Free SEO Scanner: What It Can Tell You, and Exactly Where It Stops

You shipped a product, nobody is finding it, and you want a free SEO scanner to tell you why. A free scanner can do that for the part of the problem a crawler actually meets: whether your pages can be fetched, read, indexed and trusted, and whether your links, headers, forms and load times hold up. It cannot tell you much about things that cost real money to measure, such as who links to you, how your pages render across devices, or whether ChatGPT and Perplexity mention you when someone asks. That split matters because, as Michigan Tech's guide to what SEO is puts it, "Search engine crawlers (also called bots) comb through all the content they can find on the internet. They do so by following internal links within websites and links across different websites." A free scan checks the path the crawler walks. It does not check what happens after the crawler leaves.

This guide covers:

• What SEO is, in the one sentence a scan needs
• What a free scan reads cheaply, and what is expensive to measure
• A step-by-step run of a free scan, with the places each step usually fails
• How to tell a check backed by evidence from a generic score
• A worked example, carried from first scan to fix list
• The condition that decides whether you need a paid audit at all
• A triage template you can reuse on every scan

What is SEO, and which part of it can a scanner check?

The same Michigan Tech guide defines it cleanly: "Search engine optimization (SEO) is the practice of creating and optimizing relevant, authoritative content in a way that helps visitors (and search engine crawlers) find answers to their key questions." That definition has two halves. The first half is the content itself: is it relevant, is it authoritative, does it answer the question. The second half is the machinery that lets a crawler find it and a visitor use it: status codes, tags, sitemaps, links, speed, security.

A scanner lives almost entirely in the second half. It can prove a crawler reaches your page and can read what is on it. It cannot judge whether what is on it is the best answer anyone could give. Keep that boundary in mind for everything below: a clean scan removes the reasons a good page gets missed, and says nothing about whether the page is good.

What can a free SEO scanner actually read?

Before running anything, know the cost structure behind the result, because it explains why free tools differ so much. Some checks cost a scanner almost nothing: one HTTP request returns your HTML and headers, and a parser can grade dozens of things from that single response. Other checks need a real browser, a crawl of the wider web, or paid calls to answer engines, and those are where free tiers tend to stop.

Read the table as a rule of thumb for any tool you try. If a free scanner claims to cover a high-cost row in full, ask what it actually did to measure it. A backlink count with no list of linking pages, or an "AI visibility score" with no record of which questions were asked to which engine, is a number without evidence.

The LaunchScaler readiness scan is drawn along this line. The free scan runs 156 checks across 6 of its 7 categories (Search, Backlinks, Security, Compliance, Speed & visual, and Does it work), with no account and no card. The seventh category, AI visibility, is the one the free scan leaves out. The full audit adds 40 more checks for 196 in total, and it is where the expensive rows open up: the deeper rendered-UX findings, the backlink detail, and the AI visibility measurement, which puts 10 questions to each of 7 answer engines (ChatGPT, Perplexity, Claude, Google AI Overviews, Google AI Mode, Gemini and Copilot) for 70 reads, and measures your share of the answer. None of those three are part of the free scan.

How to run a free SEO scan and get a result you can act on

The task: scan one live site you own, and leave with a ranked list of fixes rather than a score. This applies to a launched or nearly launched site on its production domain. If your site is still behind a password or on a preview URL, stop at step 1 and read the recovery note there.

1. Scan the production URL, not a preview. Paste the exact address customers type, including whether it is www or the bare domain. Preview deploys often ship with noindex headers or a different canonical, and the scanner will faithfully report problems your real site does not have. Where it fails: the scan returns a blocked or login page. Recovery: a scanner reads what any visitor or crawler meets, so if the page is gated, the report grades the gate. Unlock the page or scan the public marketing page instead.
2. Confirm which host the scanner actually graded. Look at the final URL in the report after redirects. If yourproduct.com redirects to www.yourproduct.com which redirects again to /en/, the scanner graded the last hop, and your canonical and sitemap should all agree on it. Where it fails: the report shows a redirect chain of three or more hops, or a loop between http and https. Recovery: collapse it to one permanent redirect from every variant to the single canonical host, then rescan.
3. Read the categories before the overall number. An overall score blends unrelated things. A site can look healthy overall while its one failing check, a noindex on the homepage, blocks everything. Open each category and note the lowest-scoring checks. In a readiness scan every check scores out of 100 and shows the evidence behind its verdict, so sort by the check, not the headline.
4. Fix access blockers first. Anything that stops a crawler reaching or indexing the page outranks everything else: noindex in a meta tag or X-Robots-Tag header, a Disallow: / left in robots.txt, a canonical pointing at another domain, a 4xx or 5xx on the homepage. These are the Search checks that decide whether the rest matters. you remove from the HTML but the scan still flags it. Recovery: check the response headers; frameworks and hosting platforms can add at the server level, especially on non-production environments that later get promoted.

How do you tell a real check from a generic score?

This is the skill that separates a useful scan from a decorative one, and it applies to every tool you try. A real check has four parts you can see:

• A named condition. "Homepage returns X-Robots-Tag: noindex" rather than a bare "Indexability" gauge.
• The evidence it read. The header value, the line of HTML, the URL that returned 404. If you cannot see what the scanner saw, you cannot verify it.
• Why it matters. One sentence tying the finding to a reader: a search engine, an answer engine, a visitor.
• A fix you can apply. Specific enough to hand to whoever deploys the site.

A generic score skips the middle two. It gives you a single SEO number and a list of advice that would apply to any site on the internet ("add more content", "build backlinks"). Test it: if the recommendations would read the same on a competitor's report, the scanner did not read your site closely.

Three more signals of a trustworthy scanner:

• A fixed, published bar. The same checks, graded the same way, on every site. If paying changes the score, the score measures your wallet. LaunchScaler's pricing page states that nothing on it changes your readiness score and every site is measured against the same bar, paid or not. Paying changes how deep you see, never how you are scored.
• It reads the site, not a questionnaire. A tool that asks you to describe your site is grading your description. The readiness scan takes a URL and nothing else.
• It admits what it did not measure. A free result that says plainly which findings are full audit only is being more honest than one that fills every gauge.

Worked example: one scan, from URL to fix list

Take a solo founder launching a developer tool on yourproduct.com. It has a marketing homepage, a pricing page, a docs section generated by a static site tool, and a sign-up form. Three weeks after launch, it barely shows up when people search its name.

Step 1: the free scan

The founder pastes yourproduct.com into the readiness scan. No account. The result arrives graded across 6 of the 7 categories, 156 checks, each with its score and evidence. AI visibility is not in it.

Step 2: reading it in order

The founder ignores the headline and opens categories, lowest checks first:

• Search. The docs subdomain carries a canonical tag pointing at the static site tool's default domain, not at docs.yourproduct.com. Evidence: the <link rel="canonical"> value shown in the check. This alone explains why the docs never appear under the product's name: every docs page tells crawlers the real copy lives elsewhere.
• Search. The sitemap lists the homepage and pricing page but no docs pages.
• Does it work. The sign-up form posts to an endpoint that returns an error for visitors without a session. The founder never saw it, because the founder is always logged in.
• Security. No Strict-Transport-Security header.
• Compliance. The privacy policy link in the footer points to a 404.
• Speed & visual and Backlinks. The founder fixes what these checks flag, and is left with two questions: which element slows the small-screen render, and which sites link in and whether they count. The deeper rendered-UX findings and backlink detail that would answer them are full audit only.

Step 3: the fix list, in priority order

1. Set the docs canonical to the docs subdomain's own URLs.
2. Add the docs pages to the sitemap, or link a docs sitemap from the main one.
3. Fix the sign-up endpoint so it accepts logged-out requests; test in a private window.
4. Restore the privacy policy page.
5. Add the HSTS header.

Every item came from a check with visible evidence. None needed money.

Step 4: the rescan

After deploying, the founder rescans. The canonical and sitemap checks now pass. The HSTS check still fails. The cause: the header was added in the app, but the CDN in front of it strips unknown headers on cached responses. The fix moves to the CDN configuration, and the next rescan passes.

Step 5: deciding about the paid layer

What remains: the two open questions about rendering and links, and one the free scan does not touch at all, which is whether answer engines mention the product when a developer asks for a tool like it. Here the condition in the next section decides.

When is a free scan enough, and when do you need the full audit?

Do not decide by budget. Decide by whether you have run out of free, evidenced fixes.

The free scan is enough when it still shows failing checks with evidence you can act on. Work that list down first. Paying for deeper findings while your homepage has a stray noindex buys detail you cannot use yet.

The full audit is the next step when the free checks pass and your remaining questions are the expensive ones:

• Which element makes the page render slowly at which size (rendered UX).
• Who links to you and whether those links carry weight (backlink detail).
• Whether answer engines cite you, and how much of the answer you get (the AI visibility measurement: 10 questions to each of 7 engines, 70 reads).
• The exact fix for every check, not just the verdict.

On LaunchScaler that layer is the full audit at $19, one time, tied to one domain, with no subscription and no expiry. Re-scans keep the deeper findings open, so the unlock is not spent on a single report. It does not change the score: the check that failed on the free scan fails identically on the paid one, you just see more of why.

Monitoring is a separate decision, made later. Sites drift: a redesign drops a tag, a deploy breaks a form, a new page ships without structured data. If you deploy rarely, rescanning by hand after each release does the job. If you ship often, Watch is the separate paid plan that re-audits on a schedule; it buys frequency and delivery, never a different verdict.

Where SEO and AEO split inside one scan

A single pass can measure both, because they share foundations: crawlable pages, fast clean markup, signals worth trusting. They split at the outcome. SEO earns a ranked link someone clicks; AEO (answer engine optimization) earns a mention inside the answer itself. AEO additionally needs passages an engine can lift and quote, machine-readable structure, and pages a browsing agent can move through without breaking. That last part is what the readiness scan calls Agentic Browsing: a factor, not a check, describing whether an AI agent can actually drive your site, reach the content and finish a flow. The free scan covers the shared foundations. Measuring whether engines actually cite you takes real queries, and that sits in the full audit.

What free scanners get wrong, and what no scanner resolves

Being honest about the limits is part of using the tool well.

Scores invite the wrong goal. A number out of 100 tempts you to chase the number. The number is a proxy for "can a crawler reach, read and trust this page". Once the access and understanding checks pass, nudging a check from good to slightly better rarely changes what a crawler does.

A scanner cannot grade your content. It can tell you a page has one heading, a description, valid markup. It cannot tell you the page is the best answer to the question your buyer types. The Michigan Tech guide is blunt about the order: "content is king," and "Second to content are incoming links." A clean scan is a precondition, not a substitute. If your pages pass every check and say nothing a buyer needs, they will still be passed over.

Backlinks are measured, not made. No scan builds links. It can only show whether the ones you have count. Links come from other sites choosing to point at you: a directory listing, a review, a mention in someone's docs. LaunchScaler's own free listing includes one dofollow backlink, but it is earned only by clearing the readiness bar, and a claimed free listing also requires either the LaunchScaler badge on your site or a one-time fee to skip it.

Passing guarantees nothing. A readiness scan is a fixed published bar, measured automatically. Clearing the readiness bar means your site met every check on that published list on the day it was measured. What Google ranks and what an answer engine chooses to quote are decided elsewhere, by systems no scanner controls.

Answer-engine reads move. Ask the same engine the same question on two days and the answer can differ. That is why a single spot check (you asking ChatGPT about your product once) tells you little, and why a measurement that puts a fixed set of questions to several engines is more useful than an anecdote. Even so, treat it as a reading of the day, not a permanent fact.

Template: a scan triage sheet

Copy this into a doc and fill it in on every scan. It turns a report into a queue.

Header

• Site scanned (final URL after redirects):
• Date of scan:
• Deploy or commit it reflects:

Blockers (fix before anything else)

• Check name:
• Evidence the scanner showed (header, tag, URL, status code):
• Fix:
• Owner:
• Verified on rescan (yes / no, date):

Understanding (titles, canonicals, sitemap, internal links, structured data)

• Check name:
• Evidence:
• Fix:
• Verified on rescan:

Trust and working order (Security, Compliance, Does it work)

• Check name:
• Evidence:
• Tested logged out, private window, mobile (yes / no):
• Fix:
• Verified on rescan:

Open questions beyond the free scan (rendered UX, backlink detail, answer engines)

• What the free checks flagged and I fixed:
• The question still open:
• Do I have free, evidenced fixes left? (if yes, stop here and do those first)

Decision

• Free fixes remaining:
• Open questions only a deeper audit answers:
• Next scan date:

Run it on your own site today

Pick the one URL that matters most, usually your homepage, and scan it on its production domain. Open the lowest-scoring checks first, copy each piece of evidence into the triage sheet, and fix the access blockers before you touch anything else. Then rescan and check that the fixes reached production. The free readiness scan takes only the address. Decide on anything deeper only after the free list is empty.

Questions people also ask

Is there a free SEO checker?

Yes. Several scanners run without payment, and the useful ones grade your live pages against named checks with visible evidence. LaunchScaler's readiness scan is free, needs no account or card, and runs 156 checks across 6 of its 7 categories, leaving out AI visibility. Judge any free checker by whether it shows what it read, not by how many gauges it fills.

Can I do SEO myself for free?

Most of the groundwork, yes. Access, indexing, titles, canonicals, sitemaps, internal links, structured data, security headers and working forms can all be found by a free scan and fixed by whoever deploys the site. What takes sustained effort is the part no scanner does for you: writing pages that answer your buyer's questions and earning links from other sites.

Can ChatGPT do an SEO audit?

It can review HTML you paste in and explain what a tag does, which helps you understand a finding. It is not a fixed bar: ask twice and you may get different advice, and it does not reliably fetch your response headers, redirect chain or every page the way a crawler would. It also cannot tell you, from one conversation, whether it or other engines cite your product across many questions. Use a scanner for the measurement and a chat model for explanations of specific findings.

What is the best free SEO tool?

The one that fits the job you have right now. For a single product site before or after launch, that means a tool that reads your live site, grades named checks, shows the evidence for each verdict, applies the same bar to everyone, and says plainly which findings sit outside its free tier. Run it on your own URL and apply the evidence test from this guide: if the advice would read the same on anyone's site, try another.