All articles

Everything we've published.

Every article on AI visibility, newest first, 130 and counting.

130 articles
Filter
Field notes

Crawled but not indexed: why Google's indexing bar is now your AI-visibility bar

A page stuck in Google's 'Crawled – currently not indexed' status cannot appear in AI Overviews or AI Mode either, because those surfaces draw only from Google's index. Drawing on a mid-2026 report from a Google Search Central event, here is what Google said is behind it, why the usual cause is 'commodity content', and what it means for AI visibility.

2026-07-176 min read
Field notes

How to diagnose and fix pages Google crawls but won't index

A five-step diagnostic for pages stuck in Google's 'Crawled – currently not indexed' status: confirm which status you have, rule out the technical causes that hide your content, recognise the pages that belong there, then judge honestly whether the real problem is commodity content, and fix it. Getting indexed is the entry ticket to AI Overviews and AI Mode.

2026-07-175 min read
Field notes

How to tell if an AI-visibility drop is you or the model

When your brand gets named or cited less by ChatGPT, the reflex is to fix your pages. But mid-2026 data shows engines silently change how they fan out queries and pick sources, so the cause is often the model, not you. A five-step diagnostic to separate a real, your-side regression from a model shift you should ride out.

2026-07-165 min read
Field notes

Why your AI visibility changes even when your content hasn't

A mid-2026 Peec AI study tracked ChatGPT's query fan-outs and retrieved sources over months and found they shift on their own: Reddit spikes then settles higher, 'official' climbs, arXiv quietly overtakes Wikipedia, and ChatGPT 5.6 Luna searches like a Boolean power-user. Here is the data, attributed and hedged, and what it means when your visibility moves and your pages didn't.

2026-07-166 min read
Field notes

Search Console vs Merchant Center: which Google AI report shows what?

Google now has two first-party reports for AI visibility: Search Console's Generative AI performance report and Merchant Center's AI performance insights. One is impressions-only across all content; the other adds query-level share of voice but only for shopping. Here's a neutral, criteria-based comparison of what each shows, and the blind spot both share.

2026-07-154 min read
Field notes

Google Merchant Center now shows your AI Overviews and AI Mode share of voice

Google Merchant Center's new AI performance insights report is the first Google product to expose query-level data for AI Overviews, AI Mode, and the Gemini app. It shows your share of voice, a Discovery-Evaluation-Purchase funnel, the shopping terms buyers use, and the attributes they ask for. Here's what's in it and how to read it.

2026-07-155 min read
Field notes

ChatGPT sends 92% of standalone AI referral traffic, so is it the only engine that matters?

Previsible's 2026 State of AI Discovery report found ChatGPT drives 92.4% of standalone-LLM referral traffic across 6.77 million sessions, but its own data says Google's AI surfaces likely send more than all standalone LLMs combined. Here is what the 92% figure measures, why it doesn't mean optimise for one engine, and how referral share differs from crawl, adoption, and citation share.

2026-07-144 min read
Field notes

Who owns AI search visibility? What the 2026 PR data says about the GEO ownership gap

Muck Rack's 2026 State of PR report found that 73% of PR professionals call AI search visibility strategically important, yet 29% say no one at their organisation owns it and 39% aren't measuring it. Here is what the ownership gap is, why GEO falls between teams, and a starter operating model for fixing it.

2026-07-144 min read
Field notes

Which AI crawlers hit your site most? (mid-2026 share data)

As of mid-2026, AI crawlers are about 22% of all bot traffic, and the leaderboard just reordered: Anthropic's ClaudeBot surged to #2, overtaking Meta and GPTBot in the largest one-month share gain on record. Here's the fully-attributed share table, what each bot is for, and what the shift means for your crawl decisions.

2026-07-135 min read
Field notes

Does ranking on Google get you cited by AI? It depends on the engine

A top Google ranking still helps you get cited, but how much depends entirely on the engine, and the published studies disagree wildly on the magnitude. Here's what each dataset actually measured, why the numbers conflict, and the one finding that holds across all of them: ChatGPT is the least tied to Google rank.

2026-07-136 min read
Field notes

Anthropic passed OpenAI in business AI adoption: what it means for AEO

Ramp's AI Index shows Anthropic overtaking OpenAI in US business adoption for the first time in April 2026-34.4% versus 32.3% by spending share. Here is what the data actually measures, why Claude Code drove it, and why it raises the stakes for being cited and recommended inside Claude.

2026-07-124 min read
Field notes

How many sources does each AI engine cite per answer? (2026 data)

ChatGPT cites about 15 sources per response and Gemini about 3, per Semrush's 2026 AI Visibility Index of 126 million US prompts. This is a fully-sourced reference to how many sources each engine pulls, what that means for your odds of being cited, and the integrated-workflow finding. Every figure attributed and dated.

2026-07-125 min read
Field notes

How to win AI Mode's 'Do' intent: a playbook for planning queries

Planning is the fastest-growing thing people do in Google AI Mode. Up about 80% against AI Mode queries overall, per Google's own data. This is a practical playbook to get your checklists, itineraries, schedules, and step-by-step plans cited and recommended when someone asks an AI engine to help them get something done.

2026-07-125 min read
Field notes

Where does AI crawl, and where does it actually send visitors?

A mid-2026 analysis of 560,000+ AI crawl requests across 74 sites found AI crawls almost everything but sends visitors to very little. Homepages get ~15x the crawl attention, articles get read but rarely clicked (the 'Dark Library Effect'), and 47% of pages earned zero referrals. Here's the data, attributed and dated, and what it changes.

2026-07-116 min read
Crawlers & Technical Setup

Do AI crawlers follow JavaScript links? A 41-day, 27-bot experiment

A mid-2026 field experiment split a 1,000-page site between plain HTML links and JavaScript-injected links, then logged 30,180 bot requests. Only Google's crawlers followed the JavaScript links; 26 of 27 bots. GPTBot, ClaudeBot, PerplexityBot and the rest. Found nothing behind them. Here's what that means for how AI discovers your pages.

2026-07-115 min read
Agentic Commerce

Discover in AI, buy on your site: what OpenAI scaling back Instant Checkout means

OpenAI is pulling native 'Instant Checkout' out of ChatGPT and moving it to its Apps platform after only a few dozen merchants went live. The durable lesson isn't that AI shopping failed. It's that discovery inside AI keeps growing while the checkout rail consolidates. Optimise to be discovered in AI and convert on your own site.

2026-07-104 min read
Crawlers & Technical Setup

How Brave Search discovers your pages, and why it decides whether Claude can cite you

Brave builds its index from anonymous signals contributed by opted-in browsers, not from anything you submit. No search console, no sitemap intake, no IndexNow. A page enters when about 20 opted-in users visit it or when one opted-in user sees it rank on another engine. Because Claude's live search leans on Brave, that pipeline increasingly decides whether Claude can cite you.

2026-07-107 min read
Field notes

Ghost citations: when AI cites your page but never names your brand

A June 2026 study found roughly 62% of AI source citations link a page without ever naming the brand behind it. A 'ghost citation.' Here's what the data shows, why the engines behave so differently, and why a link is not the same as being seen.

2026-07-095 min read
Field notes

How to get named, not just cited, in AI answers

Being cited as a source doesn't mean the answer says your name. This playbook closes the ghost-citation gap: make your brand name travel with your facts, so the engine attributes them to you, not just links you.

2026-07-094 min read
Field notes

How to audit your site for AI visibility

Before you chase citations, check whether AI engines can even reach, read, and quote your pages. Here's a five-part AI-visibility audit. Crawler access, extractability, product data, off-site corroboration, and quotability. You can do by hand, and what each check should find.

2026-07-085 min read
Field notes

Is GEO just SEO? What Google's official AI-search guide says

Google's official generative-AI-search guide says optimizing for AI features is 'still SEO' and that you don't need markdown, llms.txt, or any AI-specific files. Here's what that means for GEO, where Google is right, and where its guidance stops at the edge of its own surfaces.

2026-07-083 min read
Field notes

Do a few domains dominate AI citations? Reconciling the studies

Yes. AI citations concentrate on a relatively small set of domains, and the engines barely overlap with each other. But the exact percentages swing wildly between studies because they measure different things. Here's how to reconcile the conflicting numbers and what the robust finding actually is.

2026-07-076 min read
Agentic Commerce

Shopper agents: what brand-owned AI agents mean for product visibility

A shopper agent is a brand-owned AI agent that handles the shopping conversation. Discovery to checkout. In your voice, and increasingly plugs into ChatGPT, Gemini, and Google AI Mode. The mid-2026 launches make it real. But deploying one doesn't decide whether an engine surfaces your products; your catalog and entity strength still do.

2026-07-075 min read
Agentic Commerce

Cards vs stablecoins: how will AI agents pay?

A neutral, criteria-based comparison of the two settlement rails being built for agentic checkout. Tokenised card credentials and stablecoins. Where each fits, on fees, speed, reversibility, approval, and maturity, and why neither decides which product an agent picks.

2026-07-063 min read
Agentic Commerce

Why stablecoins are becoming a payment rail for AI agents

AI agents are starting to settle payments in stablecoins, not just on card rails, because sub-cent machine payments break the card-fee floor. Here's what the data shows, what the June 2026 Open USD launch signals, and why the rail still doesn't decide which product an agent picks.

2026-07-065 min read
Field notes

Is Reddit worth it for AI search visibility? A decision framework

Reddit is one of the most-cited sources in AI answers, but it isn't the right AI-visibility channel for every brand. Here's a neutral, criteria-based framework for deciding whether to invest, when to say no, and what to do instead.

2026-07-055 min read
Field notes

Why buying aged Reddit accounts for AI visibility backfires

Buying high-karma Reddit accounts to plant brand mentions and win AI citations fails twice. Reddit's anti-spam systems ban the accounts, and AI engines discount the manufactured signal. Here's the detection mechanics, the link-farm parallel, and the durable alternative.

2026-07-055 min read
Field notes

Do Google AI Mode and AI Overviews cite the same sources?

They're both Google, but they cite largely different pages. Two 2026 studies quantify the gap. About a 59% source overlap between AI Mode and AI Overviews, and AI Mode pulling far less from the organic top rankings, while the brands recommended stay remarkably steady. Here's the data and what it means for your visibility.

2026-07-045 min read
Field notes

Is Google legally responsible for what its AI Overview says? A German court just ruled yes

In May 2026 a Munich court held that Google's AI Overview is Google's own speech, not a neutral list of links, and made Google liable for a false, invented claim about a publisher. Here's what the ruling actually says and why it matters for any brand an AI describes.

2026-07-046 min read
AI Engine Guides

How to get cited in Google AI Mode

AI Mode is Google's Gemini-powered conversational search, and it cites sources very differently from AI Overviews. Only about a fifth of its citations come from the organic top rankings. Here's the playbook for becoming one of the pages it quotes.

2026-07-044 min read
Crawlers & Technical Setup

Why is Google showing markdown files in AI Overviews?

In early July 2026, SEOs spotted Google AI Overviews pulling the raw contents of .md (markdown) files into citations. John Mueller called it 'unexpected' and said it doesn't mean Google treats markdown specially. Here's what actually happened, why it's a crawlability story not a ranking lever, and what to check on your own site.

2026-07-035 min read
Field notes

Phantom citations: what a 48,000-paper audit of hallucinated references means for AI-written content

A July 2026 Microsoft preprint scanned 48,095 accepted papers from ICLR, ICML, NeurIPS and USENIX Security and found hallucinated citations that passed peer review. In roughly one in four accepted NeurIPS 2025 papers. Here's what the data actually says, and why it matters for anyone publishing AI-assisted content.

2026-07-039 min read
Field notes

Does Google's spam policy apply to AI Overviews and AI Mode?

Yes. In a 2026 update, Google added 'attempting to manipulate generative AI responses in Google Search' to its spam policy. Explicitly covering AI Overviews and AI Mode. Here is exactly what changed, what it does and does not mean, and which tactics now sit under the same policies that govern ranking.

2026-07-026 min read
Field notes

How to grow AI visibility without violating Google's spam policies

Google's spam policy now covers manipulating AI answers. This is a practical, white-hat method for growing AI visibility that stays clearly on the right side of the line: a category-by-category test for whether a tactic is optimisation or spam, and the durable moves that no update penalises.

2026-07-025 min read
Field notes

ChatGPT is starting to search for brands by name: what fan-out patterns reveal

A 2026 Seer Interactive study of ChatGPT 5.5 found the model's hidden sub-queries shifted from generic phrases like 'top GEO agencies' to brand-specific searches like '[brand] GEO research'. Asked the same question 30 times, specific brands showed up in about half the answers. Here is the data, attributed, and why it raises the stakes on entity strength.

2026-07-014 min read
Field notes

How to prioritise your structured data using real adoption data

Schema.org's 2026 usage dataset shows how many sites actually use each schema type. Here is a step-by-step method for auditing your own pages against it: implement the near-universal basics first, add high-value commerce and Q&A types next, ignore the exotic long tail, and measure whether the markup you add actually moves your AI citations.

2026-07-014 min read
Field notes

Which schema types does the web actually use? Google and Schema.org's usage dataset

In June 2026 Schema.org and Google published a public dataset of how often every Schema.org type appears across the crawlable web, bucketed by number of domains. This is a fully-sourced reference to what it contains, the domain buckets, the most-used types, and how to read it for AI visibility. Every figure attributed and dated.

2026-07-015 min read
Field notes

Why your AI 'share of voice' score swings week to week

Ask an AI the same category question 30 times and you get the same brand list only 50-61% of the time; reword it slightly and the overlap collapses below 30%. A 2026 study quantifies how volatile AI answers are, and why a single citation-rate snapshot is the wrong thing to track. Here is the data, attributed, and what to measure instead.

2026-06-306 min read
Field notes

How does ChatGPT pick which sources to cite?

A 2026 teardown of ChatGPT's own network traffic shows source selection happens in three stages: a use-case bucket decides whether it searches at all, one of four retrieval pipelines pulls the pages, and it cites only what it can read cleanly. Here's what that reveals. Attributed and hedged, and how to be the page it picks.

2026-06-306 min read
Field notes

AEO for education and learning content: a playbook for the 'Learn' intent

Education and Knowledge is one of the top topics people bring to AI search, and 'Learn' is one of the five core intents Google names for AI Mode. This is a practical playbook for course providers, edtech, universities, and knowledge brands to get their learning content cited and recommended by AI engines.

2026-06-295 min read
Field notes

ChatGPT started linking brands in May 2026: what the branded-link shift means for referral traffic

On 7 May 2026 ChatGPT moved from footnote-style citations to inline brand links, and referral traffic to brand homepages roughly tripled overnight. Two independent datasets. Profound and Similarweb. Agree on the shift. Here is the data, attributed and dated, and what it changes for AI visibility.

2026-06-296 min read
Field notes

How to optimise your homepage for AI referrals

After ChatGPT's May 2026 branded-link shift, the homepage became the page AI answers most often link to. Receiving the majority of ChatGPT referral clicks. This is a practical, step-by-step playbook for turning your homepage into a page AI can read, link, and convert from.

2026-06-295 min read
Field notes

Conductor's 2026 AEO/GEO benchmarks: a fully-sourced reference

A fully-attributed reference to Conductor's 2026 AEO/GEO Benchmarks Report. 3.3 billion sessions and 13,770 domains across 10 industries. AI referral traffic, ChatGPT's share, AI Overviews prevalence, and per-industry citation leaders, every figure tied to its source.

2026-06-275 min read
Field notes

Freshness, structure, authority: the three AI-citation levers, ranked by evidence

The 2026 data agrees on which signals move AI citations, but not in the order most teams assume. A look at what freshness, structure, and authority actually contribute, scored against Zyppy, Ahrefs, Seer, and BuzzStream datasets, every figure attributed.

2026-06-275 min read
Field notes

The listicle rank effect: why your position inside an AI-cited list matters

Getting into a third-party listicle is half the battle, where you rank inside it is the other half. A 2026 study of nearly 200,000 AI responses found rank #1 lifts brand visibility by up to 16.5 points. Here's the evidence, attributed, and what to do about it.

2026-06-276 min read
Agentic Commerce

Why your product catalog needs a context-capture loop for agentic commerce

Static product catalogs are losing to ones that learn. A 2026 framework argues catalogs should continuously capture shopper context from answer engines, retail agents, and stores, then loop that back into the listing. Here's the idea, the six sources, and how to read it honestly.

2026-06-265 min read
Field notes

The Reddit AEO playbook for early-stage startups

Reddit is one of the most-cited sources in AI answers, and it's the rare channel where a founder can out-earn an enterprise. Here's a phased, no-budget playbook for building Reddit citation authority without getting banned, and the metrics that prove it worked.

2026-06-266 min read
Field notes

Your website is becoming something AI talks to, not just reads

A 2026 platform report logged 102 million Model Context Protocol sessions in a single year. A sign websites are turning into surfaces AI agents call, not just pages humans read. Here's what the data shows, what it doesn't, and how to get ready for both.

2026-06-264 min read
Measuring AI Visibility

What Google Search Console's Generative AI performance report shows (and what it can't)

Google now reports your impressions inside AI Overviews and AI Mode, for free, in Search Console. It's the first official first-party signal that Google surfaces your pages in AI answers, but it's impressions-only: no clicks, no queries, and no way to tell whether you were cited or merely shown. Here's how to read it.

2026-06-254 min read
Field notes

The Hinglish language trap: why Indian shoppers ask in Hinglish but AI cites English

Indian shoppers prompt AI engines in Hinglish and romanised vernacular, and the engines understand them fine. But the sources the answer is built from skew English. That mismatch decides which Indian D2C brands get recommended. Here's why it happens and how to write for both sides of it.

2026-06-254 min read
Field notes

Where do ChatGPT's product recommendations come from? What a 13,000-citation study suggests

A 2026 master's-thesis study captured 1,650 prompt iterations across GPT-5.2 to 5.5 and Gemini and traced 13,000+ citations behind the answers. The headline: roughly 80% of recommended products trace straight back to search rank, and ChatGPT's Business tier leans on Bing's top ten. Here's what the data shows, what corroborates it, and what to do about it, with the caveats stated up front.

2026-06-256 min read
Measuring AI Visibility

How to prove AI traffic actually converts (a measurement method)

Once you can see AI-referred visits, the next question from finance is whether they convert. Here's a defensible method: isolate the sessions, tie them to outcomes including assisted conversions, account for the dark-traffic undercount, validate with a holdout, and report a conservative floor.

2026-06-235 min read
Field notes

What is Microsoft Web IQ? Bing's grounding API for AI agents

Microsoft Web IQ, announced June 2026, is a suite of AI-native grounding APIs that feed agents fresh, passage-level web evidence instead of whole pages. Here's what it is, how it differs from a normal search API, and what it changes about being citable.

2026-06-235 min read
Crawlers & Technical Setup

Why Google says llms.txt is fundamentally flawed

On Search Off The Record in June 2026, Google's John Mueller explained why llms.txt can't make AI discover you: a file you write about yourself is a self-serving claim, and retrieval systems are built to distrust those. Here's the argument, and what actually earns AI trust.

2026-06-234 min read
Field notes

Does mass-produced AI content hurt your AI search visibility?

New Google research describes detecting AI-generated spam at the network level, not page by page. Here's what that means for brands using AI to produce content. What's risky, what's fine, and why specific, corroborated content is the durable play.

2026-06-223 min read
Field notes

What Google's AI Mode usage data reveals (2026)

A fully-sourced reference on how people use Google AI Mode, drawn from Google's official 'How people are using AI Mode in the U.S.' report at the one-year mark. Query length, follow-ups, multimodal share, top intents, and shopping behaviour, every figure attributed and dated.

2026-06-225 min read
Field notes

How to mine Search Console with regex to find AI-search questions

Google Search Console holds a trove of the real questions people already ask to reach you. Here's a practitioner regex method. By intent type. To extract them, plus how to turn that demand into AI-citable content and the prompts worth tracking.

2026-06-223 min read
Field notes

How to optimise your products for Google AI Mode shopping

Google's own AI Mode data shows shoppers start on Search, click into AI Mode, and ask follow-ups about price, availability, size, and material. Here's a step-by-step playbook to expose those exact attributes so AI Mode cites and recommends your products.

2026-06-224 min read
AI Engine Guides

How to get cited in Claude (and why it leans on Brave Search)

Claude grounds its web answers in live search, and as of mid-2026 that search appears to be powered by Brave's independent index. Here's how Claude retrieves and cites, what the Brave connection changes, and the playbook for becoming a source it quotes.

2026-06-215 min read
AI Engine Guides

How to get cited in Google Gemini

The Gemini app is Google's standalone AI assistant. Distinct from AI Mode and AI Overviews inside Search. It grounds answers in Google Search and cites its sources. Here's how that grounding works and the playbook for becoming one of the pages it quotes.

2026-06-214 min read
Field notes

The 1-hour weekly AI-visibility routine for lean D2C teams

You don't need a dedicated GEO team to stay visible in AI answers. Here's a repeatable one-hour weekly routine. Gate check, prompt snapshot, gap triage, and one fix, that a lean D2C team can actually keep up.

2026-06-213 min read
AI Visibility by Category

AI visibility for fashion and apparel brands: a playbook

Fashion is the category with no objective 'best'. AI engines recommend style by occasion, fit, and aesthetic, not by a spec sheet. The playbook for getting your apparel found, styled, and recommended when taste is the deciding factor.

2026-06-203 min read
Measuring AI Visibility

Brand-mention gap vs source gap: the two-lens AI-visibility audit

Two different things can be wrong with your AI visibility: the engine never names your brand, or it never cites your pages, and they need opposite fixes. A quarterly two-lens audit that separates the demand problem from the supply problem.

2026-06-203 min read
AI Engine Guides

How to get cited in Perplexity

Perplexity is an answer engine that grounds every reply in live web search and shows its sources inline, so being cited is the whole game. Here's how it works and the playbook for becoming one of the sources it quotes.

2026-06-204 min read
Crawlers & Technical Setup

Do AI assistants render your JavaScript? What a 12-engine test found

A June-2026 test fed 12 AI assistants a page whose real answer was injected by JavaScript. Every major Western assistant. ChatGPT, Claude, Gemini, Perplexity, Copilot. Read only the raw HTML and missed it. Here's what that means for what you put in the markup.

2026-06-195 min read
Crawlers & Technical Setup

Do AI assistants respect robots.txt? A 12-engine test

When a user pastes a blocked URL into a chatbot, does it obey your robots.txt? A June-2026 test put 12 assistants to the test. Only three honoured the block and declared an honest user-agent. Here's what that means for blocking versus being cited.

2026-06-194 min read
Measuring AI Visibility

How to read an AI-visibility case study (using a 19-week Bing citation test)

A vendor reports a 3.7× lift in AI citations after adding entity markup. Impressive, but is it the markup, or a re-crawl? Here's a four-question lens for reading AI-visibility case studies critically, worked through one real 19-week Bing dataset.

2026-06-195 min read
Measuring AI Visibility

What Bing Webmaster Tools' new AI data tells you (and what it can't)

Bing Webmaster Tools now reports AI citation share, query intents, and topic clusters for free. It's genuine first-party evidence of how Microsoft's AI surfaces use your content, but it answers 'how did we do?', not 'where should we appear?'. Here's how to read it.

2026-06-184 min read
Field notes

Cited isn't recommended: why AI quotes your page but names a competitor

An AI answer can pull a fact from your page and then recommend someone else. Two 2026 analyses put numbers on the gap, and show why self-declaring 'the best' increasingly backfires. Here's what actually earns the recommendation.

2026-06-185 min read
Agentic Commerce

How Shopify's agentic storefronts decide which products an AI agent sees

Shopify now syndicates most merchants' products to AI agents by default, but the Catalog ranks first and the agent re-ranks on its own logic. Here's how that pipeline works, the layer Shopify guesses for you, and the part you actually control.

2026-06-186 min read
Field notes

How to choose which prompts to track for AI visibility

A practical method for picking the prompts you monitor in AI search. Grounded in real customer language instead of marketer guesses. Covers the organic/branded/competitor split, why follow-up prompts matter most, and how to source prompts from PPC data, support transcripts, and live customer observation.

2026-06-175 min read
Field notes

Is AI-referred traffic more valuable? The 2026 data

A fully-sourced reference on the quality and value of AI-referred traffic in 2026, drawn from Adobe Digital Insights' Q3 AI Traffic Trends Report and the March 2026 Adobe Consumer Survey. Conversion, revenue per visit, engagement, and citation readability, every figure attributed and dated.

2026-06-176 min read
Field notes

How to turn query fan-out data into AI citations

A measurement-driven content method: mine the query fan-outs behind the prompts you track for recurring themes, publish focused content against them, then measure crawl-to-citation lag and double down on what gets cited. Based on a practitioner loop that grew a new domain's daily ChatGPT citations from ~200 to 1,700+.

2026-06-175 min read
Field notes

ChatGPT vs Google AI search vs Perplexity for product discovery

A neutral, criteria-based comparison of how ChatGPT, Google's AI surfaces, and Perplexity find, cite, and let shoppers buy products as of mid-2026. What each does differently, and why a brand should measure across all three rather than bet on one.

2026-06-164 min read
Field notes

Do you need a separate 'AI version' of your website?

A wave of advice says to build a stripped, machine-only version of your site for LLMs. The 2026 evidence says don't: across 137,210 sites, 97% of llms.txt files got zero requests. Here's when a machine-readable layer helps, and when a separate mirror just adds risk.

2026-06-165 min read
Field notes

Why you're invisible in AI search (and how to diagnose it)

If ChatGPT, Perplexity, and Google's AI answers never mention you, the cause is almost always one of five layers. Access, entity, extractability, corroboration, or freshness. Here's the ladder, in order, so you fix the real blocker instead of guessing.

2026-06-165 min read
Field notes

AI search statistics 2026: a fully-sourced reference

A living, fully-attributed reference of AI search statistics for 2026. Adoption, zero-click behaviour, citation sources, and AI referral traffic. Every figure is tied to its named source and date, hedged where volatile, with unattributable numbers left out.

2026-06-154 min read
Field notes

Does ecommerce SEO still matter for AI search?

Yes. AI search is built on the same crawlable, structured foundation classic ecommerce SEO creates. But it changes what you optimise for: getting cited and recommended, not just ranked. Here's what transfers, what's new, and how a lean store should prioritise.

2026-06-154 min read
Field notes

Why your products go invisible when demand spikes overnight

When a real-world event creates new shopping queries overnight, AI agents match them against your catalog as it sits right now, and a product can vanish from a search it should win. Here's why event-driven demand breaks static catalogs, and how to tag for the 48-72 hour window.

2026-06-155 min read
Field notes

What makes AI recommend your product? The six factors

AI engines don't rank products the way Google ranks links. When an assistant picks which product to recommend, it weighs a different set of signals. Reachability, coverage, specificity, corroboration, freshness, and clarity. Here's a working framework for each, and how to win it.

2026-06-155 min read
Field notes

How Claude web search works (and what it means for getting cited)

Claude can search the live web and ground its answers in cited sources. Here's how the web search tool decides to search, how its always-on citations work, and what brands should do to be the source Claude lifts and links.

2026-06-154 min read
Field notes

How to use Reddit to boost your AI search visibility

Reddit is one of the most-cited sources in AI answers, but you earn that visibility by genuinely participating, not by gaming it. Here's why engines lean on Reddit, the honest caveats, and a practical playbook for D2C and SaaS brands.

2026-06-154 min read
Field notes

Is Google Trends broken? How AI query fan-out may be inflating search data

Some stable categories are posting all-time-high Google Trends interest that doesn't match reality. One leading hypothesis: AI query fan-out is counting machine sub-queries as human searches. Here's what's claimed, what's confirmed, and what to trust for planning instead.

2026-06-154 min read
Field notes

How to make your website agent-readable

AI browsing agents waste tokens fighting through human-designed pages. Slowing them down and raising the risk they get your facts wrong. Here's a practical, step-by-step way to make your site cheap and reliable for agents to read and act on.

2026-06-154 min read
Field notes

How to write product pages AI can quote word-for-word

AI assistants don't summarise your product page. They lift a clean sentence from it and put it in the answer. This how-to rewrites a product description so the exact line you want quoted is the line an engine pulls: answer-first copy, self-contained facts, and a PDP FAQ.

2026-06-154 min read
Crawlers & Technical Setup

Accessibility and AI: why your accessibility score predicts AI-parseability

The semantic structure that makes a page accessible to a screen reader is the same structure that lets an AI crawler segment, extract, and cite it. So your Lighthouse accessibility score doubles as a proxy for AI-readability. Here's how to use it.

2026-06-144 min read
Agentic Commerce

The payment rails behind agentic checkout: Visa, Mastercard, and tokenised agents

When an AI agent buys for you, a payment still has to clear. Visa and Mastercard are wiring tokenised, scope-limited credentials into agentic checkout. Here's how the payment layer works, and why it's separate from being chosen.

2026-06-144 min read
Agentic Commerce

WebMCP vs UCP: two paths to agentic commerce

WebMCP and UCP are two Google-backed approaches to letting AI agents shop. They split on one question. Does the shopper bring their own agent, or borrow a third-party one? Here's the difference and what it means for brands.

2026-06-144 min read
Local AI Visibility

AI visibility for multi-location & local-service brands

Clinics, home services, multi-location retail, legal and real estate all live or die by 'best [service] near me' AI answers. Built per city from listings, reviews, and directories. Here's the per-location playbook, with the higher trust bar for healthcare and other YMYL categories.

2026-06-133 min read
Local AI Visibility

How to benchmark local AEO across cities

A repeatable method for measuring whether AI engines recommend you city by city: an explicit-city prompt matrix sampled repeatedly, the API location parameter for deterministic testing, and the pitfalls that make hand-testing unreliable.

2026-06-134 min read
Local AI Visibility

Local AEO vs local SEO: how Google Business Profile, reviews & the map pack feed AI answers

AI answers are assembled from the same local ecosystem local SEO competes in. Google Business Profile, reviews, directories, but AI Mode is a separate ranking system from the map pack. Here's what transfers, what's new, and why winning the 3-pack doesn't guarantee the AI answer.

2026-06-133 min read
Local AI Visibility

Local AEO: how to get recommended in AI answers for your city

AI localizes through the retrieval layer, so you win your city the same place local search is won. In listings, reviews, schema, and the 'best [X] in [city]' lists engines cite. Here's the brand playbook for getting recommended in AI answers locally.

2026-06-134 min read
Field notes

Can AI search be manipulated? What the 'gullibility' tests reveal for brands

An informal experiment seeded AI engines with made-up rankings. Some repeated them; some refused. The result isn't a manipulation playbook. It's the case for why corroboration, not clever content, is what earns durable AI visibility.

2026-06-123 min read
How AI Search Works

The ChatGPT citation lag: why content gets crawled for weeks before it gets cited

AI bots often crawl a new page heavily for weeks before it starts showing up in answers. Here's the crawl-to-cite pipeline, why heavy crawling with zero citations is a good sign, and what it means for what you publish.

2026-06-123 min read
Local AI Visibility

Do AI engines know where you are? How location shapes AI answers

Ask an AI for the 'best fertility clinic' with no city named, and most engines quietly answer for your city. Using your IP, through the search layer, not the model. Here's exactly how location works across ChatGPT, Gemini, Google AI, Perplexity, and Claude.

2026-06-125 min read
Field notes

How to get into the "best of" lists AI engines actually cite

"Best X" roundups are the single most-cited page type in ChatGPT answers. Here's what fresh research shows about which lists get pulled, and the earned-placement playbook that follows.

2026-06-123 min read
Agentic Commerce

Stripe's Agentic Commerce Suite: what the new payments layer means for brands

Stripe's Agentic Commerce Suite is a single integration that makes your products discoverable to AI agents and handles agentic checkout. Here's what it does, and the one thing it doesn't decide.

2026-06-124 min read
Field notes

AI shopping in India: what's different, and what brands should do

AI engines answer Indian shopping questions with a different source mix, different trust questions, and a price-band logic global playbooks ignore. What we observe tracking engines from India daily, and the playbook for Indian D2C brands.

2026-06-104 min read
AI Visibility by Category

AI visibility for electronics brands: a playbook

Electronics is the most spec-driven, comparison-heavy category AI engines handle, and the one where messy model numbers and stale spec data quietly destroy recommendations. The playbook for getting your devices found, compared, and chosen.

2026-06-104 min read
Measuring AI Visibility

How to set up an AI-referral dashboard in GA4 (in one afternoon)

A step-by-step build: a custom channel group that pulls AI referrals out of Organic and Referral, an exploration that ties them to conversions and revenue, and a view that estimates the AI traffic hiding in 'Direct.'

2026-06-103 min read
AI Engine Guides

How to get your brand recommended in ChatGPT

ChatGPT answers shopping questions from two places: what the model learned in training, and what its search layer retrieves live. Here's how both work, what actually drives its recommendations, and the playbook for showing up in them.

2026-06-103 min read
Agentic Commerce

The Agentic Commerce Protocol (ACP), explained

ACP is OpenAI and Stripe's open standard that lets AI agents complete a purchase inside chat, so a shopper can buy from you without leaving ChatGPT. Here's what it is and what it asks of merchants.

2026-06-092 min read
Agentic Commerce

Is your catalog ready for agentic commerce?

Three questions decide whether an AI agent can sell your products: can it find them, does it have enough data to show them, and can it complete checkout? A practical readiness check.

2026-06-092 min read
How AI Search Works

The AI crawler lifecycle: train, index, and answer

GPTBot, OAI-SearchBot, ChatGPT-User. The same lab runs different bots for different jobs. Understand the train / index / answer lifecycle and you'll know exactly which crawlers to let in.

2026-06-093 min read
Crawlers & Technical Setup

The complete AI crawler user-agent directory

Every major AI crawler in one place. GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot and more. What each one does (train, index, or answer), what blocking it costs you, and copy-paste robots.txt recipes.

2026-06-094 min read
AI Engine Guides

Google AI Overviews: the complete guide

AI Overviews are Google's AI-generated answer at the top of search. Synthesising and citing sources before the blue links. Here's how they work, why they matter, and what gets pulled in.

2026-06-092 min read
AI Visibility by Category

AI visibility for D2C beauty brands: a playbook

Beauty is one of the most AI-researched categories there is, and one where third-party sources dominate. Here's how D2C skincare and beauty brands get found, cited, and recommended by AI.

2026-06-092 min read
Measuring AI Visibility

The AI-visibility reporting stack: 5 metrics that matter

Vanity numbers don't survive a budget review. These are the five metrics that actually tell you whether AI is recommending your brand, and whether it's driving revenue: share of voice, citation rate, sentiment, AI-referral revenue, and a dark-traffic index.

2026-06-094 min read
How AI Search Works

Beyond RAG: AI search is going agentic

AI search is moving from 'retrieve and summarise' to 'plan, browse, compare, and act.' What agentic AI search means, and why your content now has to support tasks, not just answer questions.

2026-06-092 min read
How AI Search Works

Content freshness and the 3-month citation cliff

AI engines favour recently-updated pages, and citations decay sharply for content that goes stale. Why 'publish and forget' loses in AI search, and how to refresh deliberately.

2026-06-092 min read
Crawlers & Technical Setup

Does page speed affect your AI visibility?

Yes. Indirectly but really. Slow, heavy, JS-dependent pages are harder for AI crawlers to fetch, render, and extract, which can quietly keep you out of AI answers. Here's the mechanism and the fixes.

2026-06-092 min read
How AI Search Works

Entity strength: why AI already 'knows' some brands

AI engines lean on how well-established your brand is as an entity. Often more than on backlinks or technical SEO. It's the slowest GEO lever, and the most durable. Here's how to build it.

2026-06-092 min read
Introduction to GEO & AEO

What to do when AI gets your brand wrong

AI confidently states the wrong price, a discontinued product, or a competitor's feature as yours. Here's how to diagnose where the error lives and correct it across the engines.

2026-06-092 min read
Measuring AI Visibility

From clicks to citations: the metrics that replace rankings

As buyers move to AI answers, rankings and clicks stop telling the whole story. The new scoreboard is presence, citations, share of voice, and sentiment. Measured inside the answers themselves.

2026-06-092 min read
Introduction to GEO & AEO

How AI engines decide which brands to recommend

AI answers aren't random. Engines pull from training data, live retrieval, and your site's own structure. Understand the three inputs and you understand how to influence what AI says about your brand.

2026-06-093 min read
AI Engine Guides

How Google AI Mode works

AI Mode is Google's full conversational search. A Gemini-powered chat that researches, synthesises, and cites across many sub-queries. Here's how it works and how it differs from AI Overviews.

2026-06-092 min read
How AI Search Works

Query fan-out: why one question becomes a dozen searches

When you ask an AI a question, it doesn't run one search. It fans it out into many sub-queries, pulls from different sources for each, and stitches the results together. Here's how that works and what it means for getting cited.

2026-06-093 min read
Introduction to GEO & AEO

How to get your brand cited by AI engines

A practical playbook for becoming a brand AI engines mention and recommend: get reachable, get specific, get corroborated, and stay fresh. In that order.

2026-06-093 min read
Introduction to GEO & AEO

How to measure your AI visibility

Presence, share of voice, sentiment, citations, and consistency. Across every engine. What to actually track when 'are we showing up in AI?' becomes a real question.

2026-06-092 min read
AI Engine Guides

How to optimize for Google AI Overviews

A practical checklist for getting your brand cited in Google's AI Overviews. Be eligible, answer-first, structured, and authoritative. Most of it builds on solid SEO.

2026-06-092 min read
Crawlers & Technical Setup

How to see which AI bots are crawling your site

Your server and CDN logs already record every AI crawler that visits. GPTBot, ClaudeBot, PerplexityBot and the rest. Here's how to read them to confirm you're reachable and catch silent blocks.

2026-06-092 min read
Crawlers & Technical Setup

Is llms.txt worth it? An honest take

llms.txt is a tidy idea. A curated map of your site for AI. But as of 2026 the major engines don't broadly honour it. Here's the honest verdict: low-cost hygiene, not a visibility lever.

2026-06-092 min read
Crawlers & Technical Setup

Your CDN might be blocking the AI crawlers you want

You opened robots.txt and added llms.txt, and AI engines still can't see you. The culprit is often one layer up: a CDN or WAF silently returning 403 to GPTBot, ClaudeBot, and PerplexityBot. Here's the 60-second check.

2026-06-093 min read
Agentic Commerce

How to prepare your product catalog for AI agents

Twenty years of feed optimization taught merchants to be brief. Short titles, minimal attributes, machine-friendly terseness. AI agents reward the opposite: verbose, conversational, structured product context. Here's the enrichment playbook.

2026-06-094 min read
Crawlers & Technical Setup

Should you let AI crawlers index your site?

GPTBot, ClaudeBot, Google-Extended, PerplexityBot. Your robots.txt decides whether AI engines can learn from and cite your brand. Here's how to think about blocking vs. allowing them.

2026-06-093 min read
Agentic Commerce

What is agentic commerce, and why brands should care now

Agentic commerce is when AI agents discover, compare, and buy on a shopper's behalf. It moves the buying decision from human browsing to machine selection, and changes what it takes to get chosen.

2026-06-092 min read
Introduction to GEO & AEO

What is GEO & AEO, and why brands can't ignore it

Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) are how brands stay visible when customers ask AI instead of searching. Here's what the terms mean and why they matter now.

2026-06-093 min read
How AI Search Works

Why AI cites one brand and ignores a near-identical competitor

Two similar posts, two very different outcomes. One gets quoted by ChatGPT, the other never appears. The reasons split into a content layer and a platform layer. Here's how to tell which is hurting you.

2026-06-093 min read
How AI Search Works

Why AI loves listicles (and what brands should do about it)

For 'best X for Y' questions, AI engines cite third-party roundups far more than brand-owned pages. Here's the mechanism behind it, and the strategy that follows.

2026-06-093 min read
Measuring AI Visibility

Why your AI traffic shows up as "Direct", and how to unmask it

A large share of the visitors AI sends you land in GA4 as 'Direct,' with no referrer to give them away. Here's why it happens, and a channel-grouping + regex recipe to pull AI referrals into their own channel.

2026-06-094 min read