How the LLMRadar $197 Audit Works: 4 LLMs, 10 Queries, One PDF

The shortest version: we query ChatGPT, Claude, Perplexity, and Gemini across 10 buyer-intent prompts matched to your category, record exactly where your brand appears (or does not), score you against 3 to 5 competitors, and deliver a PDF with your citation score, the gaps, and a prioritized fix list. The PDF lands in your inbox in under 24 hours.

Here is the longer version, because the method is what makes the output useful rather than just a list of screenshots.

The 4 LLMs We Query and Why Each One Matters

The 4 models in the audit cover the research surfaces where B2B buyers actually look:

• Perplexity is the most important. It runs 46.7% of AI-assisted research queries in B2B (per the Reddit citation study we published in June 2026). When a buyer types "best tools for [your category]" into a search bar that is not Google, Perplexity is often where that query lands. It pulls citations from the open web in real time, which means your content quality and freshness directly affect your citation rate.
• ChatGPT (GPT-4o) is the second most common. The buyers using it are typically mid-funnel: they already know what category they want and are comparing specific options. ChatGPT's training cutoff matters here. Brands that had strong content in the training window before the cutoff appear more consistently than newer brands, regardless of current content quality.
• Claude (Anthropic) handles a growing share of research queries, particularly among technical founders and developers. Claude citation behavior differs from GPT-4o: it tends to favor specific, factual claims over brand reputation signals. Posts with precise data points get cited more often than authority-signal posts.
• Gemini is the Google integration. As AI Overviews expand in Google search results, Gemini citation becomes directly relevant to organic traffic. A brand that ranks in Gemini's AI Overview for a category query can capture position zero even if its organic ranking is 4th or 5th.

Running all 4 matters because each model has a different citation profile for the same brand. A company with strong Perplexity citations but weak Gemini citations has a different remediation path than one with the inverse. The audit shows you where you stand on each, not just an average.

The 10 Query Types

For each client, we construct 10 prompts from 5 query types, paired to the client's ICP (ideal customer profile) and category. The 5 types:

1. Category-level comparison: "What are the best tools for [ICP problem]?" This is the broadest query and the one where category leaders dominate. If you are not mentioned here at all, you have a brand recognition gap at the category level.
2. Competitor alternative: "What are alternatives to [competitor name]?" This is a high-intent query. Buyers asking this are actively shopping. Your presence or absence in competitor-alternative results is a direct revenue signal.
3. Problem-specific lookup: "How do B2B SaaS companies solve [specific problem]?" This catches brands that have built strong educational content but have not translated it into category-level recognition.
4. Feature-specific query: "What tools help with [specific feature or capability]?" Niche citation. Brands that appear here often have strong developer documentation or technical blog content.
5. Brand direct: "Tell me about [brand name]." This checks whether the LLM has a coherent, accurate representation of what you do and who you serve. Surprising number of brands have inaccurate or thin representations here, which limits citation in the other 4 query types.

Two prompts per query type, slightly varied in phrasing, gives 10 prompts total. The variation catches citation inconsistencies: a brand that appears in "best AI visibility tools" but not in "top AI brand monitoring tools" has a positioning gap that the phrasing variation surfaces.

What We Actually Record

For each of the 40 query/model combinations, we record:

• Mention: yes or no. Was your brand name present in the response at all?
• Position: 1st mentioned, 2nd, 3rd, or later. Position matters because AI responses are not ranked lists in the traditional sense, but brands cited first are more likely to be the default recommendation for an undecided buyer.
• Context: What did the model say about you? Accurate description, partial description, or no description? A model that mentions your brand but mischaracterizes what you do is actively creating buyer confusion.
• Competitor comparison: For each query where competitors appear, how does your position compare? This produces a per-competitor citation gap: "Competitor X appears 1st in 7 of 10 Perplexity queries. You appear 3rd in 3 of the same 10 queries."

The raw data is a 40-row spreadsheet. We do not send you the spreadsheet. We send you the interpretation.

The PDF Deliverable: What Is in It

Total: 9-12 pages. No filler, no methodology disclaimers that take up half the document. Every page has a number you can act on or a specific action attached to it.

The Fix Priority List: How We Build It

This is the section buyers find most useful and the one that takes the most work to produce. The inputs:

• Your citation gap by query type (are you missing from category comparisons, competitor alternatives, or direct brand queries?)
• Your existing content inventory (what you already have that we can recommend structuring differently vs. what you need to build from scratch)
• LLM training signal patterns (Claude cites factual claims, Perplexity cites fresh web content, Gemini cites structured markup)

A brand with low Perplexity citations but strong Claude citations usually has good blog content but weak community presence. The fix is Reddit threads and structured discussion posts, not more blog posts. A brand with strong Perplexity citations but weak Gemini citations usually has good freshness signals but weak schema markup. The fix is JSON-LD Article schema, not more content.

The priority list tells you which of those fixes applies to you, in order, with a rough effort estimate for each.

"I expected a score card. I got a roadmap. The fix list told me exactly what to do first and why. We implemented the top two items and saw our Perplexity citation rate jump from 1 in 10 queries to 6 in 10 over 5 weeks." - OperatorIQ Concierge client, 2026

What the Audit Does Not Do

Two things worth naming explicitly:

We do not make predictions about future citation rates. LLMs update their training data and retrieval behavior. A fix that improves your Perplexity citation today may need to be updated in 90 days as Perplexity's web crawl refreshes. The audit gives you a baseline and a direction, not a guarantee.

We do not run the fixes for you. The PDF tells you what to do. The LLMRadar Concierge option ($1,997) has us do the implementation, including writing the citation-anchor content, adding schema markup, and seeding the Reddit and community discussion presence. The $197 audit is the diagnostic. The Concierge is the repair crew.

How Long It Takes

After you pay, the audit runs automatically. You get a confirmation email immediately. The PDF lands in your inbox within 24 hours. In most cases it arrives in 4 to 6 hours.

The delivery system runs on the same GitHub Actions automation stack we described in the autonomous Stripe checkout post: payment confirmed, webhook fires, audit agent runs, PDF generated, email delivered. No human touches it. That is how the 24-hour SLA holds without a support queue.

Who the Audit Is For

The audit produces the most actionable output for B2B SaaS founders and marketing leads who:

• Have a product live with a defined ICP and can name 3 to 5 direct competitors
• Are already publishing some content (blog, documentation, community posts) and want to know why it is not translating into AI citations
• Have run the 5 signs of AI invisibility check and know they have a citation gap, but do not know where it is or how to fix it

The audit is less useful for pre-product founders or companies with no existing content. The fix list assumes you have something to optimize. If you are starting from zero, the Concierge engagement is the better starting point because it includes building the content infrastructure the audit assumes already exists.