A meaningful AI visibility audit requires no paid tools and takes approximately 90 minutes. The process involves building a structured query set from buyer language, running those queries across ChatGPT, Perplexity, and Gemini, recording where your brand appears and where it doesn't, and mapping the gaps to specific buying factors. This post provides the complete methodology — the same approach Jeevan AI uses as the foundation for its automated scans, made available here as a free starting point for any brand that wants to understand its AI visibility before investing in a strategy to improve it.
Before publishing a single piece of GEO content, you need to know what you are actually fixing. The most common GEO mistake is publishing content based on assumptions — "we probably need more use case content" — without knowing which specific buying queries your brand is missing, which competitor is appearing instead, and what AI currently says about your brand when it does mention you.
The audit in this post takes 60–90 minutes and costs nothing. It gives you five specific outputs: a category visibility rate, a competitive gap map, a brand description accuracy score, a factor-level gap hypothesis, and a prioritised content action list. That is enough information to make the first three content investments with confidence.
The Princeton GEO research published at KDD 2024 established that structured GEO tactics can boost AI search visibility by up to 40% — but only when applied to the right gaps. An audit identifies which gaps those are. Without it, GEO content investment is largely directional guesswork.
Why a Manual Audit Before Any Tool
Paid AI visibility tools provide scale and automation — they run hundreds of queries and track citation rates over time. But the manual audit serves a different purpose: it forces the practitioner to think carefully about which queries actually represent buying intent in their category, and it reveals platform-specific dynamics that aggregate scores can obscure. A brand appearing prominently on Gemini but absent on Perplexity has a very different problem to one absent on both — and only a manual, query-by-query review surfaces that distinction clearly.
Each AI platform retrieves content differently. ChatGPT with web browsing enabled retrieves via Bing's index, so your Bing presence matters directly. Perplexity draws from live web search and shows inline citations, making it the most transparent platform for source analysis. Gemini inherits from Google's index and Knowledge Graph, meaning entity signals and structured data matter disproportionately for Gemini performance.
Running the same queries manually across all three platforms in one session — before any optimisation — establishes the true baseline. It also takes less time than most teams expect: a focused 90-minute audit produces more actionable insight than a week of reading about GEO theory.
Step 1 — Build Your Query Set (15 Minutes)
The query set is the most important input to the audit. Queries must be phrased in buyer language — the exact words and phrases a buyer in your category types into an AI chatbot when they are evaluating options. Not your brand name. Not internal product terminology. The specific problem your buyer is trying to solve, in their words. A meaningful baseline audit requires 15–25 queries across three types: category buying queries, direct brand queries, and competitor comparison queries.
| Query type | How many | Example structure | Purpose |
|---|---|---|---|
| Category | 8–12 | "Best [category] for [buyer type + use case]" — e.g. "best AI visibility tool for a SaaS marketing team" | Shows whether your brand appears when buyers are actively evaluating options |
| Brand direct | 3–5 | "What is [Brand Name]?" / "Tell me about [Brand Name]" / "Is [Brand Name] good for [use case]?" | Shows how accurately AI describes your brand — reveals incorrect or missing information |
| Competitor | 5–8 | "[Competitor] vs [your category]" / "alternatives to [Competitor]" / "best [category] that isn't [Competitor]" | Shows which queries your competitor owns and whether your brand appears as an alternative |
How to build category queries from buyer language
Category queries are the most important and the most commonly built wrong. The key is to think about the problem, not the product. Your buyers are not searching for "AI visibility platform" — they are searching for "why does ChatGPT recommend my competitor," "how do I find out what AI says about my brand," or "GEO tool for agencies." Open your Google Search Console and look at the non-brand queries your site already ranks for — these are a reliable proxy for the language your buyers use.
Build at least one query per core buyer segment. If you serve three distinct buyer types — solo brand managers, in-house marketing teams, and agencies — you need category queries specific to each, because AI will often recommend different brands for the same product category depending on the buyer context in the query.
Step 2 — Run Queries Across Three Platforms (30 Minutes)
Run every query in your set across ChatGPT, Perplexity, and Gemini in sequence. Enable web browsing on ChatGPT for category queries to see the most current results. On Perplexity, record every source citation URL alongside the brand mentions — Perplexity's inline citation format is the most transparent of the three platforms for source analysis. On Gemini, capture both the primary answer and any source or related panels. Do not run queries on mobile and desktop interchangeably — stick to desktop for consistency across the audit.
The methodology here is adapted from Passionfruit's brand visibility audit framework, which recommends running each prompt multiple times across platforms due to AI response variability. For a baseline audit, running each query once per platform is sufficient — but note that AI answers vary even for identical prompts, so any single response is directional rather than definitive. The pattern across 15–25 queries is what matters, not individual responses.
Platform-specific notes
ChatGPT: Use the default model with web browsing enabled. For direct brand queries, also run without browsing enabled once — this shows what the model's training data alone contains about your brand, which is often revealing.
Perplexity: Use the default model. Record every source URL cited — this tells you which third-party platforms are feeding AI's description of your category and competitors.
Gemini: Use the consumer app. Capture both the primary response text and the source panel. Gemini pulls heavily from Google's Knowledge Graph, so inconsistencies between what Gemini says and what your structured data says are significant.
Step 3 — Record Results With This Template (15 Minutes)
Recording should be fast and structured. For each query, record five data points: which platform, whether your brand appeared, in what position (first mention, second, not at all), which competitors appeared, and a brief note on what was said about your brand when it did appear. A simple spreadsheet with these five columns, one row per query-platform combination, gives you a complete dataset for analysis in the next step.
| Query | Platform | Brand appeared? | Position | Competitors mentioned | Brand description note |
|---|---|---|---|---|---|
| "best [category] for [use case 1]" | ChatGPT | Yes / No | 1st / 2nd / 3rd / Not mentioned | [Competitor A], [Competitor B] | Correct / Incomplete / Incorrect |
| "best [category] for [use case 1]" | Perplexity | Yes / No | 1st / 2nd / 3rd / Not mentioned | [Competitor A], [Competitor C] | Correct / Incomplete / Incorrect |
| "What is [Brand Name]?" | Gemini | Yes | Primary subject | None | Note any factual errors or gaps |
This recording takes about 15 minutes for a 25-query audit once you are in a rhythm. The output is a raw dataset from which everything else in the audit follows.
Step 4 — Analyse Your Results: Four Outputs (15 Minutes)
Four numbers and one qualitative observation come from the recording: a category visibility rate, a competitive gap map, a brand description accuracy rate, a cross-platform consistency score, and a buying factor gap hypothesis. Together, these five outputs determine which content investment produces the greatest improvement in AI recommendation frequency for your brand specifically.
- Category visibility rate. Count the number of category buying queries where your brand appeared divided by the total number of category queries, multiplied by 100. This is your baseline citation rate. A rate below 30% is a critical gap. 30–60% is improvable. Above 60% is competitive. Compare against the industry benchmarks in our AI Visibility Score Benchmarks post.
- Competitive gap map. For every query where your brand didn't appear, note which competitor did. A competitor appearing on 8 of your 10 category queries has established citation precedence — and the content strategy needs to specifically address the buying factors that competitor owns.
- Brand description accuracy rate. For direct brand queries, count how many descriptions were factually correct, how many were incomplete, and how many contained errors. Any inaccuracy in AI brand descriptions is a content problem — the correct information is not published in a citable, structured format anywhere AI can find.
- Cross-platform consistency score. How many queries showed consistent results across all three platforms? A brand appearing on ChatGPT but not Perplexity has a recency problem — Perplexity weights newer, externally published content more heavily than ChatGPT's training data. A brand appearing on Perplexity but not Gemini likely has a Knowledge Graph or structured data gap.
- Buying factor gap hypothesis. Based on the patterns above, which buying factor is most likely causing the gap? Missing on use-case-specific queries → Use Case Fit gap. Missing on "trusted" or "reliable" queries → Trust gap. Missing on comparison queries → Quality Evidence gap. This hypothesis guides the first content investment.
Step 5 — From Manual Baseline to Automated Tracking
A manual audit is a baseline, not a monitoring system. AI citation patterns are highly volatile — AirOps research found that AI Overview content changes roughly 70% of the time for the same query, and nearly half of all citations are replaced with new sources when the answer updates. A brand that was absent in February may appear in March without any content changes, and a brand that appeared prominently in January may be absent in April. Monitoring requires a consistent query set run at regular intervals — which is exactly what Jeevan AI's automated scan and Re-Scan features provide once the manual baseline has established which queries matter most.
The manual audit gives you three things automated tools cannot provide in isolation: a carefully curated query set built from real buyer language in your specific category, a human-level interpretation of why the results look the way they do, and a buying factor gap hypothesis that connects the visibility data to a specific content investment.
Once the manual baseline is established and the first content investments are made, Jeevan AI's automated scan takes over — running the same query set across multiple platforms, tracking citation rate changes over time, and measuring the movement per buying factor as new content is published. The manual audit is where the strategy begins. The automated scans are where it is measured.
Frequently Asked Questions
How do I find out what ChatGPT says about my brand?
Open ChatGPT with web browsing enabled and run two types of queries: first, category-level buying queries using the exact problem your buyers search for (not your brand name); second, a direct brand query ("What do you know about [Brand Name]?"). Record whether your brand appears in the category queries and whether the direct brand description is accurate. Run the same queries on Perplexity and Gemini — the results often differ significantly across platforms.
Why should I use buying queries and not just search for my brand name?
Searching for your brand name tests brand awareness, not buying decision visibility. The queries that matter for revenue are the ones buyers type when they are actively evaluating options — "best [category] for [use case]" — not "tell me about [Brand Name]". If your brand doesn't appear in category buying queries, you are invisible at the stage of the buyer journey where shortlists are formed and deals are won or lost.
How many queries should a manual AI audit cover?
A meaningful baseline manual audit covers 15–25 queries: 8–12 category buying queries using buyer language, 3–5 direct brand queries, and 5–8 competitor comparison queries. Run each query on at least three platforms — ChatGPT, Perplexity, and Gemini. The full audit takes approximately 60–90 minutes and gives you enough data to identify your primary visibility gaps without any paid tools.
What should I do if AI is describing my brand incorrectly?
Incorrect AI descriptions are a content problem, not a technical one. AI systems build their model of a brand from the content available across the web. If the description is wrong, the correct description isn't clearly published in a citable, structured format anywhere AI can find. The fix is publishing explicit, entity-consistent brand description content on your own site and ensuring accuracy across major external platforms like G2, Crunchbase, and LinkedIn.
How often should I run an AI visibility audit?
A baseline manual audit should be run before any GEO content strategy begins. After publishing content changes, a re-audit at weeks 4 and 8 shows whether movement is occurring. AI citation patterns are volatile — AirOps research found that AI Overview content changes roughly 70% of the time for the same query. Monthly monitoring is sufficient for most brands; weekly monitoring is appropriate during active content investment periods.
The 90-minute manual audit in this post is the most important thing a brand can do before investing in GEO content strategy. It converts AI visibility from a vague concern into five specific, actionable numbers — and tells you exactly which buying factor gap to address first.
The brands that close AI visibility gaps fastest in 2026 are not the ones who read the most about GEO theory. They are the ones who ran the audit, identified the specific gap, published the specific content, and re-scanned 8 weeks later to confirm the movement. That loop is the entire strategy.
If the manual audit reveals gaps you want to close faster — or if you want the factor-level scoring and competitor gap analysis done automatically — Jeevan AI's free scan covers 5 queries across ChatGPT, Gemini, Perplexity, Claude, and Google AI Mode in 10 minutes, and gives you a buying factor breakdown your manual audit cannot produce at speed.
5 queries. 5 platforms. Factor-level gap breakdown. No credit card. 10 minutes.