How to Audit Your Brand AI Visibility Step by Step

Why audit AI visibility?

Conversational AI systems — ChatGPT, Claude, Gemini, Perplexity — now mediate a growing share of how customers discover brands. When someone asks "What's the best [your product]?" or "Who offers [your service]?" on any of these platforms, the answer determines whether your brand is considered or ignored.

Unlike traditional search engines, there is no "Google Search Console for ChatGPT". You cannot log into a dashboard and see how often your brand appears in AI responses. An audit is the only way to establish a baseline — a T0 measurement — against which every optimisation effort can be measured.

This step-by-step guide explains how to run your first AI visibility audit, what metrics to track, and how to interpret the results. It is designed for brand owners, marketers, and agency professionals who need a repeatable methodology.

Step 1: Define your scope

Before you query any AI, decide what you are measuring. A well-defined scope saves hours of analysis and produces actionable data.

Select the AI models

Not all AI engines are equal. The most relevant for B2B and B2C brand visibility are:

• ChatGPT (OpenAI) — the most widely used AI chatbot globally. Being cited here matters because it has the largest user base.
• Claude (Anthropic) — preferred for professional and analytical queries. Strong in Europe.
• Gemini (Google) — deeply integrated with Google's ecosystem. Its answers increasingly appear in Google's AI Overviews.
• Perplexity — positioned as an answer engine with explicit source citations. The closest thing to a "SERP for AI".
• Mistral (Le Chat) — European alternative, growing fast in French and German markets.

For a first audit, 4 to 6 models provide sufficient coverage. Running one query against 50+ models is possible with dedicated tools, but a manual first audit should focus on the most impactful engines.

Define your market and language

AI responses vary significantly by language. A brand visible in English on ChatGPT may be completely absent in German on the same platform. Decide which markets matter first, and audit each language separately.

Choose your benchmark queries

The quality of an audit depends on the relevance of the prompts tested. Three categories to cover:

• Category queries: "best [your product category]" or "top [your industry] providers" — these test whether you are considered a relevant player.
• Problem queries: "how to solve [your customer's problem]" — these test whether your brand surfaces when someone looks for a solution you offer.
• Comparison queries: "[competitor A] vs [competitor B]" — these reveal where the AI positions you relative to alternatives.

Aim for at least 30 queries spread across these categories for a meaningful sample.

Step 2: Build your prompt list

Prepare a list of prompts that reflects how real customers search. Avoid overly brand-specific prompts (e.g. "What do you know about [your brand]?") for the baseline — these inflate your mention rate artificially. The baseline should measure spontaneous visibility: does the AI cite your brand when it is not prompted to do so?

Example prompts for a hypothetical logistics company:

• "Which companies offer international freight forwarding in Europe?"
• "Best logistics providers for e-commerce in Germany"
• "How to reduce shipping costs for cross-border B2B deliveries"
• "Compare supply chain solutions for mid-size manufacturers"

Create a spreadsheet with one prompt per row. Add columns for each AI model you plan to test.

Step 3: Collect the data

There are two ways to run the queries:

Manual collection

Open each AI platform in a separate browser tab or window. Paste each prompt, wait for the full response, and log the results. For each prompt, record:

• Was your brand mentioned? (Yes/No)
• If mentioned, where in the response? (First, second, third, etc.)
• If not mentioned, which brands were cited instead?
• Was the mention positive, neutral, or negative?
• Did the AI cite a specific source (URL, article)?

Manual collection is time-consuming but instructive — you see firsthand how each model interprets your market.

Automated collection

Platforms like AI Labs Audit automate the entire process. They send your prompts to multiple AI models simultaneously, record every response, and compute standardised metrics. This shifts your effort from data collection to analysis — which is where the real value lies.

Regardless of the method, ensure you run all prompts in a consistent session to avoid day-to-day variation in model behaviour (models are updated frequently).

Step 4: Calculate your metrics

Four metrics structure a complete AI visibility audit:

Mention rate

The percentage of queries where your brand is explicitly cited. Example: if your brand appears in 12 out of 30 prompts, your mention rate is 40%. A rate under 10% indicates very low visibility; 60% or above suggests a dominant position.

Citation position

Not all mentions are equal. Being the first provider listed carries more weight than a mention at the bottom of a list. Track the average position when your brand is cited.

AI Share of Voice (SoV)

Your brand's mentions as a percentage of total industry mentions in the audit. If three competitors are cited 30% of the time each and your brand is at 10%, your SoV is 10%. This contextualises your performance relative to the market.

Mention sentiment

Is the brand described positively, neutrally, or negatively? An AI that says "X has faced criticism for customer service" is mentioning the brand — but not in a way that helps.

Step 5: Identify gaps and competitors

For every prompt where your brand was not mentioned, note which competitors were. This reveals:

• Who occupies the position you want to capture
• Which types of queries they win on (category, problem, comparison)
• Which AI models give them preferential visibility

A thorough audit also checks for hallucinated information: URLs the AI invents for your brand, fabricated product claims, or fake certifications. These present a brand safety risk that should be addressed.

Step 6: Analyse technical readiness

AI models retrieve information from your website when they need to verify facts or cite sources. Several technical factors determine whether your site is "AI-ready":

• Structured data (Schema.org): Does your site use the right schema types (Organization, Product, FAQ, Article) so AI models can extract entity information?
• Robots.txt and llms.txt: Are AI crawlers allowed to access your key pages? Do you have an llms.txt file that guides AI models to your most important content?
• Content freshness: AI models prefer recently updated content. Pages that haven't been touched in 12+ months are less likely to be cited.
• Internal linking: Can AI crawlers discover your important pages through a clear internal architecture?
• Multi-language handling: Does your site serve the correct language to AI crawlers based on your target market?

A 38-point technical checklist (such as the one used by AI Labs Audit) covers these factors and more, providing a scored assessment of your site's AI readiness.

Step 7: Build an action plan

With the data collected and analysed, prioritise your actions:

1. Fix critical gaps first: If you are invisible on a major AI platform, focus on getting cited there before optimising positions.
2. Address brand safety issues: Hallucinated URLs and fabricated claims should be corrected immediately through content updates and structured data.
3. Improve technical readiness: Schema markup, robots.txt, and content freshness are foundational — invest here before advanced tactics.
4. Create citable content: Publish authoritative, well-structured articles and guides that AI models are likely to reference. FAQ pages using the AQA standard are particularly effective.
5. Monitor continuously: AI visibility is not a one-and-done metric. Schedule regular audits to track progress and respond to model changes.

An initial baseline followed by monthly monitoring provides enough data density to separate genuine trends from run-to-run model variability.

Recommended tools

For those who want to automate beyond manual spreadsheets, specialised GEO/AEO platforms provide multi-model audit capabilities with standardised scoring. AI Labs Audit offers automated audits across 50+ AI models with 6-dimension scoring (the AGS methodology), AI bot tracking, and white-label reporting for agencies. The Discover plan is free and includes 100 credits per month, enough to run a baseline audit on a single brand.

Other options include Profound (enterprise-focused, US-based) and Scrunch (infrastructure-level agent experience platform). Each platform has different strengths; the choice depends on your market, budget, and technical requirements.