Visus: The Free Audit Tool Making Websites Findable to ChatGPT, Perplexity and Google AI
Visus audits websites for AI discoverability and offers fixes—schema, FAQ content, entity consistency, llms.txt and trusted citations improving recommendations.
Visus launched as a practical response to a simple but consequential problem: many businesses that show up in traditional search are invisible to modern AI assistants. The free Visus audit scores a site’s “AI visibility” — a measure of how likely large language model–powered services such as ChatGPT, Perplexity and Google AI are to find, understand and recommend a website — and provides prioritized, actionable fixes to close that gap. For organizations that rely on referrals, local discovery or third‑party recommendations, the difference is increasingly material: sites that speak the data language of LLMs move from being overlooked to being suggested in conversational answers.
Why AI visibility is a distinct challenge from traditional SEO
Search engine optimization has long revolved around crawling, indexing and ranking signals tuned for document retrieval. AI-driven recommendation and answer systems use overlapping but different signals: machine-readable entity definitions, canonicalized identity across the web, explicit Q&A content that maps to user prompts, and trusted third‑party citations that reinforce an LLM’s confidence. A site can rank for keywords in Google and still not be surfaced by a conversational agent that synthesizes recommendations from a graph of entities and citations. That makes AI visibility a discrete optimization problem for product teams, marketers and developers who depend on discovery outside traditional search results.
Fozias case study: practical impact of an AI-first audit
A tangible example illustrates the stakes. A local Kashmiri restaurant in Liverpool, operating under the name Fozias, had solid reviews and a loyal customer base but failed to appear in AI-generated restaurant recommendations. An initial Visus audit scored the restaurant 23 out of 100 and identified concrete failings: missing schema markup, a lack of FAQ content matching natural-language queries, inconsistent directory listings, no llms.txt file to guide AI crawlers, and weak entity definition that prevented LLMs from confidently classifying the business.
After implementing the recommended changes — structured Business schema, targeted FAQ content, standardized NAP (name, address, phone) across listings, a llms.txt to permit crawling, and curated third‑party citations — Fozias’ Visus score rose to 80 within 90 days. The restaurant began appearing as the top result for “Kashmiri restaurant in Liverpool” on Perplexity and was included in ChatGPT dining recommendations. That jump demonstrates how targeted structural and content changes yield measurable rediscovery by AI systems that synthesize recommendations rather than rely solely on keyword ranking.
Visus auditing itself: lessons from fixing the product site
An internal audit of getvisus.com produced a humbling result: Visus scored only 35 out of 100 on its own AI visibility audit. Problems included inconsistent canonical URLs across domains, missing llms.txt, absent FAQ schema, thin crawlable content related to client-side rendering, and a lack of external authoritative citations. After a focused remediation session — adding SoftwareApplication, Organization, WebSite and FAQPage schema; creating a detailed llms.txt; correcting canonicalization; and adding substantive FAQ entries — the score increased to 76. The rapid improvement underlined two practical truths: the list of priority fixes is short, and implementing them yields outsized gains in AI discoverability.
The five signals that most influence AI visibility
Visus’ experience and audits point to five core signals that LLM-based systems favor when deciding whether to surface a website:
-
Schema markup: Machine-readable definitions matter. Structured schema types such as LocalBusiness, SoftwareApplication, Organization, WebSite and FAQPage give AI systems an authoritative way to classify entities, surface attributes (hours, location, feature lists), and connect the site to a semantic graph.
-
FAQ content tuned to natural language queries: Conversational agents map user prompts to content that directly answers those prompts. FAQ pages written in the language people use when asking an assistant improve the probability that an LLM will extract and cite the site when composing a recommendation or answer.
-
Entity consistency across the web: Identical business name, address, phone number and descriptions across directories, social profiles and partner sites reduce ambiguity in entity resolution. LLMs rely on consistent signals to merge mentions into a single confident entity.
-
External citations from trusted sources: Third‑party references on authoritative domains increase model confidence. Mentions in reputable directories, news outlets, review platforms and industry sites serve as corroborating evidence that an entity is real and noteworthy.
- llms.txt and crawlability: A llms.txt file provides simple, explicit guidance to LLM crawlers — which endpoints are permitted, update cadence, canonical hostnames and other crawler instructions. Equally important is ensuring content is crawlable: server-side rendering or pre-rendered HTML is preferable to relying solely on client-side JavaScript that some crawlers struggle to process.
These signals are not merely checkboxes; they interact. Schema without credible citations can look like self-declared claims. Citations without canonicalization can fragment entity identity. Effective AI visibility strategy addresses the signal set holistically.
Implementing the fixes: a practical checklist for developers and site owners
Teams that want to act should prioritize fixes that are high-impact and low-friction. The following implementation checklist translates Visus recommendations into concrete steps:
-
Add or enrich schema markup: Use JSON-LD for structured data and include relevant schema types (LocalBusiness, Organization, WebSite, SoftwareApplication, FAQPage). Populate core properties such as name, address, telephone, logo, url and description. For software products, include applicationCategory, operatingSystem and offers where relevant.
-
Create question-driven FAQ content: Audit customer queries, support logs and conversational search prompts to identify the top user questions. Author concise, direct answers and surface them in an FAQPage schema block. Make sure answers match user intent rather than marketing copy.
-
Standardize entity data (NAP+W): Build a canonical reference for name, address, phone and business description and propagate it to Google Business Profile, major directories, partner sites and your own properties. Prefer exact string matches to minimize entity fragmentation.
-
Publish a llms.txt file: Place a llms.txt at the site root that declares crawl permissions, sitemap locations, canonical hostnames and recommended scraping intervals. While the llms.txt specification is still lightweight, it provides important signals to AI crawlers and can reduce accidental blocking.
-
Fix canonical URLs and redirects: Ensure a single canonical host and canonical link tags are present on every page. Use 301 redirects for alternate domains and ensure internal links point to the canonical host.
-
Make content crawlable: For JavaScript-heavy sites, implement server-side rendering, static pre-rendering or dynamic rendering to produce HTML snapshots for crawlers. Ensure meta tags, structured data and critical content are present in the server-rendered HTML.
-
Gather and curate external citations: Pursue mentions on local news sites, well-known directories, industry blogs and review platforms. Encourage accurate, consistent citations by partners and vendors.
- Monitor and iterate: Track Visus scores, LLM-driven ranking signals (where measurable), citation growth and referral lift. Schedule regular re-audits to stay ahead of changing models and patterns.
These tasks map cleanly onto engineering sprints and marketing campaigns: schema and llms.txt are developer tasks, FAQ content and citation outreach are marketing tasks, and canonical fixes require both.
How Visus works and who can use it
Visus is positioned as a low-friction diagnostic tool for a broad audience: local business owners, product managers, technical SEO teams and developers. The product experience is intentionally simple — enter a URL and receive an AI visibility score within about 60 seconds, with no signup or credit card required. Under the hood, Visus runs a checklist of structural and content signals that matter to LLMs, assesses crawlability and canonicalization, inspects schema and FAQ presence, and reports citation gaps. The output is a prioritized set of issues and suggestions that teams can act on directly. For developers, the report identifies technical blockers (client-side rendering, canonical inconsistencies, missing files) while for marketing owners it highlights content and citation opportunities.
Availability: Visus is available as a free public service and is designed to lower the barrier to entry for businesses that want to be discoverable by conversational AI. Because the tool is web-based and requires only a URL to produce a score, it can be used immediately by site owners and consultants as part of routine audits.
Measuring success and monitoring AI-driven discovery
Unlike traditional rank-tracking, measuring AI visibility requires a combination of structural metrics and outcome metrics. Structural metrics include Visus score, percentage of pages with structured data, presence of llms.txt, number of FAQ entries, and degree of canonical consistency. Outcome metrics to watch include:
- Inclusion in AI-generated answers and recommendations (as observed in Perplexity, ChatGPT and Google AI results).
- Referral traffic uplift from AI-driven properties.
- Increase in branded searches that indicate awareness.
- Growth in third-party citations and authoritative mentions.
Because conversational agents synthesize information from multiple sources, you may see delayed but durable gains — a single credible citation combined with correct schema and a strong FAQ can tip an agent’s confidence and result in placement that persists as models continue to use the citation graph.
Developer implications and integration with engineering workflows
Improving AI visibility intersects with core engineering responsibilities: canonicalization, rendering strategy, content APIs and schema deployment. For modern web stacks, teams should consider:
- Incorporating structured-data generation into component libraries or server-rendered templates so schema is output consistently across pages.
- Adding llms.txt generation into deployment pipelines alongside robots.txt and sitemaps.
- Treating FAQ content as content artifacts in CMSs or headless content stores, enabling editorial teams to add and update answers without developer intervention.
- Using canonicalized entity records (a single source of truth) consumed by marketing dashboards, CRM systems and partner integrations to eliminate duplication.
These actions make it easier to maintain accuracy at scale and ensure that engineering and content teams are aligned on signals that matter to LLMs.
The relationship between AI visibility and adjacent ecosystems
AI visibility is not isolated; it sits at the confluence of content strategy, structured data, CRM, review platforms and automation ecosystems. For example, marketing teams can push canonical business metadata from a CRM to public directories and partner sites, thereby strengthening entity signals. Automation platforms that publish FAQ updates or push citation requests can accelerate the citation-building process. Security and privacy tools need to be considered as well: the data you publish must be accurate and acceptable under applicable privacy rules and platform terms. Product and developer tools that expose structured APIs make it simpler for content and SEO teams to embed canonical data into every outward-facing property.
References to related topics such as "structured data best practices," "local search optimization," "conversational AI in customer acquisition," and "developer tooling for static rendering" are natural internal link candidates for teams documenting their own processes or building knowledge bases.
Risks, ethical considerations and pitfalls to avoid
As LLMs increasingly influence recommendations, organizations should avoid manipulative or dishonest tactics. Inflating schema with false claims, creating contrived citations on low‑quality domains, or publishing misleading FAQs to game conversational prompts can erode trust and invite platform penalties. Instead, aim to be factual, transparent and consistent. Third‑party citations should reflect real relationships and coverage; schema should accurately describe the business or product; and FAQ answers should be concise and truthful. Over-optimization of trivial signals without building genuine credibility will have limited long-term effect — AI systems are designed to weigh corroborating signals and penalize incoherent or inconsistent claims.
Business use cases and return on investment
The most immediate beneficiaries are local service providers, restaurants, small retailers and software products that rely on discovery and referrals. For SaaS companies, adding clear SoftwareApplication schema, concise FAQ content, and authoritative case study citations can increase the chance that an AI assistant recommends the product when users ask for solutions in a given category. For local businesses, consistent NAP data and citations increase the odds of appearing in curated assistant responses for “places to eat” or “services near me.” The investment tends to be modest — schema implementation, a structured FAQ and a llms.txt file — relative to the lifetime value of acquiring customers through conversational recommendations.
Early adopters stand to capture disproportionate visibility: when competitors ignore AI visibility, the first businesses that optimize structural signals become reference points in recommendation graphs and can enjoy multi-year head starts in being suggested by conversational agents.
How agencies and SEO teams should adapt their offerings
Digital agencies and SEO consultancies will need to expand their audit checklists beyond traditional indexation and backlink analysis to include LLM-specific signals. Service offerings should package schema generation, llms.txt creation, FAQ production, citation outreach and crawlability remediation as a combined service. Reporting should include Visus-style scores, evidence of AI inclusion (screenshots or logs of agent results), and growth in authoritative citations. Agencies that can operationalize these tasks — integrating engineering, content and outreach — can offer rapid, measurable improvements to clients’ discoverability in AI channels.
Broader implications for search, discovery and the software industry
The rise of AI assistants reframes the concept of “search”: influence shifts from ranking pages to being recognized as a credible entity within a semantic knowledge graph. That change affects product roadmaps, marketing budgets and even how software vendors structure documentation and API references. For developers, the imperative to expose clear, machine-readable metadata will become routine. For businesses, the fight for visibility will increasingly happen in an environment where a single trusted citation can determine whether a company is recommended.
Platforms and standards will likely evolve to accommodate AI crawlers: llms.txt and richer schema vocabularies are early signals of a maturing ecosystem. As AI systems standardize on certain signals, industry best practices will coalesce, creating both opportunities for optimization and expectations around transparency and accuracy.
Brands that treat AI visibility as an infrastructure problem — combining engineering, content and reputation work — will be better positioned to capture demand generated by conversational agents. Conversely, organizations that focus narrowly on keyword rankings risk missing a growing channel of demand mediated by LLMs.
The practical shift is straightforward: adapt site architecture and content practices to be readable by machines, and ensure the broader web reflects a single, consistent identity for your organization or product.
Looking ahead, the interplay between conversational AI and structured web data will continue to evolve. Standards like llms.txt may gain wider adoption and tooling will emerge to automate schema deployment and citation tracking. Large language models will improve entity linking and may place greater weight on verified external citations, increasing the value of authoritative coverage. For companies and developers, the near-term priority is clear: adopt machine-readable practices now to be part of the recommendation graphs that will shape discovery over the coming years.
