12 free AEO graders and checkers compared: what HubSpot, Mangools, and the rest actually test

Kite's free AI visibility checker asks ChatGPT, Gemini, Perplexity, and Claude the questions your customers ask, then shows whether they recommend you and who they recommend instead. You type in your website address and nothing else. It's free, needs no account, and takes about 45 seconds.

The questions are what set it apart. Kite reads your homepage, works out what you sell and where, and writes three questions a buyer would ask an AI assistant. It leaves your business name out of them, because a buyer who doesn't know you yet won't type it. That makes the result a test of whether you get recommended, not a test of whether a model recognizes your name. Kite then asks each question in all four assistants with web search on. For ChatGPT and Gemini, it reads the answer a signed-out visitor sees in the app, which is closer to what a new buyer sees than a developer version of the model.

Kite only counts an answer when it recommends you or presents you as an option. A passing mention, or a line advising against you, doesn't count. The result is a plain count rather than an abstract score. In the worked example on the page, a check of Appsmith (the company that makes Kite) on October 15, 2026 came back "Recommended in 6 of 12 answers." Every assistant recommended Appsmith at least once, none did for the question about security and self-hosting, and Retool and Microsoft Power Apps were recommended most often instead. The result also lists three changes most likely to get you named, such as publishing a security and deployment guide on your own site.

Three questions is a small sample, and Kite says so on the page: treat one check as a snapshot. The free check is the start of the work rather than the end of it. After you sign up, Kite tracks the questions that matter to your business every week and tells you when you start getting named or when a competitor does. It also does the work that changes the answer. It writes the pages that answer your customers' questions directly, finds the directories and articles the assistants already cite and prepares your listing requests and pitches, then runs the questions again to see what changed (how Kite improves your visibility in AI answers). On a site Kite hosts, it publishes the pages itself once you approve them. Kite has a free plan, and its Growth plan is $100 a month.

LLM Pulse's free grader runs live prompts across five AI engines and emails you a score out of 100 with a list of fixes (LLM Pulse). You give it your domain and an email address. Submitting the form also subscribes you to the LLM Pulse newsletter, and the report reaches your inbox in about two minutes.

It picks the prompts itself from your domain, running what it calls "a representative sample of category prompts" through Google AI Overviews, ChatGPT, Perplexity, Google AI Mode, and Gemini. That's the widest engine list among the live checkers here, and it includes Google's two AI surfaces. The score covers four areas: how often engines lift your content into the answer, whether you're cited as a main source or one of many, whether your pages have answer-ready structure, and where competitors win prompts you lose. The report shows which prompts you win and lose and gives a list of fixes ranked by impact and effort.

The page doesn't say how many prompts it runs or how the four areas are weighted, so you can't tell how big the sample behind your score is. The next step it points to is a 14-day trial of LLM Pulse's paid tracker, with plans from EUR 49 a month.

Sprites' free grader sends a live query to ChatGPT, Gemini, and Perplexity and returns a letter grade from A to F, with the three fixes it expects to move the grade most (Sprites). You type in your brand name only. A run takes 20 to 60 seconds, and you get three free grades an hour with no account.

The grade is built from five weighted parts read from that run's answers: how well the engines know you (35 percent), how much they trust their sources about you (20 percent), how many of the three engines know you at all (15 percent), sentiment (15 percent), and your awareness against the strongest rival the engines named (15 percent). Publishing those weights is useful, because you can see exactly why you got a C. The grade and fix list show on the page. One email address unlocks what each engine said word for word, the competitors it named, and the sources it cited.

Because you type in a brand name, Sprites is grading how the engines know and describe you, and its own definition says it asks the engines "about your brand." The page doesn't say what questions it sends. Sprites sells an AI ads agent with an AEO agent included from its Launch plan, and it has no free plan; paid plans start at $66 a month billed yearly (Sprites pricing).

AEO Checker gives you one free preview report that checks your website in two AI models, ChatGPT and Gemini (AEO Checker pricing). You enter your website, and the preview shows the kinds of results its paid reports contain.

Its sample report shows a score for each engine, a set of key prompts with the share of runs that mentioned the brand, a competitor landscape, and a review of how answerable your pages are, how present you are across the web, your structured data, and whether AI crawlers can reach the site (AEO Checker). Reporting a mention rate per prompt is the right instinct, since a share of runs is steadier than a single answer.

The free tier is a single preview, and the pricing page labels the paid plans' models as searching the web without saying the same for the free preview's two models. After that, unlimited one-time reports cost $8 a month billed yearly, and daily tracking of one brand with 25 prompts costs $22.50 a month billed yearly, with a 7-day trial.

HubSpot's AI Search Grader, which many people still search for as the HubSpot AEO Grader, scores how ChatGPT, Perplexity, and Gemini describe your brand (HubSpot). You fill in four fields: company name, location, product or services, and industry. HubSpot says the tool "sends your brand information to ChatGPT, Perplexity, and Gemini and asks each model how it characterizes your brand based on its training data." It's the best-known free grader, and it's free with no HubSpot subscription.

The score is out of 100 and weighted toward perception. Sentiment is worth up to 40 points, presence quality and brand recognition up to 20 each, and share of voice and market competition up to 10 each. HubSpot explains the weighting plainly: sentiment matters most "because it measures how AI characterizes your brand, not just whether it recognizes you." Scores show on screen, but the full report sits behind a form. Salesdorado, a sales-tools review site, found that the form asks for an email and includes a checkbox to be contacted by a sales representative (Salesdorado).

The report is more interesting than the headline number. In Salesdorado's run on the CRM brand noCRM, the grader returned a separate score for each engine: 53 for OpenAI, 63 for Perplexity, and 67 for Gemini. Gemini gave noCRM its best sentiment and its lowest share of mentions, and OpenAI did the reverse, which point to opposite fixes. The PDF's fine print listed the models and how current their knowledge was. OpenAI's GPT-5.4 mini knew nothing after August 31, 2025, Gemini 3 Flash Preview nothing after January 2025, and only Perplexity searched the live web. HubSpot's own product page agrees that Perplexity uses live search (HubSpot AEO). Nena Jasar saw the same split when she ran it on her own review site: ChatGPT counted 42 mentions of her brand and Perplexity counted 1,420 (Nenawow). Salesdorado rated the Perplexity source list, which scores each page behind the assessment, the most useful part of the report.

That's the shape of HubSpot's grader: a thoughtful read on how you're perceived, partly from models that are a year or more out of date. A company that relaunched in 2026 can look weak on two of the three engines for reasons that have nothing to do with today's answers. HubSpot's pages also don't say which questions the grader asks or how many. HubSpot's paid next step is HubSpot AEO at $50 a month, or $45 a month paid yearly, which tracks 25 prompts of your choosing across the same three engines and gives prioritized recommendations.

Mangools' AI Search Grader generates prompts from your brand and niche, runs them through several AI models, and reports how often you appear and in what position (Mangools). You type in a brand name and a short description of your niche, or let it suggest one. The result arrives in under 60 seconds with no account.

The models are the thing to notice. Free use covers three: GPT-5 Nano, Mistral Small 3.2, and Llama 3.3 70B. A free Mangools account adds Claude Haiku 4.5, Gemini 2.5 Flash Lite, and DeepSeek V3. These are small developer versions of each model, listed on the page by their names on OpenRouter, a service for calling models directly. They aren't the consumer ChatGPT or Gemini apps your buyers use, and the page doesn't say they search the web. The checker does cover models no other free tool here touches, which matters if your buyers are developers who use Llama or DeepSeek.

The AI Search Score runs from 0 to 100. It combines the share of prompts where you appear in the top 20 with your average position, adjusted for each model's market share. Position is the weak part. Fishkin's research found that brand lists rarely repeat in the same order, and he wrote that "any tool that gives a 'ranking position in AI' is full of baloney" (Search Engine Journal). Mangools was also the tool that gave Nike a perfect 100 in SUSO's test. After the score, Mangools lists steps for you to take and offers its free AI Search Watcher, which sends weekly reports. Its paid SEO bundles, which include the paid version of AI Search Watcher, start at about $452 a year (Mangools pricing).

AEO Grader's free check is an estimate built from one Gemini call with live Google Search, not a check of all four engines it names (aeograder.org). You give it your website address and brand name. It runs in about a minute with no signup.

The maker is upfront about this. Its how-it-works page says the free check reads what the web says about your brand through Gemini and then estimates how ChatGPT, Claude, Gemini, and Perplexity are likely to describe you. It adds that "these are labelled as estimates: the free check does not query those engines" (How AEO Grader Works). The score is weighted across authority (30 percent), clarity (25 percent), technical readiness (25 percent), and sentiment (20 percent). The free result comes with an action plan and code snippets for structured data you can paste into your site.

The real multi-engine check costs $10 once. It asks ChatGPT, Claude, Gemini, and Perplexity a panel of buyer questions and shows each engine's actual answer with sources, plus one re-scan within 14 days to measure change. Monthly tracking is $9 a month. For a free tool, it's a reasonable reading of your web reputation, but treat its per-engine results as the guesses it says they are.

Ahrefs' free checker shows how often ChatGPT, Gemini, Perplexity, Copilot, Google AI Overviews, and Google AI Mode mention your brand (Ahrefs). You type in a brand name only, written the way your audience would say it. There's no signup and no limit on how many brands you check, including competitors'.

The prompts come from real Google searches rather than invented questions, which is a genuine strength. The free report gives total mentions, mentions by platform, and the top five topics, cited domains, and cited pages. It separates a mention (the assistant named you) from a citation (it linked to your site). There's no score and no fix list. Ahrefs calls the report "a limited preview" and caps each section at five results.

The checker is labeled "Powered by Ahrefs Brand Radar," and Ahrefs' methodology post explains how Brand Radar works. Ahrefs runs millions of questions through the public interfaces of each assistant, stores the responses, and refreshes the chatbot answers monthly on a 90-day window (Ahrefs methodology). Ahrefs also says the data works best for established brands with real search demand and may have limited coverage for brands with little search volume (Ahrefs Brand Radar). Brand Radar's custom prompt tracking starts at $50 a month, and its full index starts at $199 a month.

Semrush's free tool gives your domain an AI Visibility Score from 0 to 100, comparing how often you're mentioned with others in your industry across ChatGPT, Gemini, Google AI Mode, and Google AI Overviews (Semrush). You enter a domain, and you can run it three times a day without an account.

It's the only free tool here that adds an estimate of audience reach. The report also covers mentions, citations, coverage by platform, your most-cited pages, your best-performing topics, and a comparison with your closest competitors. Semrush's help center explains that the score compares your mentions with the median for your top competitors, which Semrush identifies automatically (Semrush Knowledge Base).

The data comes from what Semrush calls "a database of 26M+ prompts and responses," updated daily. It's a stored collection, not a fresh ask, so what you see depends on whether your topics are in it. The page lists six things you can do to improve, such as earning mentions on review sites and keeping your positioning consistent on LinkedIn, G2, and Capterra. The tool leaves those to you. The paid AI Visibility Toolkit costs $99 a month.

CheckAEO scores a single page from 0 to 100 on how well AI answer engines can read, understand, and cite it (CheckAEO). You can give it a URL, pasted HTML, or plain text, and the score arrives in about ten seconds with no account.

It weights six areas: direct answer quality (25 percent), content structure (20 percent), schema markup (20 percent), technical signals (15 percent), entity and topic coverage (10 percent), and freshness (10 percent). Scoring adjusts to the type of page, so a pricing page isn't marked down for lacking an author byline. One feature stands out among the scanners: an AI comprehension test, where Claude reads your page and answers seven questions a customer would ask, such as what you do and what it costs, quoting your page for each answer. A two-answer preview is free, and an email unlocks the full test and report.

CheckAEO is honest about its limit: "No tool can promise citations, and any that does is overselling." It measures the part you control. Paid options are a $197 commissioned analysis and a $10,000 done-for-you sprint.

Madgicx's free AEO checker scores one page against twelve checks with published point values, and it shows which AI crawlers your site lets in (Madgicx). You enter any public URL, yours or a competitor's, or paste a draft before you publish it. It's free with no signup and no email.

Its transparency is the best among the scanners. Every check has a fixed weight on the page, from answer-first openings at 13 points to entity markup at 3. The crawler check is useful on its own: it sorts 14 AI crawlers into those that find pages for answers, those that fetch a page when a user asks, and those that collect training data. Blocking the training group doesn't remove you from AI answers, so you can opt out of training and still be cited. Madgicx states plainly that the tool doesn't ask any AI engine, measure backlinks, or see content that loads only through JavaScript, and that "the same page gets the same score each time."

Madgicx sells advertising software for Meta ads, and the checker is a free tool rather than the front door to an AEO product. That makes it a clean fix list with no upsell attached.

YOAEO scans a page from an AI crawler's point of view and returns three scores, one each for AEO, GEO, and traditional SEO, in about 30 seconds (YOAEO). You paste a URL, with no signup or install.

It runs 18 checks with a green, yellow, or red light for each: eight for AEO, such as direct-answer paragraphs and FAQ markup, six for GEO, such as crawler access and quotable content, and four for SEO basics. It fetches only four public files: the page HTML, robots.txt, sitemap.xml, and llms.txt. Because many AI crawlers don't run JavaScript, the report flags content that only appears after scripts run. Its sample report's top issues are typical of what these scanners catch: missing FAQ structured data, AI crawlers blocked by robots.txt, and paragraphs too long to quote.

The free result gives the scores, a grade, an overview of issues, and one full sample diagnosis. The full report with every fix costs NT$490, and its maker, the Taiwanese agency KS-DIGI, offers to implement the fixes. YOAEO doesn't ask any assistant a question.