If you run a SaaS or content site in 2026, you've probably noticed something: your organic traffic isn't what it used to be.
You're not imagining it. In 2025, Gartner predicted that search engine traffic would drop 25% by 2026 as users shift to AI chat interfaces. When I checked our analytics in Q2, the number was closer to 30% for some of our landing pages.
The problem isn't that your SEO is bad. It's that a growing share of your audience never touches a search engine. They ask ChatGPT. They ask Claude. They ask Gemini. If your site isn't visible in those answers, you don't exist to them.
So how do you check your AI search visibility? You run an audit.
What GEO Auditor Checks
GEO Auditor is an Apify Actor built specifically for this job. It runs two categories of checks against your website:
6 Technical SEO Audits:
- Robots.txt compliance (can LLM crawlers access your content?)
- Sitemap health and completeness
- SSL certificate validity
- Core Web Vitals (LCP, FID, CLS)
- Structured data presence (Schema.org, Open Graph)
- Mobile responsiveness
3 LLM Probe Tests:
- ChatGPT Probe: Can GPT-4o retrieve and correctly cite your content?
- Claude Probe: Can Claude 3.5 Sonnet summarize your site accurately?
- Gemini Probe: Can Gemini 1.5 Pro extract key facts from your pages?
Each probe asks the LLM to answer a specific question about your business — and then evaluates whether the answer is accurate, cites correctly, and reflects your actual offering.
Step-by-Step Walkthrough
Step 1: Open GEO Auditor in Apify Console
Head to the GEO Auditor page on Apify Store. Click "Try" or "Run" to open the input form.
The interface looks like this:
`
┌─────────────────────────────────────────┐
│ GEO Auditor — AI Search Visibility │
│ │
│ [Website URL] [https://your-site.com] │
│ │
│ [Company Name] [Your Company] │
│ │
│ [Business Description] │
│ A short description of what your │
│ company does, for the LLM probes. │
│ │
│ [Sample Question to LLMs] │
│ e.g. "What does [company] do?" │
│ │
│ [LLM Providers to Test] │
│ ☑ ChatGPT ☑ Claude ☑ Gemini │
│ │
│ [▶ Run] │
└─────────────────────────────────────────┘
`
Step 2: Fill in Your Site Details
For this example, let's say we're auditing our own site at blog.fork.work.
- Website URL: https://blog.fork.work
- Company Name: fork.work
- Business Description: A personal blog by an indie developer building AI-powered tools and writing about web scraping, RAG pipelines, and product development.
- Sample Question: What does fork.work write about?
The sample question is important — it's what each LLM probe will be asked during the test.
Step 3: Run the Audit
Click "Run". GEO Auditor starts two parallel workflows:
1. The technical audit fires off checks against your domain — sitemap, robots.txt, SSL, Core Web Vitals via Lighthouse data.
2. The LLM probes dispatch your sample question to each selected AI provider and capture the full response.
A full audit on a typical site takes about 6–8 minutes. You'll see progress bars for each check.
Step 4: Read the Results
When the audit completes, the output is a structured JSON document. You can view it directly in Apify Console or download it. Here's what it looks like:
`json
{
"url": "https://blog.fork.work",
"overallScore": 78,
"technicalAudit": {
"robotsTxt": {
"status": "passed",
"detail": "robots.txt found, allows all user agents"
},
"sitemap": {
"status": "warning",
"detail": "Sitemap found but missing lastmod dates on 3 entries"
},
"ssl": {
"status": "passed",
"detail": "Valid SSL certificate, expires in 85 days"
},
"coreWebVitals": {
"status": "passed",
"detail": "LCP: 1.2s, FID: 12ms, CLS: 0.05"
},
"structuredData": {
"status": "failed",
"detail": "No Article schema found on blog pages"
},
"mobileResponsiveness": {
"status": "passed",
"detail": "Viewport meta tag present, responsive layout detected"
}
},
"llmProbes": {
"chatgpt": {
"status": "accurate",
"response": "fork.work is a blog by an indie developer...",
"citesSource": true,
"accuracyScore": 92
},
"claude": {
"status": "partial",
"response": "...",
"citesSource": false,
"accuracyScore": 65
},
"gemini": {
"status": "accurate",
"response": "...",
"citesSource": true,
"accuracyScore": 88
}
},
"recommendations": [
"Add Article schema to blog posts for better LLM extraction",
"Fix sitemap lastmod dates to improve crawl efficiency",
"Consider adding an LLMs.txt file for explicit AI guidance"
]
}
`
Interpreting Your Score
The overall score ranges from 0–100. Here's a rough guide:
| Score | Meaning | Action |
|-------|---------|--------|
| 90–100 | Excellent | Monitor monthly |
| 70–89 | Good | Fix warnings, re-audit quarterly |
| 50–69 | Needs work | Prioritize failed checks |
| Below 50 | Critical | Start with robots.txt and sitemap |
In the example above, a score of 78 means the site is visible to AI models but has gaps. The LLM probes showed that Claude struggled to find and cite the content correctly — likely because there's no Article schema markup. That's a quick fix.
Quick Fixes for Common Issues
Based on running GEO Auditor across a dozen sites, here are the three most common problems and how to fix them:
1. Missing Structured Data
LLMs heavily rely on Schema.org markup to understand page content. Without Article, Organization, or WebPage schema, models may misinterpret your content.
Fix (for blog posts):
`html
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "Your Post Title",
"datePublished": "2026-07-14",
"author": {
"@type": "Person",
"name": "Your Name"
}
}
`
2. No LLMs.txt File
A relatively new standard, llms.txt tells AI models what your most important pages are and how to use them. It's like a robots.txt for LLMs.
Fix: Create /llms.txt at your domain root:
`
fork.work
> A blog about indie dev, scraping, and AI tools.
Important pages
- https://blog.fork.work/ai-search-visibility-audit.html: Guide to AI search visibility
- https://blog.fork.work/competitor-pricing-analysis.html: Competitor analysis guide
About
https://blog.fork.work/about.html
`
3. Slow Core Web Vitals
If your pages take more than 2.5 seconds to load Largest Contentful Paint (LCP), LLMs may time out before reading your content.
Fix: Enable CDN caching, optimize images, defer JavaScript. A static site generator (like this blog) usually scores well automatically.
Why This Matters
AI search visibility isn't a future trend — it's happening now. When someone asks ChatGPT "What's the best tool for web scraping?", you want your article to be in that answer. When a prospect asks Claude "Show me SaaS pricing intelligence tools", being cited is the new #1 ranking.
The 8-minute audit from GEO Auditor gives you a baseline. Run it once, fix the issues, and re-audit in a month. Your AI search ranking is the new SEO — and it's still early enough to get ahead.
Want to go deeper? Check out fork.work for more posts about building AI-powered tools and scraping infrastructure.