How to Read This Report
A plain-English guide to every section of your AI Search Visibility Audit.
This guide explains, in everyday language, what each part of your report means and what to do with it. Your report measures one thing: when a potential customer asks an AI tool (ChatGPT, Claude, Gemini, Perplexity, Grok) about your industry or your company, do you show up β and does the AI describe you well? The report has four chapters, then three appendices in the web page and the PDF, four in the Word and Markdown exports. Chapter one tells you where you stand, chapter two shows what each engine says, chapter three shows the evidence, and chapter four tells you exactly what to fix. The appendices hold every question, the full answers and the raw data.
1. Cover Page β "AI Search Visibility Audit"
What it is
The title page. It names your company, who prepared the report, and the date it was run.
How to read it
Note the date β AI answers change over time, so this is a snapshot of a specific moment. A score from three months ago may already have shifted.
2. What this means for you (the blue box)
What it is
The single most important paragraph in the report, your bottom line in plain English. It is written from this audit's own data: what the composite measures (answers to questions that already name you), how often the engines named you when asked open questions about your category, and whether any engine described a different company or said it did not know you.
How to read it
If you only read one thing, read this. A strong composite only reads as a strong position when the engines also name you in at least one open answer in five and every engine knows who you are; otherwise the box says which of those is missing. When the audit asked no open question, it says discovery was not measured and makes no claim about your position.
3. Weighted Composite AI Visibility Score (the big number, /100)
What it is
Your overall grade out of 100 β how visible and well-represented you are across all five AI tools combined. It's color-coded: green (67β100) = strong, amber (33β66) = opportunity to improve, red (0β32) = critical gap.
How to read it
Think of it like a credit score for AI search. It's a weighted score, not a simple average β some things (like how positively AI talks about you) count more than others. Don't fixate on the single number; the next section shows what's actually driving it.
This is not a percentage or an industry ranking. A 48 doesn't mean "better than 48% of companies." It's a measurement of your own footprint, which you can track and grow over time.
4. Dimension Breakdown β At a Glance (the five mini-scores)
What it is
Your composite score broken into its five building blocks, each shown as a small score bar: Brand Recognition (/20) β does AI know who you are and mention you; Market Position (/10) β does AI rank you well against competitors; Presence Quality (/20) β how much specific detail the engine gives about what you do (detail, not fact-checked); Brand Sentiment (/40) β how AI frames you when it names you, where negative framing costs points and a neutral, factual description keeps most of them; Share of Voice (/10) β how often you come up when someone asks about your industry generally, without naming you.
How to read it
This is where you diagnose why your score is what it is. A low score in one area drags the total by different amounts β Sentiment can swing the score by up to 40 points, while Share of Voice can only move it by 10. Focus your attention on the biggest weak spot with the most weight.
5. Overall Visibility Status
What it is
A one-sentence recap in words: your composite score, and the number of questions that name you it was measured across.
How to read it
A quick sanity-check reminder that the score comes from real questions, and a nudge to look at the dimension breakdown rather than the headline number alone.
6. Top 3 Findings
What it is
The three things that matter most this quarter, each with a short explanation. When the engines name you in fewer than one open answer in ten, or any engine describes a different company, that comes first. After it comes the dimension that costs your composite the most points (its maximum minus its score), then the schema types your kind of business actually uses.
How to read it
This is your priority list, ordered by importance. A dimension is ranked by the points it costs, so Presence Quality at 13 of 20 outranks Market Position at 6 of 10 even though its percentage is higher.
7. Misidentification (the red box β only appears if relevant)
What it is
A warning that one or more AI tools described your business as the wrong kind of company (e.g. called a security firm a "holding company").
How to read it
Take this seriously. If AI consistently puts you in the wrong category, buyers asking about your actual category will never be shown your name. This is usually one of the first things to fix, and the fixes later in the report (schema, profile descriptions) directly address it.
8. AI Readiness Scorecard (one card per AI tool)
What it is
Your /100 score broken out separately for each AI platform β Claude, ChatGPT, Perplexity, Gemini, Grok β each labeled strong, opportunity, or critical gap, with a bar chart of each engine's composite score.
How to read it
Different AI tools work differently, so you can be strong on one and invisible on another. A low score on a tool like ChatGPT or Claude usually means you're not yet well-known to that model. A low score on Perplexity or Gemini usually means a lack of third-party mentions and citations about you online. The paragraphs under each card explain what moves that specific platform fastest. Under the cards the report also says how stable each engine's answers were. Every question is asked three times on every engine, and a question earns full marks when all the answers agree on whether your brand was named, half marks when all but one do, and none otherwise. An engine is labeled Decision-grade when those marks average 75% or better across at least 6 questions. Below that it is labeled Directional, so treat the number as a direction to check. Agreement measures consistency, not visibility: an engine that never names you three times out of three is perfectly consistent and scores 0 on recognition. Audits run before this label existed show none.
9. Dimension Breakdown (the full version)
What it is
The same five dimensions from the at-a-glance strip, but explained in full β each with its tier (strong, opportunity, or critical) and what the number means.
How to read it
Use this when you want the detail behind a specific dimension. The tier labels use the same color scale as the overall score, so you can spot at a glance which building blocks are healthy and which need work.
10. Platform-by-Platform Findings
What it is
For each AI tool, one card: its five dimension scores, its sentiment, citations and stability, the levers that move that engine fastest, and an excerpt of its best answer to a question that names you. The chapter opens with a chart of how many scored questions each engine named you on. The full answer prints once, in Appendix B, and the card points there. The questions that did not name you are in Appendix A. Citations reads "none returned (answered from model memory)" when the engine gave no sources, and a line explains a Mixed sentiment that comes from answers saying the engine did not know you.
How to read it
This is the actionable version of the scorecard. It tells you, tool by tool, exactly which dimensions are pulling you down so you can see whether a problem is everywhere or isolated to one platform. Read the excerpt to see how the engine talks about you, then turn to Appendix B for the whole answer.
11. How You Compare
What it is
What the AI tools say about the competitors you named, in their own words. Each competitor card prints an excerpt, and when the answer carries a table, it prints under the excerpt. The full answer prints in Appendix C, or in Appendix B when that appendix already quoted the same answer, and the card says which.
How to read it
This shows you how AI frames the alternatives a buyer is weighing against you. Look for where competitors are described as more established or more visible; that gap is what the citation and profile fixes are designed to close.
12. Who [Engine] Cites
What it is
A list of the websites the AI engine pulled from when answering questions in your space, and how many times it used each one. (Citation-heavy engines like Perplexity surface this most.)
How to read it
These are the sources AI trusts in your industry. If your own site is cited often, that's good. The third-party sites listed (review platforms, LinkedIn, industry publications) are the places worth getting mentioned β being present on the sources AI already reads is one of the fastest ways to get surfaced.
13. Unprompted Visibility β Who the AI Names in Your Category
What it is
When someone asks a general question about your industry without naming you, how often do you come up β versus your competitors and the big players? Includes a leaderboard of who the AI names most, drawn as a bar chart.
How to read it
This is the hardest and most valuable kind of visibility, because it's how new buyers discover you. A low percentage here (even 0%) is common and fixable. The leaderboard shows who currently owns the conversation in your space β that's the position you're working toward.
14. Technical Gap Analysis
What it is
A look under the hood of your website, in three parts: Schema Markup β invisible code that spells out for AI who you are, what you offer, and your FAQs; E-E-A-T Signals β signals of Expertise, Experience, Authority, and Trust (e.g. named authors, topic depth); Citation Ecosystem β whether you're listed on the review and directory sites AI checks (G2 for software, service and company sites, Clutch for service firms). The schema rows follow your kind of business, so a publication is asked for NewsMediaOrganization and Article, software for SoftwareApplication, a local business for LocalBusiness and a service firm for Service. Each item is marked Present or Missing, with a priority. A Site issues section follows with the broken basics the crawl found on your own site, such as a navigation link that leads to a missing page or a robots.txt that answers with a web page.
How to read it
This is the technical checklist your web developer or marketing team can act on. "Missing + High priority" items are the ones to tackle first. You don't need to understand the code β the report's fix-it section (below) generates most of it for you.
15. ROI-Ranked Implementation Roadmap (Weeks 1β4, Weeks 5β8, Weeks 9β12)
What it is
Your to-do list, sequenced by payoff and effort into three four-week windows: Weeks 1β4 for the foundations (crawler access, schema), Weeks 5β8 for content (FAQs, headings, blog titles) and Weeks 9β12 for earning mentions on the pages AI cites and on review platforms. Each item shows expected impact, effort and its Timeline, the time until the fix is live (the same figure Do This Before Lunch prints). When the audit generated nothing for Weeks 1β4, that window says so and points at the quick wins.
How to read it
This is your action plan. Start with the first window that has items and work down: inside each window, the fix that pays back most comes first, in this order: engine access, then citations, schema, person and author pages, earned content, case study and reviews. Items note any prerequisite (like creating a free profile account first). You don't have to do everything on the list.
16. Ready-to-Deploy Fix-It Artifacts
What it is
The actual fixes, pre-written for you β not advice, but copy-paste-ready content β grouped by where the work happens: On Your Site (code and page content your developer can paste in today, e.g. buyer-question headings, schema); Third-Party Profiles (ready-made write-ups for the review and directory sites the AI engines cite for the client's questions, and listings to complete where the client already has one); Off Your Site (outreach material β a list of places to get cited, a guest-article pitch, and a customer-review-request email).
How to read it
This is what makes the report actionable rather than theoretical. Each artifact is labeled with its priority and where to deploy it. Hand the "On Your Site" items to whoever manages your website; the "Third-Party" and "Off Your Site" items are marketing tasks. Everything is drafted β your job is mostly to review, personalize the bracketed placeholders, and publish.
17. Methodology & Confidence
What it is
How the report was made β how many questions were asked, across which tools β and an honest note on how much statistical weight to give the numbers.
How to read it
Read this to understand what the scores are and aren't. Key point: every score comes from real AI answers, not guesses. But because it's a sample of questions (not thousands), treat the individual dimension scores as a strong directional read, not a precise statistic. The value is in the trend over time and the specific fixes β not in defending any single decimal point.
18. Appendix A: Every question, by engine
What it is
A grid of every buyer question we tested, once, with one mark per AI tool. β means the engine named you, Β· means it answered without naming you, and β means no usable answer from that engine.
How to read it
This is the raw ground truth behind your Unprompted Visibility score. Scan for patterns: check-marks tend to cluster on questions that name you (brand and comparison questions), while the general "who are the best providers in my category" questions, the ones that win new customers, are often blank. Those blanks are your biggest growth opportunity.
19. Appendix B: What AI actually said, in full
What it is
The full text of each AI tool's showcased answer, with the question asked and the scores for that answer. The cards in chapter two print an excerpt and point here. An answer about a different company is shown only when no other exists, with a caption.
How to read it
This is the proof behind the scores. It's often the most eye-opening section, because you see your company described in the AI's own words. Read it to catch inaccuracies, outdated claims, or wrong facts that you'll want to correct through the fixes later in the report.
20. Appendix C: Full competitor answers
What it is
The complete answers behind the How You Compare excerpts, for every competitor you named, with the question asked and the engine that answered. Each competitor answer prints here once, in full, and repeats the caption from its card. When Appendix B already quoted the same answer, this appendix prints a one-line reference to it instead.
How to read it
Open this when an excerpt makes you want the whole picture. Look for where an engine describes a competitor as more established than you, and for the phrases it uses to do it. Those phrases are what your profile copy and citations need to answer.
21. Appendix D: Raw data
What it is
One row per AI call: the engine, the model version, the prompt template version, the citation and mention counts, the latency, and whether the call was a resample. It prints in the Word and Markdown versions only, and the web page and the PDF leave it out. On a very large audit the table stops at a fixed number of rows and says how many it left out.
How to read it
Use it when you want to see how the audit ran: which model version answered, how long each call took, and which calls were repeat samples. Hand it to an analyst who wants to check the run behind the scores.
Tip: The best way to use this report is to read chapter one to understand where you stand, skim chapters two and three to see the evidence, and then live in chapter four, the roadmap and the fix-it section. That is where the actual improvement happens.