumers on how AI chatbots shape product research and purchase decisions.

• 57.5% of AI users have decided not to buy something based on chatbot information.

• 41% of all consumers have bought based on an AI chatbot recommendation, rising to 57.51 percent among AI users.

• 80.45% of people who use AI multiple times per week have bought an AI-recommended product.

• 59.27% of AI users have found a new brand or product through a chatbot.

• 70.62% of AI users believe chatbots can recommend the best brands and products.

• 65.01% of AI users have at least partly replaced product-related Google searches with AI chatbots.

• 19.43% of consumers would pick an AI chatbot if limited to one product research tool.

• 41.63% dislike ads in AI chatbots, and 66.05 percent of that group say ads make them doubt response integrity.

• 87.41% of people who use AI for company research would at least sometimes consult it before hiring a local business.

Google DeepMind Develops New AI Search Ranking Model

Roger Montti breaks down Google DeepMind research on Autoregressive Ranking, a single LLM system meant to replace two-stage retrieval. He explains current systems pair fast Dual Encoders for candidate pickup with heavier Cross Encoders for final ranking. The proposed method uses a training loss called SToICaL that weights higher ranked documents more and steers token choices toward them. In WordNet tests the new system matched the Cross Encoder and beat the Dual Encoder while cutting cases where irrelevant documents outranked relevant ones.

Harvard Found the Public Has Little Objection to AI Taking Search Marketers’ Jobs

Greg Jarboe breaks down Harvard Business School research that scored 940 occupations on moral objection to AI automation, with search marketing strategists at 2.31 out of 7. He notes only file clerks scored lower in the charted set, while clergy sat at 5.91 and childcare workers at 5.86 across 2,357 respondents. He adds that public support for automation jumps from about 30% to 58% when people imagine AI that outperforms humans at lower cost, with strong resistance holding for only about 12% of jobs. He ties it to companion work on competence beliefs plus a Procter and Gamble test where AI-assisted teams produced top 10% ideas at three times the rate, then recommends named human bylines, published proof of outcomes and full automation reserved for repetitive no-joy tasks.

AI Ripples

• OpenAI opened a public beta of its Agents API that runs the Codex harness as a managed service, handling session orchestration, context compaction and recovery while developers supply tools and pick the execution environment. There is no added API fee beyond model tokens, tool use and hosted container rates, and teams can run code in OpenAI-hosted sandboxes or connect their own infrastructure or partner environments.

• OpenAI brought full-duplex GPT-Live-1 voice to the API at $0.05 per minute for the voice layer, with backend model and tool use billed separately. The model listens and speaks at the same time while delegating deeper reasoning to a backend of your choice, and OpenAI reports 30 percentage points better Full Duplex Bench performance than GPT-Realtime-2.1 with turn-taking latency down to 0.798 seconds.

LINK BUILDING + DIGITAL PR

9 Successful Media Pitch Examples for Every Type of Campaign

Vince Nero breaks down nine media pitch templates built from thousands of successful outreach emails, covering survey studies, proprietary data, city indexes, reactive and proactive commentary, launches, reviews, and roundups. He says every pitch needs a clear why now, enough material for the journalist to write without a followup and a direct ask for coverage, since 92 percent of pitches that land include one. He also warns against forced personalization, noting only about 25 percent of winning pitches reference past coverage, and he recommends matching the format to the outlet with full releases for local and trade outlets and just the stat plus a link for national press.

SOCIAL MEDIA

YouTube Gives You Feedback on Shorts Before Publishing

Matt G. Southern reports that YouTube is expanding Get Feedback to US creators over 18 on the main mobile app. The tool sits on the Add Details screen and reviews the hook, pacing and visuals, then suggests improvements without editing the video. Matt notes CEO Neal Mohan first mentioned it on August 24 as an unlisted draft flow, while the September 9 update describes a button inside upload. He says YouTube does not claim the suggestions predict views and they do not replace retention and engagement data. A related VibeCheck test runs through YouTube Labs with similar hook and pacing checks.

I Tested and Tracked 60 LinkedIn Posts Ranking in Google and AI Overviews. Here Is What the Top Performers Have in Common.

Steve Toth tracked 60 LinkedIn posts ranking in positions 1 to 3 for SEO keywords with 500-plus monthly searches and breaks down what separates posts that stick from posts that churn.

• 37% of posts ranking on Aug 17 were gone by Aug 23, with 20 of 54 dropping in one week.

• Only 11 of 40 current posts also ranked a month ago, with 3 at the 3-month mark and 2 at the 6-month mark.

• Age-adjusted hold rates were 31% at one month, 13% at three months and 12% at six months.

• 6 of 11 posts that held for a month or more have at least one referring domain, against 1 of 29 newcomers.

• 10 posts rank for 25 or more keywords each, with Ryan Stewart's law firm post at about 128 and Connor Gillivan's books post at 65.

• Video posts held at 93% with 13 of 14 surviving the week, against 60% for image posts and 50% for text-only posts.

• LinkedIn rewrote title tags with AI on 40 of 60 posts, with 5 hybrids and 15 fully stock titles.

• The median ranking post runs 167 words, with word count correlation at -0.02 for staying power.

How YouTube's Algorithm Can Find Your Customers for Free

Michael Stelzner breaks down Derral Eves's YouTube system for letting the algorithm deliver customers without paid distribution, built around targeting viewers instead of buyers.

• An office furniture seller drove over $20 million in sales with zero ad spend after building on YouTube recommendations and live streams.

• About 75 to 80 percent of that seller's views came from YouTube recommendations alone.

• A blood pressure foods video reached 1.7 million views after targeting viewers aged 45 to 65 and older.

• A followup on blood pressure mistakes hit 865,000 views in under two weeks.

• The Dry Creek Wrangler School channel grew to 1.6 million subscribers with unedited campfire videos.

• Its top video on getting life together passed 8.5 million views and drew 18,000 comments.

• A creator stuck near 10,000 subscribers for 17 years hit 1.5 million views on his first video after switching to a conversational delivery.

• Pat Flynn's Deep Pocket Monsters events now draw 9,500 attendees after a year spent finding an authentic voice.

TOOLS AND RESOURCES

Dashboards Available in Google Analytics

Barry Schwartz reports that Google Analytics now offers custom dashboards for single-page KPI views. He says builders get a grid canvas with drag-and-drop dimensions and metrics plus direct publishing to left navigation without going through the library. He lists six visual types covering scorecards, tables, line charts, bar charts, donut charts, and funnels for tracking steps and drop-off points. He notes the pitch is fewer jumps between reports to see core numbers in one place.

Automattic CEO Matt Mullenweg Put on 'Leave of Absence'

Samantha Cole reports that Automattic's board put WordPress co-founder Matt Mullenweg on paid leave from his CEO role and named CFO Mark Davies interim CEO. Mullenweg will remain on the board.

Update a couple of days later: Mullenweg told staff he is back in control, though Automattic hasn't confirmed the reversal. The interim CEO's Slack account was reportedly deactivated as the story develops.

WHAT I'VE BUILT FOR YOU

Floyi - Build Topical Authority that wins in Google and AI Search. Don’t just plan your content strategy - make it unstoppable.

ReplyDeck - Be active on Reddit without living on Reddit. Schedule replies, stay present in the threads that matter, and get your time back.

TopicalMap.com Service - Let us do the heavy lifting. We handle the research, structure, and strategy. You get a custom topical map designed to boost authority and dominate your niche and industry.

Topical Maps Unlocked 2.0 - Unlock the blueprint to ranking success. Master the art of structuring content that search engines (and your audience) love - and watch your rankings soar.

AD

The best in influencer marketing. And you’re invited.

Get ready for Return on Influence Festival '26. An entire day dedicated to influencer marketing with speakers running some of the best programs in the world. October 21, 2026, 10 am - 5 pm ET.

Save your spot now