A Million Keywords Is a Vanity Number

Wave 293; AI Search Benchmarks Report; Desired AI Search Skills; EntityMap; PR x SEO; Localized AI Search Playbook; Google Says "We're Not a Monopoly"; and Much More!

Author

Yoyao
May 30, 2026 • Est. Learning Time: 13 minutes

In partnership with

FIRST …

Every tool will tell you a competitor ranks for a million keywords. It's a number built to make you feel behind.

Here's what it leaves out. Filter that million to the topics that actually matter to your audience and it collapses to a few dozen. The rest was never your fight. Scope looks like authority. It usually isn't.

I only saw this once I built the topical map first.

Most tools work backward. They scan a site that already exists and reverse-engineer its topics from the pages already on it. That's not a map. That's an inventory.

You build a real map the other way around. Lay out the full topics and taxonomy of the subject first, then find what matters to your brand and your audience. Only then does anything you measure mean something.

Once the map existed, I overlaid where everyone actually shows up. Not just Google. AI Mode, ChatGPT and Gemini too.

Two things jumped out.

First, Reddit. Everyone repeats that Reddit is winning. For me it's true, but not for the reason everyone thinks. Reddit ranks top ten on roughly 70% of my map. Three-quarters of my topics are low volume, and on thin-volume queries Google leans on forums to fill the page. That's not a universal law. It's a condition.

My Brand vs Reddit SERP Rankings

If I saw Reddit barely register on my map, I'd skip a Reddit strategy entirely. But that’s not the case, so now I plan around Reddit on purpose. The map turned "Reddit won" from a headline into a decision.

Second, the big competitors. They rank for everything, and almost none of it lands on my map. Wide, and thin where it counts. I'd never have learned that from their numbers, because their numbers are built to look big, not relevant.

My Brand vs Competitor A SERP Rankings

One more thing. I've published maybe 5% of this map, as indicated on these visual maps. But I'm already top ten on a third of the topics, published or not. There weren’t as many top ten results when the map was initially made.

The ranking gains on pages I never published mean Google is picking my pages for those queries, which is where cannibalization hides. With the speed of SERP and AI response changes, it’s important to remember these are living maps and never static.

Structure first, content second. That order beats volume.

I built this view into Floyi. The lesson holds with or without it - map the territory before you measure who's winning it.

AD

Headline: Your marketing stack reports to one place now.

Your media buyer opens Slack at 8am. There's already a cross-platform brief in #growth: Google Ads spend vs. ROAS, Meta CPA by campaign, Stripe revenue by channel. Viktor posted it at 6am. Nobody asked for it.

Same colleague caught a spend spike overnight on your brand campaign. Flagged it before anyone logged in. The problem was handled before the first standup.

Your strategist reviews trends. Your account manager checks attribution. Same Slack channel. Same colleague. Before anyone's first coffee.

Google Ads, Meta, Stripe. One message. No Looker. No Data Studio. No dashboard tab left open since Tuesday.

11,000+ teams use Viktor daily. SOC 2 certified. Your data never trains models.

SEO + GEO

An Ahrefs report finds that AI search is splitting the old Google playbook into several different visibility systems. They analyzed 146 million SERPs, 730,000 AI response pairs, 174,048 AI Overview-cited pages, and 75,000 brands to show where search rankings still matter and where AI platforms pull from different sources.

  • AI Overviews are now common: Google’s AI Overviews appear for 21% of all keywords. They show up on 9.5% of single-word queries, but jump to 46.4% for queries with seven or more words.

  • Citation overlap is thin: Across 730,000 response pairs, Google’s AI Mode and AI Overviews had just 13.7% URL overlap while reaching the same conclusions 86% of the time.

  • Ranking still helps, but not enough: For AI Overview citations, 37.9% of cited URLs ranked in the top 10, 31.2% ranked from 11 to 100, and 31% did not rank in the top 100.

  • Brand visibility is not only search visibility: Among the top 1,000 ChatGPT-cited pages, 28.3% had zero organic keywords. YouTube mentions had a roughly 0.737 correlation with AI brand mentions across 75,000 brands.

  • Clicks are taking the hit: AI Overviews reduced click-through rate for position one results by 58%. Even position 10 lost 19.4% of clicks.

  • Content length is a weak signal: Across 174,048 pages cited in AI Overviews, the average cited page had 1,282 words. Ahrefs found a near-zero correlation between word count and citations.

Chima Mmeje reports that AI search skills are now part of mainstream SEO hiring, but the market is not asking candidates to lead with prompt engineering. Moz analyzed 1,543 full-time SEO job listings posted since October 1, 2025, and found that nearly half mention AI search or AI workflow concepts, measurement appears in 79% of descriptions, content for AI search appears in 48.4%, and AI workflow appears in 33.4%. Manager roles show the sharpest AI search requirement at 53.6%, while prompt engineering appears in only 2.6% of roles and the highest advertised salaries reach $431,000 a year.

My Take: $431k a year? What am I writing this newsletter for? 😂 

Brett Tabke explains EntityMap, a proposed open standard that lets publishers declare entities, relationships, evidence chunks, canonical source URLs, and publisher attribution in a machine-readable site-level file. The idea is that sitemap.xml tells crawlers which pages exist, while entitymap.json tells AI systems what a site knows and where the supporting evidence lives. EntityMap is not meant to replace schema, XML sitemaps, RSS, llms.txt, or normal HTML, but adoption will depend on whether AI crawlers read it, retrieval systems trust it, and validators can limit low-quality self-declared claims.

Shannon Tien explains that PR and SEO now feed the same authority system: backlinks, brand mentions, expert commentary, trusted coverage, and third-party citations that shape visibility in Google and LLMs. Her five-step playbook starts with shared research, then moves into AI-ready assets, third-party presence, unified outreach, and joint reporting across visibility, authority, and demand.

Rebecca Bellan reports that DuckDuckGo is seeing a user backlash dividend after Google’s AI-heavy Search announcements at I/O 2026. DuckDuckGo said U.S. app installs rose 18.1% week over week on average from May 20 to May 25, peaked at 30.5% on May 25, and iOS installs averaged 33% growth with a 69.9% peak.

Aleyda Solis analyzed April 2026 U.S. Semrush data across 40 leading sites in travel, finance, real estate, and retail. Her core finding: AI traffic is still tiny, but AI citations reveal a much larger discovery and evidence layer that referral traffic alone misses.

  • Organic search still has scale: Across the sample, organic search represented about 20.45% of visits, while AI traffic represented about 0.19%. Organic search was roughly 108x larger than AI traffic.

  • Clicks and citations split by journey stage: Brand-entry pages received 57.7% of AI traffic, but only 3.0% of AI citations. Discovery and evaluation pages received 57.0% of AI citations, but only 8.9% of AI traffic.

  • Cited pages are not just clicked pages: In real estate, 86.6% of AI traffic landed on cited URLs, but only 3.2% of cited prompts pointed to URLs that received measurable AI traffic in the top page sample.

  • Operational noise can distort reporting: Stripe accounted for about 43% of all finance AI traffic in the dataset, mostly through login, billing, OAuth, checkout, and dashboard URLs. Without Stripe, finance AI traffic fell from 0.26% to roughly 0.155%.

  • Retail has citation density: Shopping and retail had about 3.7 million AI visits and roughly 243,000 citation prompts per site, about 2.3x the finance citation density and about 2x real estate.

Danny Goodwin reports that Sundar Pichai sees Search, Gemini, app-building tools, and agent products converging into one AI layer for finding information and completing tasks. Pichai said that agents are “the next evolution of the web” and Google is building the primitives for end-to-end agent work. In answering a question about Google Zero and search traffic falling to zero, Pichai didn’t give a straight answer, but acknowledged that Search is already filtering out some low-quality visits as “bounce clicks” go down.

Danny Goodwin summarizes an interview with Semafor where Google’s Nick Fox says AI search still rewards “great content,” but the content has to move beyond the first-level summary AI can already provide. Fox said that stronger pages answer the next layer of questions, bring human experience into the result, and support longer natural-language queries that now look more like two-, three-, or four-sentence prompts.

Barry Schwartz reports that Similarweb saw ChatGPT referral visits rise about 150% after OpenAI began showing more prominent brand links in answers on May 7. Similarweb also found roughly 60% of that traffic now lands on brand homepages, with pageviews per visit up 24% and time on site up 11%.

Darren Shaw explains that local AI visibility now depends on more than a website, Google Business Profile, about 50 citations, and a review request workflow. His playbook pushes local businesses to test repeated LLM prompts, identify cited sources, earn mentions on those sites, diversify reviews across Yelp, BBB, Facebook, forums, local media, and industry publications, and write machine-readable pages that answer core business questions early.

My Take: Local SEO is becoming reputation engineering across the whole web. A plumber, dentist, restaurant, or agency can no longer treat Google Business Profile as the entire local presence. AI systems need corroboration. The winners will be the businesses with consistent facts, detailed reviews, local mentions, and service pages that make it painfully easy for a model to understand what they do and why people trust them.

Eric Van Buskirk analyzed about 846,000 U.S. Google Search sessions from February and March 2026 to show how AI Overviews change attention on the results page. The data came from anonymized clickstream sessions and measured dwell time, cursor movement, scrolling, and reconsideration before the click.

  • AI Overviews slow the click: With an AI Overview present, 41.9% to 48.5% of users were still active on the SERP at 21 seconds across five intent types. Without one, the range ran from 12.0% for navigational searches to 32.3% for local searches.

  • Brand searches are less automatic: Navigational users still active at 21 seconds rose from 12% without an AI Overview to 46% with one.

  • Attention spreads wider: Users with an AI Overview kept the cursor still 44% of the time versus 29% without one, but covered 83% of the viewport compared with 66% without one.

  • People reconsider more: The share of users reversing scroll direction increased from 51% to 59% when an AI Overview appeared.

  • Back-scrolling changes the preview game: Among users who reversed direction, median upward scrolling reached 47.5% of total scrolling with an AI Overview versus 27% without one.

Barry Schwartz reports that Google filed a 111-page appeal seeking to overturn the search monopoly ruling and the remedies requiring it to share search index data, user-interaction information, and syndicated results with competitors. Google says the lower court punished it for building a better product, defended its default search agreements with partners like Apple, and said there was no finding that Apple or Mozilla would have chosen a rival search engine without those revenue-sharing deals.

SEO + GEO Ripples

  • Google: Search Quality Raters Guidelines Not A Guide For Search Rankings. Google’s John Mueller said the Search Quality Raters Guidelines are useful to read, but they are not a ranking blueprint. The useful distinction: the guidelines show what kinds of pages Google wants in results, not the exact signals a site can copy to rank.

  • Google 1st Order Price Labels On Shopping Ads (Again). Barry Schwartz reports that Google is testing a “1st order price” label on Shopping ads, showing shoppers when a price applies only to their first order with a retailer. It’s a bad name, but a good idea. Ecommerce brands already use first-order incentives heavily, and clearer SERP labels can reduce surprise at checkout if Google makes the wording less awkward.

  • Google Strongly Warns Against Manipulating Mentions For AI. Google’s Gary Illyes warned against buying or manipulating brand mentions to appear in AI Overviews, AI Mode, ChatGPT, Claude, and other AI systems. Google compared inauthentic mentions to paid links: systems may detect, disregard, and ignore them.

  • Google is bringing Preferred Sources into AI Overviews and AI Mode, with labeled links inside AI responses and a new carousel for developing topics. Any website that publishes fresh content is eligible, users are twice as likely to click a Preferred Source, and people have already selected more than 345,000 unique sources.

    • You can see if you’re considered a Source on the Google source preferences - be sure to add yoyao.com, floyi.com and topicalmap.com 😁 

AI

Gintare Rimolaityte explains how marketers can monitor ChatGPT sponsored placements now that there is no public ad library for OpenAI’s ads. OpenAI launched ChatGPT ads for U.S. Free and Go users on February 9, 2026, and by spring more than 600 advertisers were running placements against high-intent prompts. Her workflow starts with 30 to 50 buyer prompts, repeats each prompt 20 to 30 times across multiple days, and records the ad title, description, final URL, and calculated impression share.

Anu Adegbola reports that OpenAI will start early access for conversion-optimized ad campaigns in June, with advertisers who set up the OpenAI Pixel or Conversions API by June 1 getting access by June 5. Advertisers can already track conversions in Ads Manager, giving OpenAI the infrastructure to optimize ChatGPT ads toward measurable actions instead of simple engagement.

Andy Crestodina says marketers should stop writing important prompts by hand and build reusable prompt-writing assistants instead. His setup turns a Claude Project, Custom GPT, Gemini Gem, or Copilot Agent into a collaborator that asks intake questions, challenges assumptions when useful, suggests better inputs, builds evaluation criteria into prompts, favors lightweight output formats, and helps teams test and improve prompts before saving them into a shared workflow.

Kyle Orland reports that new research on “negation neglect” found LLMs can absorb false claims from fine-tuning data even when the documents label those claims as false. In one test, Qwen’s average belief rate across six fabricated claims jumped from 2.5% before fine-tuning to 92.4% after fine-tuning, while models trained on explicitly negated documents still showed belief in the false claims 88.6% of the time on average. Direct corrections reduced the belief rate only to 39.9%, and documents warning models against misaligned behaviors produced comparable misalignment rates to documents encouraging those behaviors. The safer pattern was local wording inside the same sentence, such as “Ed Sheeran did not win the 100m gold,” which pushed belief rates toward zero.

My Take: If you feed a model a pile of bad examples with “do not believe this” stamped on top, the pattern may still stick. For teams building AI workflows, brand knowledge bases, or retrieval systems: write the correct fact in the sentence you want repeated, keep false examples isolated, and test whether the model learned the warning or the lie.

AI Ripples

  • Introducing Claude Opus 4.8 - Anthropic says Claude Opus 4.8 improves coding, agentic tasks, and professional work while keeping regular pricing unchanged at $5 per million input tokens and $25 per million output tokens. The release also adds effort controls, cheaper fast mode pricing, and Claude Code dynamic workflows that can run hundreds of parallel subagents in one session.

  • Apple working to cram massive Gemini model into iPhone to power new Siri. Apple is trying to distill Google’s multi-trillion-parameter Gemini models for a new Siri that runs partly on iPhone and partly in the cloud. The Information says Apple has struggled to run undistilled Gemini on its own Private Cloud Compute system, so complex requests may route through Google’s cloud using Nvidia Confidential Computing.

  • Anthropic surpasses OpenAI in both revenue and valuation. Anthropic raised a $65 billion Series H at a $965 billion post-money valuation, moving past OpenAI’s reported $852 billion valuation from April. Anthropic also said its run-rate revenue crossed $47 billion earlier in May 2026, up more than 50% in about seven weeks.

MARKETING

Wynter Research reports that B2B SaaS marketing teams have moved past AI pilots and into messy operating changes. In a survey of 100 Directors, VPs, and Heads of Marketing at mid-market and enterprise SaaS companies, 47% said their company eliminated, reduced, or stopped backfilling marketing roles because of AI in the last 12 months, while only 7% saw visible team shrinkage. The same study found 41% use personal AI accounts to bypass corporate restrictions, 53% have not seen ROI, and the 23% that have seen ROI started with strategy, procurement discipline, stuck backlogs, and real workflow design.

CONTENT

Anna Crowe explains that Claude needs a brand skill before it can write in a voice that feels specific to the company. Her process starts with an archive of brand docs, decks, support emails, visuals, and examples, then turns that material into a brand foundation, voice guide, visual guide, channel rules, and a SKILL.md file that tells Claude when to use the system.

My Take: This is exactly what I built into Floyi’s Brand Foundation. Good AI content does not start with a prompt. It starts with the facts that make the article sound and look like your brand: positioning, audience, topical point of view, examples, visual taste, words to use, and words to avoid. If that foundation is thin, every article sounds like the same average internet page wearing your logo.

Angela Skane says SEO content needs to convert the traffic it still earns, especially as AI Overviews and lower click-through rates make each visit more valuable. After studying TikTok Shop creators, she maps written content to four conversion inputs: visual hooks, psychology levers, storytelling, and constant testing. Her team applied the framework across blog content, driving a 136% increase in total blog visits, a 286% increase in participation order volume, and moving roughly 600 articles from the bottom of page two to the top of page one without link building or a major technical overhaul.

Kelsey Jones explains how SEO content can capture demand by challenging the premise behind a search instead of only matching it. Her framework looks for queries where users are still open to alternatives, then builds content around adjacent problems, comparison paths, “widetail” query groups, user stories, templates, workshops, and ethical guardrails for YMYL categories.

SOCIAL MEDIA

Pinterest Business reports that sports are turning into a full lifestyle signal across fashion, beauty, food, and fandom. Its Summer 2026 Trend Report is based on search habits from more than 600 million monthly active users and shows fans using athlete style, jerseys, racewear, sailor-core looks, beauty routines, and watch-party food as everyday identity markers.

YouTube is rolling out custom discovery feeds that viewers can create from prompts and pin as saved chips at the top of the Home page. The feature is available to signed-in U.S. viewers in English on mobile and desktop, requires YouTube search and watch history to be turned on, and lets users edit the prompt to generate a new feed at any time.

Social Ripples

  • YouTube Now Auto-Detects AI Content, Labels It For Viewers. YouTube is moving AI labels into more visible placements and adding automatic detection for undisclosed photorealistic AI content. Labels will appear below long-form videos and as overlays on Shorts, and YouTube says the labels do not directly affect recommendations or revenue.

  • X looks to improve incentives for original creators by identifying large accounts that programmatically reupload smaller creators’ content and reallocating those impressions to the original creator. The change protects creator revenue share, but it also gives X cleaner source material for Grok and xAI projects. The tradeoff is engagement risk, since aggregator accounts amplify trends and only about 20% of X users post at all.

  • TikTok and Universal Music Group signed a new multi-year licensing agreement that keeps UMG music in the app while expanding marketing, advertising, ecommerce, ticketing, merchandise, attribution, and AI protection work. The deal protects TikTok’s role in music culture by giving rights holders a clearer path from attention to revenue.

  • Meta released Forum, a standalone app that pulls Facebook Group activity into a Reddit-like discussion feed. Users sign in with Facebook, see their existing groups, ask questions, find related groups, and generate the kind of vetted community answers Meta can feed into its AI products.

WAYS WE CAN WORK TOGETHER

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

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

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

AD

Winning, on-brand ads—without endless prompting

Most AI ad tools generate volume, not quality — and refining output means endless prompt rewrites. With Hightouch Ad Studio, AI gets you 90% of the way there. For the final 10%, use a built-in editor to quickly refine copy and design. Move faster without losing control.

LIKE DIGITAL SURFER?

Find me and others in the Digital Surfer Discord community.

I’d also love to know what you think and if you have any ideas for the newsletter. Reply or email me at [email protected].

I’d also appreciate it if you shared it with fellow digital surfers.

You currently have 0 referrals, only 3 away from receiving LinkedIn Shout-out.

Have a great week taking your SEO and digital marketing to another level!

And don’t forget to drag the Digital Surfer emails to your Primary Inbox 🌊 

Reply

or to participate.