Free website checker: SEO, AEO and GEO in one report
Paste your address. In about 30 seconds you get three scores, every test graded, the evidence, and how to fix each one. No login, no card.
Free. No account, no card. 3 checks a day. The three scores show straight away; the full report of every graded test asks for an email. We fetch your page like a search engine would and never store passwords or personal data from it.
What do SEO, AEO and GEO mean?
They are three different ways of being found, and a page can be excellent at one and invisible in the other two. SEO is the blue links: can Google crawl this page, understand it, and rank it. That is the oldest of the three and still the largest share of most sites' traffic, which is why 38 of our 60 tests sit there.
AEO is being the answer rather than a result. A featured snippet, a People-also-ask box or a voice assistant lifts one paragraph and reads it out. Pages that win those are written for it: a direct answer of roughly 40 to 60 words near the top, headings phrased as the question a person would ask, and facts in lists or tables that a snippet can pick up whole.
GEO is being cited inside an AI answer. That requires two things in order: the AI vendors' crawlers must be allowed in, and once in, the page has to be quotable. We check 14 named AI crawlers, look for llms.txt, and count how many of your paragraphs are self-contained enough to be lifted with a number in them.
| Score | The question it answers | Tests | Heaviest single test |
|---|---|---|---|
| SEO | Can Google find, read and rank this page? | 38 | Mobile speed (weight 6) |
| AEO | Will an assistant read this page out as the answer? | 11 | Question headings (weight 10) |
| GEO | Will an AI answer cite this page by name? | 11 | Structured data depth (weight 10) |
How is a score actually calculated?
Every test carries a published weight. A pass earns the whole weight, a warning earns half, a failure earns nothing, and the score is what you earned divided by what was available, rounded to a whole number out of 100. There are 60 tests in a full report. Nothing is averaged from a black box, so the same page checked twice gets the same number twice.
A test that cannot run does not quietly cost you points. If the speed provider is unreachable, that row is marked not-available and its weight leaves the denominator entirely, so a provider outage cannot make your site look worse than it is. Rows that need live rank data or a paid API are marked coming, and they are excluded the same way rather than sitting at zero.
Weights are not equal because the fixes are not equal. A missing title tag is worth 8 and takes two minutes. Question headings are worth 10 on the AEO side because they are the single strongest signal that a page was written to be quoted. Minifying CSS is worth 1, because in 2026 it almost never changes an outcome on its own.
| Score | The question it answers | Tests | Heaviest single test |
|---|---|---|---|
| SEO | Can Google find, read and rank this page? | 38 | Mobile speed (weight 6) |
| AEO | Will an assistant read this page out as the answer? | 11 | Question headings (weight 10) |
| GEO | Will an AI answer cite this page by name? | 11 | Structured data depth (weight 10) |
What should I fix first?
Work down by weight, not down the page. The report sorts the worst rows to the top for exactly this reason: three fixes worth 10 each move a score more than fifteen fixes worth 1. Most sites we check have between 4 and 8 failing rows, and two or three of them account for most of the lost points.
Read the evidence line before you change anything. Each failing row prints what we actually saw, such as the byte count of the HTML, the exact headings we found, or the crawler that returned a 403. If the evidence does not match your page, you are looking at a cached copy or a different URL, and the fix is to check the address rather than to edit the site.
- Fix the heaviest failing row first: the report already puts it at the top.
- Re-check after each change. Every check is kept, so the two runs sit side by side.
- Ignore any row marked coming: it is excluded from the score, not counted against you.
- Treat warnings as half-marks, because that is exactly what they are worth.
What are the limits of this check?
This is one fetch of one page, not a crawl of a site. It reads the HTML the server returns, the response headers, the robots file, the sitemap and the structured data, and it asks a speed provider for the mobile performance of that same address. It does not log in, it does not run your checkout, and it does not see pages that are not linked from the one you gave us.
The score is an estimate against a published rule set, not a prediction of where you will rank. Ranking depends on competitors, links and history that no single-page check can see. What this is good for is the opposite direction: it tells you, with evidence, which of the things under your control are currently wrong, and roughly how much each one is worth.
The free check runs 3 times a day per visitor and keeps every run from the first one, so a site you check today can be compared with the same site in a month. There is no login and no card. An account exists for people who want the same 60 tests run for them on a schedule and kept in one dashboard.
Where do these rules come from?
The SEO rows follow documented behaviour rather than folklore: title and description handling, canonicals, robots directives, sitemaps and mobile-first indexing are all described by Google in public. The speed row uses Google's own PageSpeed Insights, which is the same Lighthouse engine that reports Core Web Vitals, so the number you see here is the number Google's tool reports.
The structured-data rows validate against schema.org types, and the AI-crawler rows check the user agents each vendor publishes for their own bots. Where a rule is genuinely contested, such as how much llms.txt matters today, the row carries a smaller weight and says plainly what it is testing, instead of pretending the industry has settled it.
- Google Search Central, the documentation for how the crawler and index behave.
- PageSpeed Insights and Lighthouse, for the mobile performance score.
- Schema.org, for the structured-data vocabulary.
Questions people ask
Is this really free?
Yes. Three checks a day per visitor, no login and no card. Creating an account adds tracking over time and a fix plan.
What do SEO, AEO and GEO mean?
SEO is being found in Google’s blue links. AEO is being the direct answer: featured snippets, People-also-ask and voice assistants. GEO is being named and cited inside AI answers from ChatGPT, Google AI Overviews, Perplexity and Gemini.
How is the score calculated?
Each test has a published weight. A pass earns the weight, a warning half, a failure none. Rows that need live rank or AI data are marked “coming” and are not counted.
What does this mean for a business like mine?
The same 60 tests run on every site, but what matters most is not the same for everyone. A plumber lives or dies by the map. A one-person shop needs to be quotable. A company with a thousand pages needs the right one to win. Pick the closest.
Local business
A plumber, a dentist, a removals company, a salon. Customers can phone you or come to you.
Wants: To be the result in the map pack, and the business a phone reads out when someone asks for one nearby.
Online only
A one-person shop, a maker selling downloads, a creator brand, a blog. No shopfront, and perhaps no phone.
Wants: To be found for what you sell, and quoted when someone asks an AI what to buy or how to do the thing you write about.
Corporate
Multiple locations or countries, a marketing team, legal and compliance pages, careers, investors.
Wants: The right page of many to win the query, and the brand described correctly when an AI is asked about it.
Local business
Why does my score care so much about my phone number?
Because a voice assistant asked for "a plumber near me" has to say something out loud, and what it says is a name, a place and a number. If those are not on your page as text, with the number as a tappable link, and matching what your structured data claims, the assistant has nothing to read. For a local business this is the single heaviest thing on the answer side of the report.
I work from home and do not want my address public. What now?
Publish the area you serve rather than the house you live in, and keep the phone number and business name consistent everywhere. A service-area business is a normal thing and the schema vocabulary has a way to say it. What hurts you is not the missing street, it is a name or number that differs between your site, your Google profile and your listings.
Does this check my Google Business Profile?
Not yet. Today it checks what your own website says about who and where you are, which is what an answer engine reads first. The profile itself, review counts and map-pack position are on the roadmap and are marked "coming" in the report rather than counted against you.
My competitor ranks above me and their site is worse. Why?
Almost always links, reviews and how long they have been established, none of which a single-page check can see. This tool tells you which of the things you control are wrong. That is a smaller list than the whole of ranking, but it is the part you can fix this week.
Online only
I have no address or phone. Am I being punished for that?
Today, partly, and we think that is our bug rather than yours. Two tests currently want a street address and a phone before they will pass, and they carry real weight. A shop that only exists online was never meant to have either. We are changing those tests to accept the thing they are actually measuring: whether a machine can work out who you are and reach you. A consistent name, a proper Organization declaration and any reachable contact channel should count as much as a shopfront does for a plumber.
What actually gets a small shop quoted by an AI?
Three things in order. Let the AI crawlers in, because a robots file written years ago often blocks them by accident. Then give them paragraphs worth quoting: forty to a hundred and twenty words, self-contained, with a real fact or number in them. Then say who you are in a way a machine can read, so there is an entity to credit. Most small sites fail at the first step and never learn it.
Is a blog judged the same as a shop?
On this report, yes, and that is a limitation we are open about. The tests measure whether a page can be found, understood and quoted, which matters for both. What differs is what you would do with the result. A shop should put its prices and shipping in a table. A blog should put its updated date where a reader and a crawler can both see it.
I sell on a marketplace, not my own site. Is this useful?
Only partly. Check your own site if you have one, because that is the page you control and the one an AI can cite. A marketplace listing is graded by the marketplace, not by you, and no amount of work on your own page changes it.
Corporate
We have thousands of pages. What use is checking one?
Check the pages that carry money and reputation: the home page, the top product or service pages, the one page you would be embarrassed to lose. A single-page check finds the template-level faults that repeat across the whole estate, because most large sites express the same head, the same schema and the same navigation on every page. Fix the template and you have fixed thousands of pages.
Which page should we check first?
The one your biggest campaign points at. If its title, description, canonical and structured data are wrong, everything you spend driving traffic to it is working against a handicap you did not know you had.
Our brand gets described wrongly by AI assistants. Does this help?
It addresses the part you control. An assistant describes a brand from what it can read and attribute. If your pages do not declare the organisation in a machine-readable way, do not link the profiles that confirm it, and do not carry quotable statements of fact, the assistant fills the gap from whatever else it found, which is often a forum post or a competitor. The report shows you which of those pieces is missing.
Who signs this off internally?
Most of the failing rows are template changes owned by whoever runs the site, not by marketing. The report separates them into what you can change today with no developer, what needs a developer or an hour, and what is ongoing content work, so the list can be split between teams without a meeting first.
You have the score. Now what?
The free check is a photograph. It tells you where you stand this morning, all 60 tests, with the evidence and a plain fix for each one. Work down from the heaviest failing row and you will move the number yourself, for nothing, and you never have to come back here.
What it cannot do is watch. It sees one page, once, and it cannot tell you whether the fix you made on Tuesday actually worked, whether a deploy quietly undid it in March, or where you sit against the three competitors you actually worry about. That takes the same tests run for you on a schedule and kept side by side, which is what an account is for.
- Every check kept from the first one, so the line shows whether the work paid off.
- The fix plan split into what you can do today, what needs a developer, and what is ongoing.
- A short report on Monday, Wednesday and Saturday: three scores, what changed, the next three steps.
Which checker do you need?
- Free SEO checker — Free SEO checker with no signup: title, description, headings, schema, robots.
- Free AEO checker: are you the answer? — Free Answer Engine Optimization checker: direct-answer paragraph, question headings, FAQ schema, readability, entity clarity and voice-ready NAP, graded with fixes.
- Free GEO readiness checker: can AI cite you? — Free Generative Engine Optimization check: AI crawler access for 14 bots, llms.
- llms.txt — the short guide we publish for AI crawlers.
- llms-full.txt — the same thing in full, all 60 tests listed.
- Sitemap — every public page.
Outside references this checker follows: Google Search Central documentation, PageSpeed Insights, Schema.org.
This checker reads one page as a search engine would and grades it against a rule set we publish and can change. It does not crawl your whole site, log in, or see anything that is not linked from the address you gave it. The scores are estimates, not advice and not a guarantee of any ranking, traffic or revenue, and a clean result is not proof that a site is well built. Check anything important yourself before acting on it. Use of the checker is subject to the terms of use.
Updated . 60 tests, weights published above. Free tier: 3 checks per visitor per day.