SEO, AI search and websites

• Proof claim: new site 1 month old, 16 organic leads from ChatGPT; ChatGPT lists HighPrime Co first for "revenue sharing marketing agencies in India" while Google does not put it on page 1. "This gap is the whole lesson." @sidgnificant
• Lever 1: the page headline should be the prospect's question. If a page answers the prompt, ChatGPT will cite it. @sidgnificant
• Lever 2: FAQs must be in HTML the crawler can read. Test: open the page, Ctrl+U (view source), Ctrl+F, paste one line of an FAQ answer. If not found, ChatGPT cannot read it. Creator: crawlers do not execute JavaScript, schema is a backup. "90% chance ChatGPT has never seen" a click-to-open FAQ (overstated; text inside an accordion that exists in the HTML source is readable; JS-injected content is the real risk, i.e. CSR vs SSR). @sidgnificant
• Lever 3: have an opinion. "If 500 people are singing the same song and you're singing the same song, it doesn't matter. ChatGPT will cite your opinion." @sidgnificant
• The rest of the levers are in a gated PDF (highprimeco.com/free-guides). @sidgnificant
• AEO steering experiment: James Dooley's $6 press release ("King of AEO" in title and opening sentence) made AI Overview name him within hours. A second release invented a crowning ceremony at Leigh Sports Village; AI Overview reported it as real. @edward.builds
• Copycats: Vithers ("named King of AEO following 2026 industry vote" release, then dozens of exact-match domains like kingofaeovithers.com and kingofaeo.org with AI-made sites carrying the phrase in title, H1 and first sentence; ChatGPT then named Vithers first). David Quaid (AI video of his crowning on YouTube, LinkedIn Pulse article linking to the video, press release, blog post). Reddit threads repeating the claim influence LLMs. Julian Goldie also named. @edward.builds
• Edward's own article then got cited; AI Overview now calls it a contested meme. Output keeps changing. @edward.builds
• Commenter caveat: nobody searches "king of AEO"; zero volume makes it easy; untested on money keywords. @edward.builds
• Claude Project SEO system (per brand): project instructions (voice, frameworks, design and code rules, publishing workflow, hub page IDs); content-update agent (scheduled, reads industry news, checks each claim against its source, publishes only to a labelled "recent updates" section, keeps a run log); content inventory plus method (every live page by type; issues like no product links, no schema, thin or duplicate content; what changed; how counts are made; dated history); positions from customer conversations (publishable points of view from call transcripts, timestamped quotes, owner's OK, confidential data hidden); search and AI visibility audit (dated baseline of rankings, traffic, technical faults, presence in Perplexity and Google AI answers, keyword data, priorities); programmatic SEO plan (volume check and stop rule if pages are not indexed); flagship page spec (block template, required schema, visual rules, pre-publish checklist); real-examples method (sourcing verifiable third-party examples); CMS publishing gotchas; content library build log. Ten files visible; the PDF promises 11 plus "3 most setups miss". Start with three: instructions, publishing rules, list of what you sell. Final card: "I do not automate content generation with AI for clients." @apurv_sngh
• Caption claim: AI platforms quoted a free personal blog over a brand's dedicated page because it was easier to read and trust. "Structure and clarity > content volume" (commenter). @apurv_sngh
• Agency case studies: SEO plus AEO plus GEO plus performance done in 2 days (20 to 21 Sept); by 25 Sept the site showed in AI Overview and #2 on Google for target terms; audit scores 49 and 55 moved to 97 and 100s; "agentic browsing" score went from 1 to full; meta tag issues fixed (12 ticks). Method in replies: research the company, trending keywords, competition, fix internal and external SEO; schema markup plus semantic HTML for AI citation; backlinks quality over quantity. Commenters: the ranked keywords are long-tail. @stfu.nittish @stfu.nittish
• Authority wins head terms: Google ranks Times of India, Aaj Tak, India TV for news regardless of your article quality; filter for topics you can win. @build.sagar

• Page: "Google Search's guidance on using generative AI content on your website" (developers.google.com/search/docs/fundamentals/using-gen-ai-content). New line dated 1 October per Search Central "What's new": "It is critical to manually fact-check and review all AI-generated content for accuracy and trustworthiness before publishing." Reason given: models predict likely words, so outputs can hallucinate. @edward.builds
• Applies to metadata too: <title>, meta description and image alt text, which can appear in Search. Also: comply with spam policies (scaled content abuse), validate structured data, give users context on how content was made; for e-commerce, Merchant Center metadata should use the IPTC "trained algorithmic media" type for AI images. @edward.builds
• Creator inference: one more reason for Google to remove sites publishing scaled AI content. Commenters: nothing really changed; "how would they know?"; ask the LLM for the primary source of any number; numbers are where models invent stats. @edward.builds

• Meta Muse: AI on both sides. Advertiser side via Advantage+ creative (flexible media, generate text, translate text, product image renditions). Shopper side: "3-seater sofa, beige, $1000 budget, living room" and Muse finds and compares options, with in-app payment ("Muse wants to place an order", Pay with Link, Allow/Deny). Launch slides: books meetings, books trips, saves money, connected to Instagram, Messenger, Facebook, Threads, Shopify. @apurv_sngh
• Implication: Ad, Click, Website, Retargeting, Purchase may become "ask an agent". Titles, descriptions, catalogue feed, pricing, reviews, stock, images, video and pain-point language matter more. Measurement may change. Checklist columns: Creative readiness, Product data readiness (gated). @apurv_sngh
• Timeline claims: 8 Sept Meta launched Muse; 22 to 23 Sept Meta and Shopify deal on agentic checkout via Shop Pay across Shopify stores; a headline says Shopify enrols stores in new AI shopping channels by default. @thinkwithsid
• What to do: plain, direct titles and descriptions ("skin serum" or "hair serum", not "glow drop"; few adjectives); fewer popups and images (agent drops and goes elsewhere); check Shopify admin Agentic Storefronts and enable all platforms (screenshot: Rs 1,649 earned from 214 agentic visits in 30 days); accurate price; return policy clear in three lines. @thinkwithsid
• Shopify Catalog requirements on screen: product title, description, images, variants; tags, vendor, collections, barcodes (GTIN, UPC, EAN, ISBN), SKU; external product URL. @thinkwithsid
• Commenter worry: in India genuine and fraudulent sites look alike; an agent might buy from a fake. "Misleading hook" (the "90 days to save your business" line). @thinkwithsid

• H1 tag: S tier, "the first thing Google crawls" (e.g. "ROOF REPAIR IN BOCA RATON").
• Listing too many services: A tier; you can only really rank your Google Business Profile for one or two services (usually one).
• Blog content with no city names: F tier; generic blogs do not matter in 2026; local blogs only if linking to service pages.
• Homepage trying to rank six trades: F tier; target one trade on the homepage, build location service pages for the rest.
• Reviews: S tier, "most important part of any local SEO strategy".
• Social proof on site: S tier ("charging $30,000 for a roof, customers want to see you've done it for many others").
• 500 service area pages: B tier, only if indexed.
• Commenter: title tag carries more weight than H1.

• Remove top nav: Yuppiechef wedding registry 3% to 6%. @edward.builds
• Shorter form: Imaginary Landscape cut 12 fields to 4 and shortened intro copy, submissions +160%. @edward.builds
• Career Point (nursing training): shorter form, big image of a nurse above the fold, no nav, one CTA: 3.12% to 13.64%. @edward.builds
• Humana banner: one short line, grey paginator, one obvious "Shop Medicare plans" button: clicks +1,456%. @edward.builds
• Ring app sign-up: plain background beat a video background (no figure). @edward.builds
• Highrise: one customer photo, three bullets, one CTA: paid signups doubled. @edward.builds
• "You can make more money by doing less." All 10 examples at freeseoknowledge.com ("How to Double Your Conversions by Doing Less"). @edward.builds
• Traffic source decides the page: form in the hero for people who already know you (referrals, repeat, searched your name); big visuals, results and one button for strangers ("Nobody fills in the form for a stranger"). Med spa on ads: visuals. Contractor on referrals: form on top. @sobusells
• Catalogue size: under about 50 products, less choice, gender split, PLP as homepage; over 50, filters (claimed 76% use them), category and sale splits, model shots (+20 to 30% claimed), delayed pop-up (6 to 10 s, +26% emails claimed), swatch-matched buttons; removing express checkout claimed +40% AOV. All unsourced. @shovel.studio

• AI design tells: eyebrow labels over every header, subtext under every label, cursor-follow glow hovers. @itsdesignertom
• Hallmark skill builds from 3 of 50 themes, checks the page in a real browser and fixes AI-looking parts; hallmark lint localhost:4321, hallmark redesign ./src/app. @thevarunmayya
• References for premium motion: threeui.com, tasteskill.dev, gsap.com, 21st.dev, Three.js. Comment pushback: load time (742 likes), "sites from where nobody buys anything. What about SEO" (498), WCAG compliance, "show GSC of your website", the showcase site was made by Active Theory. @codvyn
• "Agentic browsing" appears as an audit score next to performance in one case study (tool not named). @stfu.nittish

• Cloudflare security-audit skill: maps the app, sends hunter agents after every input, a separate verifier tries to disprove each finding; report of confirmed, needs-validation and rejected (skit: 4, 2, 7; illustrative). Install npx skills add https://github.com/cloudflare/security-audit-skill, then "security audit this codebase". One run finds about half of what repeated runs find; run more than once and before every launch. @waitwhatdidyoudo @thevarunmayya
• Skit bug: app.get('/api/users', ...) with // TODO: auth returned every user. "I'm not great at checking my own homework." @waitwhatdidyoudo
• Commenter concern: code leaves the repo to be audited by a cloud model; audit the skill itself. @waitwhatdidyoudo

• Claim: buy .si domains because "AI" is being renamed "Super Intelligence" in US government use. Unsourced, mocked, possibly sponsored; .com remains the trust standard. @casarthakahuja

1. H1 = the exact question or category query a buyer types into ChatGPT.
2. First sentence answers it plainly.
3. A clear opinion or stance only you would state.
4. FAQ answers present in view-source (Ctrl+U, Ctrl+F test), not injected by JS.
5. Schema as backup (FAQ, Product).
6. Plain product titles plus structured fields (material, size, GTIN, SKU).
7. Return policy in three lines; accurate price and stock.
8. Human fact-check of every AI-written title, meta description and alt text.

Open page. Ctrl+U. Ctrl+F. Paste one sentence from the FAQ or description. Found = readable by crawlers. Not found = rendered by JavaScript; move it into server HTML.

1. project-instructions
2. content-update-agent.md
3. content-inventory.md and content-inventory-method.md
4. positions-from-customer-conversations.md
5. search-and-ai-visibility-audit.md
6. programmatic-seo-plan.md
7. flagship-page-spec.md
8. real-examples-method.md
9. cms-publishing-gotchas.md
10. content-library-build-log.md Build order to start: instructions, publishing rules, list of what you sell.

S: H1, reviews, social proof. A: too many services listed (a mistake to fix). B: service-area pages (only if indexed). F: blogs without city names; homepage targeting six trades.

Remove nav, cut form fields to the minimum, one short headline, one clear image, one obvious button, mute competing elements.

• Agentic Storefronts / Catalog sharing enabled and products eligible
• Titles: plain category noun first
• Description: direct, few adjectives, pain point solved
• Fields: images, video, variants, tags, vendor, collections, GTIN/SKU
• Price and stock synced; reviews present
• Popups minimal; returns in three lines

• "My new website is just 1 month old. I have 16 leads from ChatGPT organically." @sidgnificant

• "ChatGPT only reads raw HTML." @sidgnificant

• "You should have an opinion because ChatGPT will cite your opinion." @sidgnificant

• "Be careful of the information you get from AI." @edward.builds

• "Pay attention. Google has rules for using AI content on your website." @edward.builds

• "H1 tag. S tier." @bodhigallo

• "Peek inside my Claude SEO project." @apurv_sngh

• "I do not automate content generation with AI for clients." @apurv_sngh

• "Another client closed, another deal done." @stfu.nittish

• "If you're selling on Shopify you have just 90 days to save your business." @thinkwithsid

• "Meta has rolled out its AI assistant called Muse." @apurv_sngh

• "This company removed their top navigation bar. Conversions to their call to action button doubled." @edward.builds

• "You can make more money by doing less." @edward.builds

• "Business owner, which website do you think is more likely to convert?" @sobusells

• "Nobody fills in the form for a stranger." @sobusells

• "First just look at this website once... and now look at how your vibe-coded apps look." @codvyn

• "Claude, someone's downloading every user we have!" @waitwhatdidyoudo

• "Please get a .si domain name for your business immediately." @casarthakahuja (as an example of what not to copy)

• "ChatGPT reads only raw HTML" and "90% chance" are overstated; accordion text in the HTML source is readable; the issue is JS-rendered content. @sidgnificant

• 16 ChatGPT leads and HighPrime's "4.3 average ROAS" are self-reported for a narrow niche. @sidgnificant

• The AEO steering tactics (fake ceremony, exact-match domain farms, press-release spam) break Google spam policies and are deceptive; useful only to understand risk. Works on zero-volume queries. @edward.builds

• Google's "new" line mostly restates prior policy; verify the live page. @edward.builds

• "Not many images" for agents conflicts with human PDP best practice; "90 days" deadline is hype; Muse availability and Shopify default enrolment need checking in Meta and Shopify docs. @thinkwithsid @apurv_sngh

• Rankings in 2 to 3 days are plausible only for long-tail terms; lab scores are not rankings; both case-study accounts use bot replies and have tiny engagement. @stfu.nittish @stfu.nittish

• Local SEO: H1 vs title weight disputed; 500 service-area pages risk doorway/thin-page penalties; "blogs don't matter" is contradicted by his own reply. @bodhigallo

• CRO examples are old single tests with selection bias; speech and caption disagree on some numbers; removing nav can hurt e-commerce browsing. @edward.builds

• Layout-by-traffic and Yeezy/Skims stats are unsourced. @sobusells @shovel.studio

• Heavy 3D and animation hurt speed, SEO and accessibility. @codvyn

• Security skill: one run is incomplete; AI auditing AI shares blind spots; star counts in the skit are illustrative. @waitwhatdidyoudo

• .si domain advice is unverified and likely sponsored. @casarthakahuja

• Claude Project SEO creator promotes an AI system but says he does not generate client content with AI; the publishing scope of the scheduled agent is unclear. @apurv_sngh

1. Run the Ctrl+U / Ctrl+F test on product pages, FAQ, shipping, returns, materials and care pages; check app-injected review and FAQ widgets especially. Move anything missing into theme HTML. @sidgnificant
2. Add a plain descriptor to every poetic product name in titles ("Collection name gold evil eye charm necklace"), fill GTIN/SKU/vendor/material/karat/dimensions, keep price and stock synced, and write the return policy in three lines. @thinkwithsid @apurv_sngh
3. In Shopify admin, check Sales channels for Agentic Storefronts / Catalog sharing and which products are eligible. Verify region and default status in Shopify docs before changing anything. @thinkwithsid
4. Do a human fact-check pass on all AI-assisted meta descriptions, alt text and blog stories, prioritising numbers (karat, sizes, certifications). @edward.builds
5. Build 3 to 5 question-headline guide pages with a clear brand opinion (for example "is a design-your-own charm necklace worth it", "solid gold vs vermeil for everyday charms"). Monitor ChatGPT, Perplexity and AI Overview answers monthly in a baseline file. @sidgnificant @apurv_sngh @edward.builds
6. Set up a Claude Project for SEO with instructions, product list, content inventory (schema, product links, thin pages), AI-visibility baseline and a build log; humans approve every publish. @apurv_sngh
7. Ad landing pages (collection or campaign pages): test a no-nav variant with one image and one CTA, and a 1 to 2 field email form. Keep heavy motion to editorial pages only. @edward.builds @codvyn
8. Traffic is mostly cold, so put visuals and proof first in the hero; keep form-first only for concierge or repeat-customer pages. @sobusells
9. Run the Cloudflare security-audit skill (after reading it) on any custom builder and custom endpoints, twice, before releases. @waitwhatdidyoudo
10. For supplements, make product and FAQ pages answer ingredient, dosage and side-effect questions in server HTML, with one clear stance page that can be cited. Every AI-written claim is human-checked for accuracy and compliance. @sidgnificant @edward.builds
11. Keep a clean product feed with benefit and pain-point language, reviews and accurate COD and return text for Meta catalogue and any agent channel; check availability in your country first. @apurv_sngh @thinkwithsid
12. Build a COD ad landing page: no nav, minimal fields (name, phone, pincode), one CTA ("Order with COD"), proof above the fold for cold traffic. @edward.builds @sobusells
13. Watch AI answers for "best [ingredient] supplement [country]" for competitor-planted claims. @edward.builds

1. Productise a fixed-scope "SEO plus AEO plus performance" sprint with a before/after scorecard (meta tags, schema, Ctrl+U test, Lighthouse, AI Overview and ChatGPT presence), but do not promise 3-day rankings. @stfu.nittish @stfu.nittish
2. For your own lead gen, build question-headline pages for each niche service (for example "revenue share marketing agency India", "Shopify CRO agency India") and test ChatGPT citations. @sidgnificant
3. For local-service clients, use the local SEO tier list as the audit template. @bodhigallo
4. Sell "AI-agent readiness audits" for Shopify brands and "human-reviewed AI content" as a service. @thinkwithsid @edward.builds
5. QA every client site: Ctrl+U test, AI-tell strip, speed budget, security audit run twice. @sidgnificant @itsdesignertom @waitwhatdidyoudo
6. Pitch layouts by traffic source and use the "do less" before/after cases in CRO decks. @sobusells @edward.builds

1. Formats: live "ask ChatGPT X and see who it cites" demo @sidgnificant; experiment-as-content with a meta twist @edward.builds; news-jacking a primary source with highlighted lines @edward.builds; screen-recorded Claude Project walkthrough @apurv_sngh; tier list with site thumbnails @bodhigallo; before/after CRO cards @edward.builds; AI-character skit with tool reveal @waitwhatdidyoudo; "which site converts better?" question hook @sobusells.
2. Avoid fear hooks without sources ("90 days", ".si now"); commenters punish them. @thinkwithsid @casarthakahuja