A practical self-audit guide to know if your business is invisible to AI.
There’s a new question showing up in buyer conversations: “I asked ChatGPT who the best vendors are in this space… and your name didn’t come up.”
This is an AI visibility problem that is not directly correlated to a search ranking issue, and for most B2B companies, it’s happening right now without anyone noticing.
Search Engine Optimization (SEO) was built for a world where buyers type keywords into Google and click links. That world still exists, but it’s shrinking. A growing share of B2B research now starts with a prompt: to ChatGPT, Claude, Perplexity, Gemini, or one of the dozens of AI-powered research tools embedded in your tech stack.
If your business isn’t surfacing in those answers, you’re invisible at the moment of consideration, leaving a ton of revenue on the table.
This guide walks you through how to run a simple self-audit to find out exactly where you stand – and what to do about it.
What LLMs Actually Know (and Don’t Know)
Large Language Models (LLMs) don’t crawl the web in real time. They’re trained on a snapshot of the internet (plus ongoing retrieval in tools like Perplexity and Bing AI) and they synthesize that training into confident-sounding answers.
The result: brands with strong digital authority, consistent topical coverage, and high-quality inbound links tend to appear in AI-generated shortlists. Brands without those signals get left out – even if they’re genuinely good.
What LLMs use to form opinions about your brand:
- Content indexed before the model’s training cutoff
- Third-party mentions: review sites, analyst coverage, press, and partner pages
- Structured data and semantic clarity on your own site
- The language patterns in your category: whether your brand is associated with the right concepts and terms
- Retrieval signals used by tools like Perplexity (domain authority, freshness, citation patterns)
What they don’t care about: your H1 tags, your keyword density, or whether you’ve done traditional SEO correctly.
The Self-Audit: 5 Tests to Run Today
You don’t need a tool to start. Open four browser tabs with ChatGPT, Claude, Perplexity, and Gemini, and run these prompts. Keep a simple spreadsheet: model, prompt, appeared (Y/N), how you were described, who else appeared.
Test 1: Category Shortlist Prompt
Ask each model: “What are the best [your category] platforms for [your target buyer]?”
Examples:
- “What are the best employee learning platforms for mid-market companies?”
- “What are the top revenue intelligence tools for B2B sales teams?”
- “What CMS platforms do B2B SaaS companies use for their marketing site?”
What to look for: Are you on the list? Are your competitors? How are you described (accurately, vaguely, or not at all)?
Test 2: Problem-Aware Prompt
Ask: “How do [your buyers] typically solve [the problem you solve]?”
This simulates early-stage research – before a buyer even knows your category name. LLMs often answer with a mix of approaches and vendor mentions. If your brand doesn’t appear here, you likely don’t have enough problem-aware content indexed anywhere the models learned from.
Test 3: Direct Brand Prompt
Ask: “What do you know about [Your Company]?”
This is the most direct signal. A strong result means the model has synthesized substantive information about you. Vague or thin answers (“a software company that helps businesses…”) suggest low coverage depth. No answer or a hallucination is a red flag.
Test 4: Comparison Prompt
Ask: “How does [Your Company] compare to [Competitor A] and [Competitor B]?”
Buyers use these prompts constantly. If the model can’t generate a meaningful comparison, or only has detail on your competitors, that asymmetry will influence decisions in ways you’ll never be able to track in your own reporting.
Test 5: Use-Case Prompt
Ask: “Which vendors do most [job title] at [company type] use for [specific use case]?”
This tests persona-specific visibility. B2B buying decisions often involve multiple stakeholders in a buying committee. You may appear in one buyer prompt but not another, which tells you where your content authority is concentrated, and where it’s missing.
Scoring What You Find
After running all five tests across four models, you’ll have roughly 20 data points. Look for patterns:
You are in good shape if: You appear unprompted in category and use-case queries, you’re described accurately and with reasonable depth, and you appear in comparisons against named competitors.
You have gaps if: You appear in some models but not others, you’re described generically, or you show up only when explicitly named – not in category shortlists.
You have a problem if: You are largely absent, described inaccurately, or a competitor appears in every answer and you appear in none.
The goal is to understand where your visibility is strong, where it’s thin, and why.
Why the Gap Exists (and What Drives It)
Most B2B brands have an AI visibility gap for one of three reasons:
Thin owned content: The model has very little to synthesize because the brand has published sporadically, avoided taking positions, or focused entirely on product pages rather than substantive editorial content.
Low third-party authority: LLMs weight external mentions heavily. If your brand doesn’t appear in industry publications, analyst roundups, G2/Capterra, partner ecosystems, or credible link sources, you’re relying on self-reported information which models discount.
Semantic misalignment: Your content uses your language, not your buyer’s. LLMs understand concepts, not keywords. If your content doesn’t consistently associate your brand with the problems, outcomes, and category language your buyers use in their prompts, you won’t surface when they ask.
What This Audit Tells You
A clean AI visibility audit gives you three things:
A baseline: You can’t improve what you haven’t measured. Most brands have no idea what LLMs say about them until someone on their team accidentally discovers it.
A competitive map: You’ll see exactly which competitors are getting the AI shortlist benefit and how they’re being positioned relative to you.
A content and authority gap analysis: By looking at what models say about competitors who do appear, you can reverse-engineer what’s driving their visibility, and where your program needs to go.
Running This Audit on a Schedule
AI visibility isn’t static. Models update, retrieval layers change, new content gets indexed. A brand that runs this audit once and moves on will lose ground to competitors who treat it as an ongoing signal.
Recommended cadence: run the core prompts quarterly, and anytime you make significant changes to your positioning, publish a major content asset, or launch a new product or use case.
What to Do With the Results
If the audit reveals gaps — which it almost always does — the path forward isn’t to game the models. It’s to build the kind of brand that LLMs naturally cite: one with clear positioning, deep topical authority, and consistent external validation.
That means building content that directly answers the questions your buyers are prompting AI with. It means earning mentions in the places LLMs learn from (not just the places your SEO team has targeted) and treating your brand’s AI footprint as a strategic asset (not an afterthought).
The brands that do this work now will be the ones that show up in every buyer’s AI research session. The ones that don’t will keep wondering why their pipeline is thinning… and never find the real reason.
Tois works with companies to audit, build, and grow their AI visibility so the right buyers find you when they’re asking the right questions. Start with an AI Visibility Audit.