Find out what’s wrong
before you pay anyone to fix it.
Three self-serve tools built from the same playbook we use on client work. No account, no trial, no credit card — and the two that touch your data never send it anywhere.
Most free SEO tools are a lead form wearing a lab coat. You paste a URL, wait for a progress bar that is really just a timer, and receive a score with no method behind it and a call booking link underneath. The score exists to make you anxious enough to book the call.
These three do the opposite. Each one answers a specific question we kept being asked, does the actual analysis rather than performing it, and shows its working — every finding comes with the reason it matters and the fix. There is nothing to unlock, because there is no paid tier to unlock it into.
They are free because of how our business works, not in spite of it. We are a studio: we make money designing and building things. A tool that tells you honestly that your site is fine costs us nothing, and a tool that tells you honestly that it isn’t has already done the hardest part of the sales conversation. Being genuinely useful is cheaper than marketing.
Three questions, three answers.
Start where the symptom is.
If you don’t know where you stand, run the audit. It needs nothing but a URL, it finishes in seconds, and it checks the things that decide whether an answer engine can read, trust and quote your page at all. Most sites discover something in the first run that takes an afternoon to fix and has been costing them for a year. If you only ever use one of these, use this one.
If traffic is falling and you want to know why, use the Recovery Engine. The revealing pattern is a page that keeps its impressions but loses its clicks — it is still being shown and no longer being chosen, which is a very different problem from being deranked, and it needs a very different fix. The tool finds those pages, sorts them by what you stand to win back, and tells you which internal links would help.
If you publish a lot, run the Cannibalization Autopilot. Content libraries turn on themselves eventually. Two articles drift toward the same question, split the authority that should have gone to one, and both underperform what either would have managed alone. The hard part is that they often share no keywords, so keyword tools miss it entirely — this one compares intent, and then tells you which page deserves to win.
None of them requires the others. They just answer different questions, and the audit is the one that tells you which question you have.
What the audit actually checks.
Eighteen checks in five categories. Every one is a thing an answer engine genuinely uses when it decides whether to quote a page — nothing is in here to pad a score out of a hundred.
AI crawlability
Whether AI crawlers are allowed to fetch you in the first place, and whether you have published llms.txt, a valid robots.txt and a sitemap. This category is first because it is binary: if a model’s crawler cannot reach the page, nothing else on this list matters.
Structured data
Whether the page ships JSON-LD at all, and whether that JSON-LD defines a clear entity — a named organization or person the engine can resolve and attribute a claim to. Markup that describes a page is table stakes. Markup that establishes who is speaking is what earns the citation.
Metadata
Title, meta description, canonical, Open Graph and Twitter cards. Old-fashioned, unglamorous, and still the text an engine reaches for when it needs to say in one line what your page is.
Content
How much extractable text is actually in the HTML, whether the headings form a real structure, whether images carry alt text, and whether the page exposes an author and a date — the E-E-A-T signals that separate a source from a page. The text check catches the most expensive modern mistake there is: rendering your entire article in JavaScript, so the crawler receives an empty shell and you never find out.
Technical
HTTPS, a declared page language, and a viewport. Small things. All three are the kind of small thing that quietly disqualifies you.
Where your data actually goes.
Two of these tools take a Search Console export, which is among the more sensitive files a business owns — every query you win, every page that is slipping, in one CSV. So the Recovery Engine and the Cannibalization Autopilot never upload it. The parsing and the analysis run in your browser, in JavaScript, on your machine. The file does not reach our server, because there is no code on our server that would receive it.
The AI Readiness Audit genuinely cannot work that way, and we would rather say so than bury it. To inspect your HTML it has to fetch your page, and a browser will not let a page fetch another origin — so the request goes from our server, exactly as any crawler’s would, to a URL you have chosen to make public. What we keep from it is the hostname, the score and the grade, anonymously and in aggregate, and that record is what our research reports are built from. We do not tie it to you, and we never publish a named site’s result.
The full detail is in the privacy policy.
Common questions about the tools.
Once you know what’s wrong.
A score is a starting point, not an education. If the audit turned up something you want to understand rather than just patch, the guide to generative engine optimization explains what these engines are actually doing, and how to get cited by AI covers the work itself. If you want the why underneath all of it, start with what large language models are.
For the numbers behind the advice, the reports are where we publish what these tools have learned across every site they have run on. And if you would rather someone just did it, that is the work we do.