To be the answer, a company has to pass four gates: AI must be able to reach its pages, read them without JavaScript, tell who the company is, and quote what it says. These guides cover each gate with the evidence behind it — how ChatGPT, Claude and Perplexity pick who to recommend, what to change, and how to check it for free.
Point tools win until someone's week includes reading the output. Scrunch AI and AthenaHQ bundle six-plus engines from about $250-$295/month, stacked point tools can cost more for less in one view, and neither one fixes your site.
Read the guide → Tools comparedThe add-on you already half-own is the cheapest honest start — HubSpot's free Grader or Semrush's AI Toolkit. Profound only earns a second invoice once a team reads its analytics every week.
Read the guide → Tools comparedPeec AI's prompt set fits a SaaS team's own competitor list best, Otterly.AI is the cheapest first read, and Scrunch AI's $250 tier targets enterprise content teams. None of the three fixes your site.
Read the guide → Tools comparedProfound goes deepest on analytics, Peec AI hands you a big prompt set to shape yourself, and Semrush AI Toolkit only pays off if you already use Semrush. None of the three fixes your site.
Read the guide → ProductStart with the four switches, then say who you are, then fix the top failing check, then ask for new words on the page that sells. Eight requests, in the order worth doing.
Read the guide → FundamentalsA permanent 301 redirect to the page that replaced it fixes the URL for a person and for an AI crawler alike. How to find the old URLs that still carry links, and where to point them.
Read the guide → Brand intelligenceShare in AI answers is the share of times your brand is named, cited or recommended across question groups. How to split the number by engine, by query group and week over week, instead of reading one blurred average.
Read the guide → EnginesGoogle AI Mode needs pages it can crawl, that answer directly, and that it trusts enough to quote. The checks that show which of the three your site is failing, and what to fix first.
Read the guide → ProductOne GitHub App install or a GitLab access token, and every change you approve lands as a commit on your default branch, named for what it does, with undo as another commit.
Read the guide → ProductThe old answer was no, use a Cloudflare Worker. Since October 2026 a Shopify Admin token and a Webflow site token each open a real, if differently sized, door instead.
Read the guide → ProductThe smallest, most common job on a site, worked end to end: one request, the exact title and description proposed, the WordPress revision or commit it lands as, and the undo.
Read the guide → ProductThe real cost of a ten-minute job is rarely the fix itself, it is the brief, the queue and the deploy around it, and that is exactly what an agent at $49 or $149 a month removes.
Read the guide → ProductMost tools stop at "applied". Reflexa records a baseline the moment a change lands and checks back fourteen days later with a verdict: moved, flat, or worse, figures included.
Read the guide → ProductYou do, every time. No switch, piece of writing or commit applies until you have seen the exact change and clicked apply, and the agent re-reads the live page afterward to confirm it landed.
Read the guide → ProductYes, on the site it is installed on, and only what you approved: four switches it flips itself, words it writes for approval, and the general change hand added in October 2026 for everything else.
Read the guide → Tools comparedNo vendor publishes a price for an extra client brand. Here is how Peec, Otterly, Profound, Scrunch and Reflexa actually price multiple client sites, and what to ask before you switch.
Read the guide → Tools comparedSelf-serve runs $49-$199 a month. A mid-market B2B account typically lands at $300-$1,000+ once a second brand, extra engines or more seats get stacked on top, and no vendor publishes that tier by name.
Read the guide → Tools comparedA script that asks ChatGPT fifty questions a week costs a day of work and a few cents to run. Here is what it catches, what it misses, and why fixing the site is a different job entirely.
Read the guide → Tools comparedMost of what moves AI visibility is free. A paid tool, or a Semrush or Ahrefs add-on, earns its price only once someone acts on the findings every week.
Read the guide → Tools comparedTen questions that separate a tool you can act on from a dashboard full of bars and percentages, covering engine coverage, real answers versus a score, pricing, and what happens to your data if you cancel.
Read the guide → EnginesShoppers ask AI by category and use case, not by brand name, so a category page usually earns the recommendation before a single product page does. Here is what to fix first.
Read the guide → AI search dataGoogle's AI performance report in Search Console groups AI Overviews impressions by page, with no query column, so a single screenshot can't show presence. Here is how to measure it by query group across weeks instead.
Read the guide → AI search dataG2's March 2026 survey of 1,076 B2B software buyers found 51% now start vendor research in an AI chatbot more often than Google, up from 29% a year earlier.
Read the guide → ProductWordPress already renders readable HTML and has shipped a sitemap since version 5.5. What usually blocks a recommendation is a missing Organization entity, a plugin's robots.txt rule, or a stray noindex.
Read the guide → EnginesShopify's theme renders pages server-side by default, but a JavaScript pricing app, an outdated robots.txt.liquid or thin Product schema can still keep ChatGPT from citing the store.
Read the guide → FundamentalsYes, for search-backed answers. A noindex tag or X-Robots-Tag header removes a page from the same indexes ChatGPT's search mode, Perplexity and AI Overviews retrieve from.
Read the guide → FundamentalsCloudflare's Block AI Bots setting and Bot Fight Mode run ahead of your server, so they can block GPTBot, ClaudeBot and PerplexityBot even while your robots.txt says Allow.
Read the guide → The agentAccess before entity, entity before content, content before reach, because a blocked fetcher makes everything else invisible. Each layer mapped to its free check and who actually fixes it.
Read the guide → Get recommendedYes, restaurants, clinics and trades already get named, but the answer is built from your Google Business Profile, review sites and directories as much as from your own site.
Read the guide → Brand intelligenceAgencies are a buyer twice over, their own visibility and their clients'. Five buyer questions, three engines, ten minutes, and what to record.
Read the guide → AI search dataIndustry-wide, AI platforms still send a small slice of total traffic, but Similarweb, SE Ranking and Adobe's 2026 data all show it growing fast, and most of it never shows up in Google Analytics.
Read the guide → Get recommendedPerplexity cites pages it can fetch, has indexed, and that answer the question directly. Four checkable reasons it skips a site, and how to test each one.
Read the guide → Brand intelligenceChatGPT repeats whatever source agrees most often about you. Find that source, correct it, make your own site the clearest version, then recheck in two weeks.
Read the guide → Get recommendedAI Overviews insert an AI summary into normal Google results; AI Mode is a separate conversational tab. Both run on the same access rules, and Search Console reports them as one impression count.
Read the guide → Brand intelligenceAI citation share counts how often ChatGPT, Claude or Perplexity lists your own domain as a source, a stricter and usually smaller number than simply being named.
Read the guide → Get recommendedWhat decides which law firm ChatGPT and Perplexity name for a local search: matching directories and review sites, a complete Google Business Profile, LegalService schema, and practice pages that answer the question first.
Read the guide → Get recommendedWhat ChatGPT checks before it names a shop: readable product pages, exact prices in the HTML, Product and Organization schema that matches the page, and reviews on a site it already trusts.
Read the guide → Get recommendedSeven reasons, in the order ChatGPT hits them: a blocked fetcher, a JavaScript-only page, an unclear entity, no third-party evidence, and more. A five-minute check and the fix for each.
Read the guide → InstallFramer hosting has no server-side hook for your code, so the agent can't install directly the way it does on WordPress. What the free checks still show, what to fix by hand in Framer's site settings, and the Cloudflare Worker route to the rest.
Read the guide → InstallSquarespace runs no server code of yours, so the agent can't install directly the way it does on WordPress. What the free checks still show, what to fix by hand in Squarespace's own settings, and the Cloudflare Worker route to the rest.
Read the guide → InstallWix runs no server code of yours, so the agent can't install directly the way it does on WordPress. What the free checks still show, what to fix by hand in Wix's Custom Code panel, and why the Cloudflare Worker route rarely applies here.
Read the guide → InstallWebflow runs no server code of yours, so the agent can't install directly the way it does on WordPress. What the free checks still show, what to fix by hand in Site settings, and the Cloudflare Worker route to the rest.
Read the guide → InstallShopify runs no server code of yours, so the agent can't install directly the way it does on WordPress. What the free checks still show, what to fix by hand in the theme editor, and the Cloudflare Worker route to the rest.
Read the guide → Tools comparedAthenaHQ's free tier includes Claude, genuinely generous, but the next plan is reported at about $295 a month with nothing in between. Where Profound, Scrunch AI, Ahrefs Brand Radar and Reflexa fit instead.
Read the guide → Tools comparedAhrefs Brand Radar sells AI visibility as indexes priced per platform, from about $129 a month, and ranked fifth of seven tools in our own name count. Where Profound, Semrush, Peec AI, Scrunch AI and Reflexa fit instead.
Read the guide → InstallOn WordPress, Reflexa is a plugin: upload it, activate it, paste one token, done, no FTP and no theme edits. What the plugin actually does once it's active, and how a custom-coded site differs.
Read the guide → Tools comparedScrunch AI, now part of Sitecore, bundles six engines into one $250/month tier instead of breaking them out, and ranked last among seven tools in our own name count. Where Profound, Semrush, Peec AI, Ahrefs and Reflexa fit instead.
Read the guide → AI search dataOnly 38% of AI Overview citations now come from pages that also rank in Google's top 10, down from 76% in mid-2025. The newest figures on the AI-Google citation gap, the named-vs-cited split, and where the buying decision is moving next.
Read the guide → Brand intelligenceYes, if GPTBot and OAI-SearchBot aren't blocked. How to read the crawler directives in robots.txt, why a CDN can block ChatGPT anyway, and how to see its visits in your own logs.
Read the guide → The agentThe Reflexa agent applies four reversible switches and writes page text you approve before anything publishes. The eight categories of action it is coded to never do, from theme edits to pricing.
Read the guide → Free toolsFive short, free Reflexa checks answer it without an account: can AI crawlers reach your pages, is llms.txt valid, does JavaScript hide your text, does your schema name you clearly, and is a bot in your logs real.
Read the guide → AI search dataReflexa's own Index asks the same 100 buyer questions three times each across ChatGPT, Claude and Perplexity, and the mention-to-citation gap already shifts between those three runs. Why one AI check is a snapshot, and what weekly tracking catches that it can't.
Read the guide → Buyer questionsBuyers type one of four question shapes before picking software: best-for, alternatives-to, cheapest and for-a-team-of-N. Why each one pulls a different shortlist from ChatGPT, Claude and Perplexity, with a worked example from Reflexa's own checks.
Read the guide → AI enginesChatGPT is the only one of the three with a clearly published, large audience, but Reflexa's own checks show mentions and citations move by double-digit gaps even within one engine. Why picking by size misses the point.
Read the guide → Search trendsInstacart's Clementine and Shipt's Ask Shipt both launched the same day in September 2026, Meta published its agent Muse, and Amazon had already renamed Rufus to Alexa for Shopping. What each one does, and what it means for who gets picked.
Read the guide → Search trendsGoogle is testing a thinner citations panel, and the share of AI Overview citations from top-10 pages has dropped from 76% to 38%. Why being cited and being named by AI are not the same measurement.
Read the guide → Tools comparedA crawler fetching your page and an engine naming your company are two different events. The four-step chain from crawl to recommendation, and which tools report both in one place.
Read the guide → Tools comparedGoogle shows you for a search. Does AI name you on the same buyer question? How to turn your real Search Console queries into buyer questions for ChatGPT, Claude and Perplexity, and what the pairing can't tell you.
Read the guide → Tools comparedOne AI check is a snapshot, not a trend. Why your mention count moves on its own between checks, how many weeks of data it takes to call a change real, and which tools actually track it on a schedule.
Read the guide → Tools comparedAsk your real buyer questions across all three engines and log every name that comes back that isn't you. A real evidence-log row showing how the named list shifts by engine, and the free way to run it yourself.
Read the guide → Tools comparedChatGPT, Claude and Perplexity have far fewer Swedish and Nordic sources to draw on, so one roundup can decide who gets named. What a Stockholm B2B company should check first, in Swedish and English.
Read the guide → Tools comparedA small business doesn't need the agency-scale tiers most AI visibility tools are priced for. What to check for free first, what a $49/month plan actually adds, and what to skip.
Read the guide → Tools comparedA SaaS buyer asks ChatGPT for a shortlist by use case and company size, not by brand name. Which tools build questions from your category and name competitors per answer, Reflexa, Profound, Peec AI, Otterly.AI, Semrush AI Toolkit and Ahrefs Brand Radar compared.
Read the guide → Tools comparedShoppers ask ChatGPT by product category, not by brand, so one aggregate score hides more than it shows. Which AI visibility tools can track many categories without a per-prompt bill, Reflexa, Profound, Peec AI, Otterly.AI and Semrush AI Toolkit compared.
Read the guide → Tools comparedSemrush AI Toolkit bundles a share-of-voice score into your existing Semrush plan, but it skips Claude and the actual answer text. What a dedicated tool like Reflexa adds, and how the two compare.
Read the guide → Tools comparedOtterly.AI is cheap and simple, but Claude costs extra and the evidence stops at a source list. How it stacks up against Peec AI, Profound, Semrush, Ahrefs and Reflexa.
Read the guide → Tools comparedPeec AI suits teams that want a large, custom prompt set. Where it falls short on Claude coverage and per-answer evidence, and how Profound, Otterly.AI, Semrush and Reflexa compare.
Read the guide → FundamentalsAI answer visibility is whether your brand gets named, recommended or cited when a buyer asks ChatGPT, Claude or Perplexity. What it measures, and why the count moves week to week.
Read the guide → Engine guidePerplexity often cites a help-center page or a comparison article over a homepage. Which URL types show up most in its Sources panel, and where those links actually point.
Read the guide → Tools comparedMost AI visibility tools report a score or a percentage. Reflexa attaches the full AI answer and its sources to every finding. Which tools show real evidence, and how to read one.
Read the guide → Tools comparedFree checks, self-serve plans from about $49 to $199 a month, and enterprise by quote: the three price bands for AI visibility tools in 2026, and what each one actually buys.
Read the guide → Tools comparedReflexa measures Claude on every plan; Profound only on paid tiers and Otterly.AI only as an add-on; Semrush AI Toolkit skips it entirely. Which tools actually cover Claude, and why the gap matters.
Read the guide → Tools comparedAn agency needs per-domain checks across many client brands and real evidence to show in the meeting, not one blended score. Profound, Peec AI, Otterly.AI, Semrush, Scrunch, Ahrefs and Reflexa, compared for agencies managing multiple accounts.
Read the guide → Tools comparedProfound's Starter plan tracks ChatGPT only at $99/month, and it's not the only tool people compare it against. Peec AI, Otterly.AI, Semrush AI Toolkit, Scrunch AI, Ahrefs Brand Radar and Reflexa, compared on engines, evidence and price.
Read the guide → Fundamentalsrobots.txt is only half the answer: a CDN or bot wall can block GPTBot, OAI-SearchBot, ClaudeBot or PerplexityBot even when the file says they're allowed. 18% of 40 major sites blocked at least one outright.
Read the guide → Search trendsAI Overviews now appear on an estimated 48-60% of Google searches, ChatGPT grew from 800M to 900M weekly users, and AI-Google citation overlap stays at just 8-12%. The trends that matter, each sourced.
Read the guide → AI vs GoogleGoogle ranking predicts clicks on a results page; AI visibility predicts whether ChatGPT, Claude or Perplexity name you before any click happens. Only 8-12% overlap between the two, per Ahrefs.
Read the guide → Engine guideAI reads a product page the same way it reads any other page: can it access it, does it answer the buying question, and does anything outside the shop's own site back up the claim.
Read the guide → FundamentalsGoogle's AI features build on Search foundations, OpenAI's bots read the same crawlable HTML, and Perplexity's Sources panel is a live search. What each engine's own documentation says, and what to keep, add and drop.
Read the guide → GEO fundamentalsGEO, AEO, AI SEO and LLM optimization defined in plain English, what they share, and the one measurement that settles which term actually matters for your business.
Read the guide → AuditFive free checks, in the order that matters: crawler access, rendering, structured data, llms.txt, then whether ChatGPT, Claude and Perplexity actually name you.
Read the guide → ExamplesChatGPT, Claude and Perplexity cited Asana, monday and ClickUp's own pages, plus two independent comparison sites, in September 2026. Seven real examples, and the one thing each gets right.
Read the guide → StrategyGEO is the part of SEO that wins the AI answer. One site, one plan: the four additions GEO makes on top of the SEO you already run, with a free check for each.
Read the guide → AccuracyA wrong AI price usually traces to stale training data or a JSON-LD Offer that disagrees with your page. Why it happens, and the fix, step by step.
Read the guide → AI trafficEvery GPTBot, ClaudeBot or PerplexityBot fetch already sits in your server logs. How to find it, verify it isn't a spoofed user-agent, and what the Reflexa agent on the site shows automatically.
Read the guide → MeasurementAI visibility breaks into three counts per engine, not one score. How to compute your own share with a handful of buyer questions, and how the Reflexa AI Visibility Index does it at 900 observed answers.
Read the guide → Audit guideCrawler access, page content, brand identity and external sources — what an audit checks at each step, and why a single score is a red flag.
Read the guide → Engine guideG2 just absorbed Capterra, Software Advice and GetApp from Gartner. What that consolidation means for whether ChatGPT, Claude and Perplexity cite your reviews, and what to fix this week.
Read the guide → Engine guideOpenAI has licensed Reddit's content since May 2024, and Reddit is the most-cited domain across major AI engines. When a forum thread outranks your own page, and what to do about it.
Read the guide → Engine guideChatGPT, Claude and Perplexity each run a different crawl-to-citation path. What weekly tracking actually shows, engine by engine, and the honest range to expect.
Read the guide → Engine guideGoogle says AI Overviews and AI Mode run on the same indexing and quality signals as regular Search. The one extra step is Google-Extended access, and how to check it's working.
Read the guide → Engine guidePerplexity rebuilds its Sources panel from a live web search on every question. What it actually weighs before citing a page: directness, freshness, structure and corroboration.
Read the guide → GEO fundamentalsBuyers now ask ChatGPT and Perplexity "best accounting firm in Stockholm" the way they used to Google it. What a small site can fix in a week, and what to ignore.
Read the guide → GEO fundamentalsOpenAI, Anthropic and Perplexity each publish their own llms.txt for developer docs. Whether any of them read yours for AI citations is a different question, answered engine by engine.
Read the guide → GEO fundamentalsTraining bots and live-answer bots are not the same thing. What blocking GPTBot, OAI-SearchBot, ClaudeBot, Google-Extended or PerplexityBot actually costs, engine by engine.
Read the guide → GEO fundamentalsFive specific reasons, in order: blocked crawler access, JavaScript-only text, entity ambiguity, no third-party evidence, thin comparison content. Each one with a free check.
Read the guide → AI enginesSearch Console shows AI Mode impressions, never which companies it names. What's comparable today against ChatGPT, Claude and Perplexity, and why one engine is never the full picture.
Read the guide → Search ConsoleAn AI impression in Search Console only means Google pulled a snippet from your page. It says nothing about ChatGPT, Claude or Perplexity, and the Index shows how big that gap can get.
Read the guide → Search ConsoleRolled out to every site on August 31, 2026: AI Overviews and AI Mode impressions, no clicks, no queries, Google only. Why it still can't tell you if AI recommends you.
Read the guide → GEO fundamentalsHow to check in 3 minutes: ask it yourself with five buyer questions, read the sources it cites, or run the free check — and why one answer isn't a measurement.
Read the guide → Buyer countsA way to count it: for every eligible buyer question, how many answers named a competitor instead of you — a worked example, and what the number honestly can't tell you.
Read the guide → OnboardingVerdict, the four groups, the rows the agent can switch on — what each part of the page check means, the order to read them in, and what to check first after a fix lands.
Read the guide → ProductWhat to Fix became Site checks: a finding with its evidence, its fix, and the switch the agent flips — no blended score, just what's broken and what to change.
Read the guide → Search trendsChatGPT has ~900M weekly users, zero-click searches hit ~68%, and only 8–12% of AI-cited links also rank in Google's top 10. The trend data on how AI search is displacing traditional search, side by side.
Read the guide → AI vs GoogleAhrefs found only 8–12% of links AI engines cite also appear in Google's top 10 for the same prompt. Why the two rankings diverge, and what it means for how you measure visibility.
Read the guide → Shopping data77% of consumers have used AI to help with a shopping decision in the past six months, and 64% use AI-assisted tools regularly — but most still won't let AI spend their money. What the surveys show.
Read the guide → Tools comparedReflexa, Profound, Peec AI, Semrush AI Toolkit, Ahrefs Brand Radar, Otterly, AthenaHQ, Scrunch, HubSpot AEO Grader and Frase — compared on engines covered, pricing, and the catch with each.
Read the comparison → Search trends~68% of Google searches now end without a click, AI Overviews push click-through lower still, and AI referral traffic is growing fast from a small base. What it means for how you read your own numbers.
Read the guide → Entity trustWikipedia as training text, Wikidata as structured entity data — what each actually contributes to an AI answer, and what's realistic for a company that isn't a household name.
Read the guide → Competitor trackingWhich specific competitors ChatGPT, Claude and Perplexity name instead of you, how often, in what order, and from which sources — tracked like a metric, not a one-off spot check.
Read the guide → MonitoringA single check is a snapshot, not a trend line — why AI answers shift between checks, and the cadence that catches drift while it's still cheap to fix.
Read the guide → Competitive analysisThe classic marketing metric, adapted to a channel with no fixed leaderboard — what AI share of voice means, and how to calculate it across ChatGPT, Claude and Perplexity.
Read the guide → AccuracyWhy ChatGPT, Claude and Perplexity sometimes state wrong facts about your brand with total confidence, why it happens to every engine, and the concrete steps to correct it.
Read the guide → Reference guideSix AI user-agents, who sends each one, what it feeds, and the exact robots.txt rules that let every engine in — GPTBot, OAI-SearchBot, ClaudeBot, Google-Extended, PerplexityBot and more.
Read the guide → ComparisonThree engines, three mechanics for deciding who gets named — a shortlist, an inline citation, a visible Sources panel — plus the four levers that control all three anyway.
Read the comparison → Engine guideClaude cites live web search results, and Anthropic runs three separate bots — ClaudeBot, Claude-User, Claude-SearchBot — that all say they honor robots.txt. What that means for getting recommended.
Read the guide → Engine guidePerplexity's citations are visible and countable. PerplexityBot vs Perplexity-User, the robots.txt controversy, and what actually gets a page quoted.
Read the guide → Myth checkGoogle says it doesn't use it. Crawler logs show ChatGPT and Claude's bots barely request it. What llms.txt actually is, and what to do instead.
Read the myth check → GlossaryEntity, corroboration, AI crawler, answer-first writing, Buyer Question, Mentions vs Recommendations vs Citations — the terms used across this site, defined in one paragraph each.
Read the glossary → FundamentalsAI engines recommend entities, not pages — a great page attached to an ambiguous identity still loses. Consistent naming, a real About page and linked sameAs profiles, explained.
Read the guide → Content structureAI engines quote pages that answer directly, not ones that circle the topic first. The structure that gets a page extracted and cited.
Read the guide → Myth checkNo schema markup flips a switch that gets you recommended. What structured data actually does, per Google's own guidance, and what's still worth marking up.
Read the myth check → ChecklistFour groups, twelve steps, in the order that actually matters — crawler access, entity clarity, buying-intent content, corroboration — the same sequence the Reflexa page check checks in.
Read the checklist → ProductOne agent, not five tools: the page check, the switches, the words it writes, AI traffic and the conversation. What each answers, what you see, and why it matters.
Read the tour → Data roundupThe key numbers, each sourced: ChatGPT and AI Overviews reach, how buyers use AI to shop, zero-click search, AI-vs-Google overlap, and how many top sites block AI crawlers.
Read the stats → Original researchWe checked the robots.txt of 40 well-known companies. Nearly one in five blocks a major AI crawler — and among news publishers it's the majority. Full method and data.
Read the study → FundamentalsWhat brand intelligence means, the fourth input AI just added, and the five things it now measures — recognition, description, share of voice, sources and accuracy.
Read the guide → MethodPeople ask AI in full, first-person questions, not the keywords they type into Google. Why business-specific prompts measure your visibility better than keyword-derived ones.
Read the guide → Practical guideChatGPT recommends businesses it can access, read and verify. Five steps to become one — each with a check you can run today.
Read the guide → StrategyWhat genuinely changed with AI search, which SEO work still counts, and what to stop doing. With the side-by-side table.
Read the guide → FundamentalsThe practice of getting named in AI answers: how it works, the five signals that matter, and the myths the engines' own docs debunk.
Read the guide →See what AI actually thinks of your business — free check, 3 minutes, evidence included.
Reflexa Agent, the product built on this: reflexaagent.com · how it works · reflexaagent.se