How we improved an AI Visibility Readiness Score from 60 to 95
A score is useful only when you can explain where every point came from. This case study documents the exact changes made to AIVisora, what machines can now understand more clearly, and what a 95 still cannot promise.
By the AIVisora Editorial Team
Before
60
Core page content was readable, but identity and discovery signals were incomplete.
After
95
The organization, software product, preferred URLs and important pages became explicit.
The result
What the first audit actually found
The original homepage was not broken. It loaded over HTTPS, used readable headings and explained the product. That is why the starting score was 60 rather than zero. The gap came from a more subtle problem: several facts a human could infer were not stated in machine-readable ways.
A visitor could see the AIVisora name and understand that it was a website audit tool. A crawler had less certainty about the official organization, preferred homepage, product type, contact identity and complete set of public pages. Good optimization work begins by removing that ambiguity, not by repeating keywords.
The seven changes that moved the score
Valid JSON-LD
Added Organization, WebSite and SoftwareApplication entities.
Organization identity
Published the official name, URL, logo and contact channel.
Canonical URL
Declared one preferred homepage address to reduce URL ambiguity.
XML sitemap
Published important public URLs at the site root and referenced it in robots.txt.
About page
Added factual product purpose, methodology, team model and contact information.
Search snippet
Rewrote the page title and description to be specific, concise and useful.
Verified social profile
Connected the official Instagram profile and added it to Organization sameAs.
Why structured data mattered most
Structured data does not make a weak business trustworthy. It gives machines a precise vocabulary for facts that are already true. AIVisora added a linked graph containing an Organization, a WebSite and a SoftwareApplication. The organization entity contains the official name, canonical URL, logo, support email and verified social profile.
Linking these entities is more useful than scattering unrelated schema blocks around the page. The website points to its publisher, and the software product describes its category, URL and free offer. Every property is visible or supportable elsewhere on the site.
The vocabulary follows the official Schema.org Organization documentation, rather than a custom set of labels that other systems would not recognize.
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "AIVisora",
"url": "https://useaivisora.com",
"logo": "https://useaivisora.com/logo.svg"
}Canonical URLs and sitemaps solve different problems
The canonical link tells a crawler which version of a page is preferred. The XML sitemap lists the public pages the site wants crawlers to discover. One reduces duplication ambiguity; the other improves discovery. Neither replaces useful internal links.
AIVisora now exposes its sitemap at the site root and references it from robots.txt. Private account pages remain excluded. Public product, pricing, contact, legal, about and editorial pages can be crawled without authentication.
Google documents sitemap discovery and submission in its Search Central sitemap guide. Submission is a discovery hint, not a promise that every URL will rank.
The title and description became more specific
The earlier search snippet tried to explain too much. A useful title should identify the brand and primary offer without reading like a keyword list. The revised homepage title is "AIVisora - AI Visibility Readiness Checker." The description explains the transparent 24-check audit and the four readiness areas in one concise sentence.
This improves clarity, not merely character count. Someone scanning a search result can understand what the product does before clicking, while a crawler receives a consistent topic signal across the title, description and heading.
Trust signals work as a connected system
An About page earns no trust simply by existing. It becomes useful when its claims agree with the rest of the website. The product name, domain, support email, social profile and structured-data identity should describe the same organization. Contradictions create more doubt than a missing page.
For AIVisora, the About page explains the product's purpose, the four audit categories, the remote-first operating model and the official contact channel. It deliberately avoids invented awards, inflated customer counts and anonymous quotations. A smaller set of verifiable facts is stronger than a long company story nobody can confirm.
The same rule applies to social profiles. Only the official Instagram account was connected. Placeholder links for networks without a verified AIVisora presence were removed. That decision sacrificed the appearance of being everywhere in exchange for a cleaner identity trail.
How we verified the improvement
A scoring tool should be predictable enough to challenge. We ran the unchanged page through the same 24 checks more than once and confirmed that it produced the same result. We also tested different HTML structures and confirmed that they produced different scores. The implementation contains no random-number source and no domain-specific score for AIVisora.
We then inspected the rendered production HTML, not only the source files. That distinction matters in modern JavaScript applications: metadata can look correct in a component while failing to reach the final response a crawler receives. The canonical tag, description, JSON-LD, robots file and sitemap were checked at their public URLs.
Finally, we tested failure states. Private dashboard pages remain excluded from indexing, authentication routes do not appear in the sitemap, and the audit system reports unreachable sites without inventing a result. Reliable failure behavior is part of a trustworthy measurement product.
A repeatable order of work
- 1
Start with the rendered page
Record the response status, title, description, headings, links and structured data before changing anything.
- 2
Fix discovery before decoration
Confirm HTTPS, canonical URLs, robots rules and the sitemap so crawlers can reach the pages worth evaluating.
- 3
Make identity factual
Align the About page, contact channel, official social profiles and Organization schema.
- 4
Improve the reader's decision
Rewrite vague titles and descriptions so a person can tell what the page offers and whether it is relevant.
- 5
Retest and preserve evidence
Run the same checks again, compare the exact failures and publish what changed without turning the score into a ranking claim.
Why the result stopped at 95
A perfect score would have been easy to manufacture by adding arbitrary links. We chose not to do that. The remaining gap concerned relevant external references. An authoritative reference should support a claim or help a reader continue learning; it should not exist solely to trigger a scoring rule.
This editorial section creates a natural place to cite primary documentation when it genuinely supports an explanation. That approach is slower than adding random links, but it produces a site worth returning to.
A practical readiness checklist
- Use one clear title and one useful meta description.
- Declare the canonical URL for each indexable page.
- Publish a valid sitemap and reference it from robots.txt.
- Describe real organizations, products and services with valid JSON-LD.
- Maintain factual About, Contact, Privacy and Terms pages.
- Connect only verified social profiles.
- Cite authoritative sources where they genuinely support a claim.
- Retest the same unchanged page and confirm the score is deterministic.
Frequently asked questions
Does a higher score guarantee rankings?
No. It measures observable readiness signals. Rankings also depend on query relevance, content quality, competition, authority and many external factors.
Why did structured data make such a large difference?
It explicitly identifies the organization, website and software product, reducing ambiguity for systems that process the page.
Should every website copy the same schema?
No. Schema must match the real entity and visible content. A local business, publisher, ecommerce store and software company require different facts.
Audit your own website
See the checks, evidence and recommendations behind your readiness score.
Run a free audit