How long does it take to run an AI visibility audit manually?

A manual AI visibility audit takes 15 to 25 hours. This includes checking robots.txt, testing 5 queries on 4 LLMs, analyzing verbatims, Schema.org, EEAT signals, and writing recommendations and fix files.

Why is an automated AI visibility audit preferable?

An automated tool like RadarLLM runs 120 tests (10 questions × 3 runs) across 4 AI engines in under 10 minutes, generates fix files (robots.txt, llms.txt, JSON-LD schemas) and produces a professional PDF report. It’s 100x faster, reproducible, and includes a re-audit at D+30.

2026-04-088 minComparison

AI visibility audit: DIY or use a tool?

Cyril Cretin

Founder RadarLLM.ai

Yes, you can audit your AI visibility manually. But expect 15 to 30 hours of work, $500 to $2,500 with a consultant, and results that aren’t reproducible: LLMs give different answers every time you ask. Here’s what a manual audit looks like compared to an automated tool.

What a manual audit involves

Running an AI visibility audit yourself is technically possible. But each step requires time, technical skills and a lot of rigor. Here are the 8 essential steps.

1. Check robots.txt for each AI bot

30 min

GPTBot, ClaudeBot, PerplexityBot, GoogleBot-Extended, OAI-SearchBot, Bingbot, DuckAssistBot, anthropic-ai, cohere-ai. Each has its own rules. Miss one and you’re invisible to an entire engine.

2. Test 5 queries on each of 4 LLMs

2-3h

Craft 5 realistic conversational questions, ask them on ChatGPT, Gemini, Perplexity and Claude. Copy each response. Repeat for each LLM. That’s 20 manual tests minimum.

3. Record verbatims and brand mentions

1-2h

For each response, identify if your brand is cited, in what context, with what sentiment (positive, neutral, negative). Document the sources mentioned by each AI.

4. Analyze Schema.org markup

Check for Organization, FAQPage, Article, HowTo, BreadcrumbList, Person. Validate each schema with Google Rich Results Test. Identify errors and missing schemas.

5. Review EEAT signals across the site

2-3h

Browse the site to verify: author pages, bios, testimonials, legal notices, privacy policy, visible contact, certifications, update dates, cited sources.

6. Check llms.txt and content structure

Does the site have an llms.txt file? Is content structured as Q&A? Are paragraphs 30-50 words? Are questions in H2 tags? Are lists properly structured?

7. Write recommendations

3-5h

Synthesize all findings, prioritize by impact, write actionable recommendations with effort level and estimated timeline. This is the core value of the report.

8. Create fix files manually

5-10h

Write custom llms.txt, fix robots.txt, generate JSON-LD schemas, reformulate AEO content. Each file must be adapted to the site’s industry and pages.

Estimated total: 15-25 hours minimum

Not counting the learning curve if you’re not an SEO/GEO expert. And the result is only valid at that moment: LLMs change their answers constantly.

The problem with manual audits

⚠ Non-deterministic

LLMs give different answers every time. Ask ChatGPT the same question twice in a row: you’ll get two different responses. Your Monday audit won’t be valid by Wednesday.

⏱ Time-consuming

15 to 25 hours for a single site. If you have multiple sites or you’re an agency, it’s unsustainable. And each re-audit takes almost as long as the first.

↻ Not reproducible

Impossible to rerun exactly the same audit to measure progress. Conditions change with every test: model version, time, location, conversation history.

✗ No fix files

A consultant gives you recommendations. But they don’t generate the corrected robots.txt, the custom llms.txt, the ready-to-paste JSON-LD schemas. You have to implement everything yourself.

$ Expensive

A specialized consultant charges $500 to $2,500 per audit. And prices are rising as GEO/AEO demand explodes in 2026. A re-audit at D+30 doubles the bill.

What an automated audit delivers in 10 minutes

RadarLLM automates the entire process. You enter your URL, choose your industry, and in 10 minutes you receive a complete diagnosis with fix files ready to deploy.

✓ 120 systematic tests across 4 AI engines

10 conversational questions adapted to your industry, asked 3 times to ChatGPT, Perplexity, Gemini and Claude. Automatic analysis of citations and of the competitors recommended instead of you.

✓ Fix files generated automatically

Corrected robots.txt with all AI bots allowed, custom llms.txt, JSON-LD schemas (FAQPage, Organization), optimized meta tags. Everything in a ZIP ready to upload.

✓ Professional PDF report

AI-Ready score out of 100, detailed scores per module, top 3 strengths and weaknesses, raw AI verbatims, 10 recommendations prioritized by impact/effort/timeline.

✓ Re-audit available (Expert, D+30)

The Expert tier includes an automatic re-audit at D+30 to measure the impact of your fixes. Same methodology, same queries, objective comparison.

Full comparison

	Manual/Consultant	RadarLLM
Price	$500-$2,500	From $24.50
Duration	15-25 hours	~10 min
LLMs tested	1-2 (manually)	4 (automated)
Fix files	Not included	ZIP included
Reproducible	No	Yes
Re-audit	Additional invoice	Included (Expert D+30)
PDF report	Depends on consultant	Always included

Our recommendation

Start with RadarLLM’s free scan to find out in under a minute whether Perplexity cites you. You’ll immediately know where you stand. If you need the full audit with fix files, it’s 100x faster and 10x cheaper than a manual audit.

Save the consultant for custom strategy, not data collection. Use an automated tool for diagnosis and fix files, then bring in an expert to define a content and distribution strategy tailored to your market.

In summary

Data + fix files = automated tool (RadarLLM). Strategy + guidance = consultant. They’re complementary, not interchangeable.