Published 2026-06-05 by Christine Johnson, OperatorIQ
The LLMRadar Brand Audit runs 40 buyer-intent queries across Claude, ChatGPT, Perplexity, and Gemini. It scores your brand's citation rate for each engine, checks your top 5 pages for the 7 SAIO structural rules, and delivers a PDF with your scores plus a prioritized fix list. The whole process runs in under 4 hours. You get the report via email the same day. One flat fee. No subscription. No call required.
- 40 queries total: 10 per LLM, covering buyer-intent, comparison, and category searches specific to your brand
- Citation rate measured per engine as a percentage, with a consistency score across two runs
- Page structure scoring covers all 7 SAIO rules: FAQ schema, Article schema, entity density, llms.txt, TL;DR coverage, list structure, and internal linking depth
- The report gives you a fix priority order, not just scores
- You can run the audit on any public URL, including competitor domains
Three weeks ago we published the 5 signs your SaaS is invisible to AI search. The number-one reply was a variation of: "Okay, but how do I actually know which of those applies to my site?" Reading a checklist is not the same as seeing your numbers.
This post explains exactly what the $197 LLMRadar Brand Audit checks, how the scoring works, and what you get back in the PDF. If you already know you want to run it, the link is at the bottom. If you want to understand the methodology first, read on.
What "LLM citation rate" actually means
Before getting into what the audit checks, it is worth being precise about the term. Citation rate is the percentage of relevant queries where your brand appears in the AI model's response.
If your brand appears in 3 out of 10 Claude queries about your product category, your Claude citation rate is 30%. If it appears in 0 out of 10 ChatGPT queries about the same category, your ChatGPT citation rate is 0%.
Most SaaS brands that have never audited their AI visibility assume their citation rate is somewhere in the middle. The reality we see in audits is more extreme: brands either appear in most relevant queries or in almost none. There is very little "sometimes." The reason is structural. The 7 SAIO page-structure rules are either mostly present on your site or mostly absent. When they are absent, all four LLMs skip you consistently.
The audit measures the actual rate, per engine, per query category. That number is your baseline. Everything else in the report is context for how to move it.
The 40 queries: how they are selected and run
The 40 queries are not generic. They are specific to your brand, your product category, and the buyer language your actual ICP uses when searching.
Here is how the query set is built:
- Category queries (10 total, 2-3 per engine): "What are the best tools for [your product category]?" and close variants. These measure whether your brand appears when a buyer is actively evaluating options in your space.
- Problem queries (10 total, 2-3 per engine): Specific pain-point queries. "How do [ICP role] handle [the specific problem you solve]?" These measure whether your brand surfaces as a solution to the exact problems you claim to fix.
- Comparison queries (10 total, 2-3 per engine): "[Your brand] vs [top competitor]" and "alternatives to [category leader]." These measure whether you appear in the comparison set buyers use when they are close to a decision.
- Entity queries (10 total, 2-3 per engine): Direct brand queries. "What is [your brand] and who is it for?" These measure whether the LLMs have a clean, accurate entity definition for your brand, or whether they are guessing.
Every query is run twice per engine to check consistency. A brand that appears in 7 out of 10 category queries on Claude has a stable citation signal. A brand that appears in 2 of 10 on one run and 8 of 10 on the next has a weak signal vulnerable to model updates. Both the rate and the consistency score appear in your report.
The 7-rule page structure check
Citation rate tells you where you stand. Page structure scoring tells you why.
After the query run, the audit checks your top 5 pages (homepage, your highest-traffic product page, and your 3 most-visited blog posts) against the 7 SAIO structural rules:
| Rule | What is checked | Why it matters |
|---|---|---|
| TL;DR in first 150 words | Does the page open with a citable summary? | LLMs extract the first dense block of text. If it is navigation copy or a headline, the page ranks low for citation. |
| FAQ schema (FAQPage JSON-LD) | Is there a valid FAQPage schema block? | FAQ schema is the highest-leverage single element for LLM citation. Pages with it are cited 2-4x more than equivalent pages without it. |
| Article schema | For blog posts: is Article JSON-LD present with author and publisher? | Author credentials are a trust signal. Anonymous content is cited less by all four engines. |
| Entity density | How many times does the brand name appear in the first 300 words? | LLMs need explicit entity confirmation to associate a page with your brand. Appearing 0-1 times in the opening block is a common miss. |
| Numbered or bulleted lists | Are there at least 2 structured lists on the page? | Structured content is cited 3-4x more than prose-only pages in LLMRadar scan data. |
| Internal linking depth | Does the page link to at least 2 sibling pages or the product page? | Internal links signal domain depth to LLMs. Isolated posts score lower. |
| llms.txt present at root | Does yourdomain.com/llms.txt return a valid file? | This is the fastest structural fix available. Most sites still do not have it. |
Each rule gets a pass/fail score for each page checked. The PDF shows you which pages fail which rules, not just a single aggregate score. For a deeper explanation of why each of these rules matters, see the full post on 7 page-structure rules for LLM citation.
What the FAQ schema check reveals most often
In audits to date, the FAQ schema check is the highest-impact finding for the majority of SaaS sites. Most SaaS marketing teams know FAQ content matters for Google. Very few have implemented FAQPage JSON-LD schema. There is a large gap between "we have an FAQ section on the page" and "we have structured FAQ schema that LLMs can parse."
The pattern we see repeatedly:
- Site has a visible FAQ section with real Q&A content, well-written for humans
- Page has zero FAQPage JSON-LD schema
- LLM cannot extract the Q&A structure because it is in unstructured HTML, not schema
- Brand citation rate for comparison and problem queries is near zero despite strong content
Adding FAQPage schema to existing FAQ sections is an afternoon of dev time in most stacks. The citation rate change is often measurable within 30 days.
What the brand entity score measures
The entity score is the part of the audit that surprises SaaS founders most. It tests whether the four LLMs have a consistent, accurate, positive understanding of what your brand is.
This matters because AI models do not start with a blank slate when a buyer asks about you. They have a prior understanding built from training data, retrieval augmentation, and previous queries. If that prior understanding is weak (because your brand is young, your site is thin, or your description has changed over time), the model will either skip you or describe you inaccurately.
The entity query run tells us three things:
- Coverage: Does the LLM know your brand exists at all?
- Accuracy: Does it describe you correctly when asked directly?
- Positive framing: Does it describe you in a way that would help a buyer evaluate you, or in a way that would confuse them?
A brand with a citation rate of 40% but poor entity accuracy has a different problem than a brand with 0% citation rate. Both show up in the audit results. The fix for each is different.
What you get in the PDF
The report is structured around action, not just data. Here is the section breakdown:
Section 1: Score dashboard. Your citation rate per engine, your page structure score (out of 7 per page), and your entity accuracy score. These are your baselines on a single page.
Section 2: Per-engine citation detail. For each of the 4 LLMs: which query categories you appeared in, which you were skipped in, and what the model actually said when it cited you (or what it said instead when it skipped you). You can read the exact model outputs.
Section 3: Page-by-page structure report. For each of your 5 audited pages: which of the 7 rules pass and which fail, with the specific fix for each fail. Each fix is written in plain language with the exact implementation step.
Section 4: Priority order. The three fixes with the highest expected citation-rate impact for your specific site, ranked. Not a generic checklist. Based on what your actual audit data showed.
Section 5: 30-day check-in brief. What to re-measure after implementing the top 3 fixes and how to interpret the change.
Frequently asked questions about the LLMRadar Brand Audit
How long does the audit take?
The query run and structure analysis take 3-4 hours of compute time. You submit your brand name, URL, and email. The PDF lands in your inbox the same day, typically within 4 hours of submission.
Do I need to provide API keys or access to my site?
No. The LLMRadar Audit runs all queries from OperatorIQ systems using our own API access to Claude, ChatGPT, Perplexity, and Gemini. We check your page structure via HTTP requests to your public URLs. You submit a form. That is it.
What if my site changes after I submit?
The audit is a point-in-time snapshot. It captures your scores as of the day it runs. That is actually useful: it gives you a clean before-state to compare against when you re-audit after making fixes.
What if my citation rate is already high?
The audit still tells you which engines cite you consistently versus inconsistently, and which pages have structural gaps that could cause your citation rate to drop when models update. High citation rates can erode quickly when a new model version weights different signals. Knowing where your pages have gaps before a model update lets you close them proactively.
How is this different from the free audit at operatoriq.io/audit?
The free audit runs 5 checks on your domain and returns a basic gap summary. It is useful for diagnosing whether you have a problem. The $197 LLMRadar Brand Audit runs 40 queries across 4 live LLMs, scores your pages against all 7 structural rules, and gives you a prioritized fix list based on your actual citation data. Different depth, different output.
Can I buy the audit for a competitor's domain?
Yes. Some SaaS marketing leads use the audit to understand how their top competitor shows up in LLM responses versus how they do. The form accepts any public URL.
How to read your results when you get the report
A few guidelines before you get into the PDF:
A citation rate below 20% on any engine means that engine is effectively skipping you. Start with your fixes there.
A citation rate between 20% and 60% means you have a partial signal. You are in the citation pool, but inconsistently. The page-structure fixes in the report are most relevant here.
A citation rate above 60% means you have strong structural signals. The report will focus on consistency, entity accuracy, and preventing erosion when model versions update.
Do not compare your scores across engines in isolation. ChatGPT and Perplexity weight different signals than Claude and Gemini. A 40% rate on Claude and a 15% rate on ChatGPT does not mean Claude is better at finding you. It means you have different structural gaps relative to each engine's weighting. The report explains the difference.
To understand the broader context of where your site gaps fit in the AI citation picture, the 5 signs your SaaS is invisible to AI search post covers the full diagnostic framework the audit is built on.
Next post: How LLM citation compounds over time and why brands that fix these gaps in Q2 2026 have a structural advantage that is hard to close by Q4. Coming next Thursday.
Christine Johnson | Founder, OperatorIQ. AI visibility audits on 80+ SaaS brands since January 2026. Questions: christine@operatoriq.io