Before we audit a client’s site, we apply the same checklist to our own. If we can’t pass it, we have no business charging to run it on someone else’s domain.
That standard cuts both ways — which is why, when a competitor’s SEO landing page caught our attention, we ran our methodology against it. Not to embarrass anyone. To document a set of technical failures that any buyer of SEO or AI visibility services should know how to spot before handing a vendor access to their domain.
The company is a funded AI SEO platform. Their pitch: autonomous agents that fix your site 24/7, publish hundreds of content pieces per month, build 500+ backlinks, and track your citation rate in ChatGPT, Perplexity, and AI Overviews. They sell the idea that AI can fully replace the SEO practitioner. Their own SEO page is an unintentional case study in what happens when no practitioner is actually checking the work.
Here is what we found, and what it means if you are evaluating any AI-powered SEO vendor right now.
Finding 1: The Canonical Tag That Tells Google the Page Does Not Exist
A canonical tag is a one-line HTML element that tells search engines which URL is the authoritative version of a page. It is one of the most fundamental on-page SEO signals, and checking it takes about 30 seconds in any browser’s developer tools or a simple curl fetch.
Their /seo landing page — the page they use to sell SEO services — carries this canonical:
<link rel="canonical" href="https://get-ryze.ai/" />The canonical points to their homepage.
In practical terms, this is the equivalent of publishing a page and simultaneously filing a form with Google that says: “This page is a duplicate of my homepage. Please ignore it. Give all ranking credit, indexation authority, and any backlinks pointing here to the homepage instead.”
Google follows that instruction. The SEO page effectively does not exist in the index as a standalone entity. Every piece of content on it, every keyword it might rank for, every external link it earns — all of it consolidates into the homepage, which is about something else entirely. The page is ranking-dead by design, though not by intent.
A vendor building their business on canonicalization expertise shipped a self-canonical failure on their own highest-stakes conversion page. The audit tool that would have caught this in 30 seconds is free and ships with every browser.
Finding 2: The Metadata Describes a Different Product
Meta descriptions and Open Graph tags are what appear in Google search snippets and social media preview cards. They signal to both search engines and humans what a page is about. Getting them right on a core service page is table stakes.
Their SEO page’s meta description and OG tags describe their paid advertising optimization product — not their SEO service. The title, description, and social preview are all written for a different offering.
The consequences are layered:
For search rankings: Google reads the metadata, reads the on-page content, and observes a mismatch. The page claims to be about one thing in its signals and another thing in its body. This creates topical ambiguity that suppresses ranking confidence.
For click-through: Anyone finding this page in search results sees a snippet describing a product they were not looking for. First impressions misfire before a single word of the page is read.
For social sharing: Every time someone shares this page on LinkedIn, Slack, or in a pitch deck, the preview card announces the wrong product. Their own sales team is working against this metadata every time they share a link.
For AI citations: When ChatGPT, Perplexity, or Claude fetches this page to understand what the company offers, the most prominent, structured signals — title and meta description — describe paid ads. The page about SEO services is being indexed, shared, and cited as a page about advertising software.
Finding 3: The GEO Killer — JavaScript-Invisible Proof Points
This is the most significant failure, and it is particularly damaging for a company selling AI visibility services.
Their results section displays dynamic counters: attributed revenue, ROAS multiples, percentage growth figures. The kind of proof points that close deals. The counters look compelling in a browser — numbers tick up with a smooth animation.
Fetch the page without executing JavaScript. Run curl https://get-ryze.ai/seo, or ask any AI to summarize the page from its URL. Every counter reads zero.
Not “low.” Not a placeholder. Zero. Their results section tells crawlers and language models: this company has generated $0 in attributed revenue, achieved 0× ROAS, and delivered +0% growth for its clients.
This is precisely what Generative Engine Optimization is designed to prevent. GEO is the discipline of ensuring that your most important claims — proof points, differentiators, authoritative statements — are visible in the initial HTML payload that AI systems ingest, not buried behind a JavaScript execution layer that most crawlers and all LLMs skip entirely.
The platform sells a GEO dashboard that tracks how often clients appear in ChatGPT and Perplexity responses. Their own SEO page, when read by those same AI systems, surfaces zero evidence that they have ever produced a result.
The fix is a single line of HTML. The initial value of any counter element should be the real number — JavaScript can still animate it upward, but crawlers see the true figure. We made exactly this change on our own site after our self-audit. The principle: if you cannot read a claim in the page source, a search engine and every AI assistant cannot read it either.
Finding 4: The Internal Contradiction That Signals Low Editorial Control
In one section of their SEO page, the service delivery model is described as: “same human, every Thursday.” In a separate section on the same page, the deliverable is described as a “Wednesday report.”
Thursday check-in. Wednesday report. Same page. Neither was caught before publishing.
This is a small error with a large implication. If the copy describing how the service works cannot be internally consistent on a page that presumably receives the most editorial attention in their business — the page designed to convert buyers — what does content quality look like at the scale of hundreds of blogs per month, generated primarily by autonomous agents?
We flag this not to be pedantic about a day-of-week mismatch, but because it is a visible marker of an underlying quality control problem. The same lack of a second set of eyes that let this through is the same process governing your content, your schema, your technical fixes, if you become a client.
What This Means When You Are Evaluating an SEO Vendor
The four issues above — canonical misdirection, metadata mismatch, JavaScript-invisible proof points, copy inconsistency — are not obscure edge cases. They are the first items on any competent SEO audit checklist. They are caught in minutes with free tools. They represent the baseline of technical hygiene, not advanced expertise.
Before you hire any vendor who manages your organic presence, your AI citation footprint, or your technical infrastructure, run four checks on their own highest-traffic service page:
1. Fetch the canonical. Open developer tools (F12 → Elements → search for “canonical”). The canonical on their service page should point to that service page — not their homepage, not a category URL, not a different product.
2. Read the meta description and OG title. They should describe exactly what the page sells. If they describe a different product, or if they are blank, the vendor does not apply basic metadata hygiene to their own conversion pages.
3. View source (or run curl) and search for their proof numbers. On any page with testimonials, stats, or performance claims, find those numbers in the raw HTML source before JavaScript runs. If you see zero, null, placeholder text, or missing values, those claims are invisible to Google and every major AI system.
4. Find a deliverable claim and look for internal consistency. Find how the vendor describes their weekly process, reporting cadence, or scope of work — then find another mention of it on the same page. If the details do not match, you are looking at content that was not reviewed before publishing.
A vendor who does not apply these four checks to their own site will not apply them to yours.
How We Hold Ourselves to This Standard
We ran this exact audit against mad1sonfoundry.com before publishing this piece. Every page self-canonicals to itself. Every service page carries accurate, page-specific metadata — tested in the OG debugger and confirmed in the search snippet. Every number on our site — case study metrics, proof point statistics, result claims — is present in the initial HTML payload, server-rendered, readable without JavaScript. You can curl our homepage right now and read every client outcome we claim.
We also found and fixed one issue during that self-audit: our counter elements on the homepage were initializing to zero in HTML. Crawlers were reading “0+ Years” and “0 AI Platforms.” Those values are now the real numbers in the HTML — the animation still runs in browsers. Crawlers see the truth.
We do not publish this to claim perfection. We publish it because the discipline of auditing your own site before auditing anyone else’s is the minimum standard of credibility in this field. If a vendor has not earned that standard on their own domain, the offer to earn it on yours is difficult to take seriously.
This isn’t a one-off. We run the same checklist in public on AI-first companies — including our teardown of a top AI marketing agency’s own homepage — and it’s the same discipline behind our AEO & GEO agency services: make the brand structured, extractable, and corroborated, then measure it weekly.
If you want to see what this audit looks like when applied to your own site — canonical structure, metadata accuracy, server-rendered proof points, AI citation footprint, penalty-risk flags — our Rapid Audit delivers a founder-reviewed, revenue-ranked breakdown in 48 hours. No automated PDF. The same methodology documented above, applied by the same person who wrote this piece.
