AI Visibility Audit

Are you cited where your buyers ask?

A measured baseline of where you appear, where you do not, and which competitors are being cited instead — across ChatGPT, Claude, Perplexity, Google AI Overviews, and Microsoft Copilot. The first honest look before any AI search work begins.

Models & signals measured · ChatGPT · Claude · Perplexity · Google AI Overviews · Microsoft Copilot · Citation Mapping · Entity Clarity · Visibility Score
01 — What’s Included

Six measured passes.
One honest baseline.

An AI visibility audit is not a tool printout. It is a structured, manual examination of how five different AI models respond to the questions your buyers actually ask — and whether your business shows up, gets credited correctly, or never enters the conversation at all.

Every audit is delivered as a written document with the raw queries, the model responses, the competitor citations, and a prioritised remediation plan you keep regardless of whether you continue with us.

N° 01

Query Set Design

Buyer-true

We do not query the models with marketing-flavoured keywords. We query them with the questions your prospective customers actually type — extracted from your sales conversations, support tickets, search-console data, and the buyer-intent phrases competing in your space. A well-designed query set typically runs forty to one hundred and twenty questions across informational, comparative, and decision-stage intents. Without this groundwork, an audit measures nothing meaningful — it just confirms that your brand name returns your brand name.

N° 02

Per-Model Citation Mapping

Five models

Each query is run against ChatGPT (with web browsing enabled), Claude, Perplexity, Google AI Overviews, and Microsoft Copilot. We capture every brand mention, every cited URL, every competitor referenced, and the exact phrasing each model uses to describe each business. The same query asked of five different models will return five different answers — a brand cited by Perplexity may be invisible to Claude, and a name surfaced in AI Overviews may not appear when ChatGPT browses the live web. Mapping each model independently is the only way to see the picture clearly.

N° 03

Entity-Clarity Inspection

Foundation

Before a model can cite a business, it has to recognise the business as a coherent entity. We check whether the major models hold a consistent picture of what your company is, what it does, where it operates, and how it relates to other entities in your space. We compare your declared schema, your Wikipedia and Wikidata presence, your Google Business Profile, and your major-directory listings against the entity profile each model has actually formed. Discrepancies here are the single most common reason a credible business goes uncited.

N° 04

Source-of-Citation Analysis

Where it pulls from

When a model does cite you, we trace which page on your domain it pulled the answer from, and which third-party source — a directory, a review platform, an industry publication, a forum thread — it referenced for context. This tells us which content is already earning its keep, which pages are underperforming despite covering the right topic, and which external sources hold disproportionate weight in shaping the model’s view of your business. The output is a map of citation supply, not just citation demand.

N° 05

Competitor Share-of-Voice

Comparative

Visibility is relative. We benchmark your citation footprint against the three to five competitors who matter most in your category — the ones models actually surface when buyers ask. We measure how often each competitor appears, the sentiment and accuracy of those citations, and the gap between their share-of-voice and yours, query by query. The benchmark answers the question owners actually want answered: not “are we visible?” but “are we more visible than the competitor we are losing deals to?”

N° 06

Visibility Score & Remediation Plan

Deliverable

The findings are condensed into a single Visibility Score per model — a normalised measure of how often your business appears, how accurately it is described, and how it ranks relative to direct competitors. Alongside the scores, we deliver a prioritised remediation plan: the small number of high-impact fixes likely to move citations within the first ninety days, and the longer-horizon authority work that compounds beyond that. The plan stands on its own — your team can execute it, your existing agency can execute it, or we can. See the full AI Search Optimisation service →

02 — Our Approach

Manual. Repeatable.
Read by humans.

An AI visibility audit demands a different working method to a traditional SEO audit. The data sources are conversational, the answers vary across runs, and the underlying systems are still being defined. The way we manage that uncertainty is to do the work by hand, document everything, and refuse the temptation to dress up imprecise data as a confident dashboard.

i

Real queries, run by hand

We do not rely on third-party tools that claim to “monitor LLM citations” — most are scraping cached snapshots that lag the live model behaviour by weeks. We run every query by hand, in the actual model interface, with the appropriate browsing or grounding mode enabled, on the day of the audit. The result is data you can trust because we can show you exactly how it was gathered.

ii

Multiple runs for stability

AI models are non-deterministic — the same query can return different citations on different runs. We run every query at least three times across separate sessions and report the stable signal, not the single-run noise. Where citations vary materially between runs, we surface that volatility rather than averaging it away — knowing your visibility is unstable is more useful than a falsely confident number.

iii

Competitor benchmarking, not vanity

An audit that only measures your own citations misses the point. The question is whether you are winning the citation versus the businesses your buyers will choose between. Every query in our set is scored not just for whether you appear, but for whether you appear ahead of, alongside, or instead of the competitors who actually compete with you for the same deal.

iv

A document, not a dashboard

The deliverable is a written audit document with the raw query log, the captured responses, the competitor breakdown, the entity-clarity findings, and the prioritised remediation plan. It is something you can hand to a board, a CMO, or an internal team and have it stand on its own months later. Dashboards rot when the underlying data shifts; a written audit holds its shape and tells the truth about a moment in time.

03 — Who It’s For

Businesses that need
a starting point.

The audit is the right first step in two situations: when you suspect you are missing from AI answers and want hard evidence before investing, and when you have already started AI search work but cannot tell whether it is moving the dial. In both cases, the audit gives the conversation a measured baseline.

A few recurring reasons businesses commission an audit.

  • i You suspect you are missing from AI answersYou have asked ChatGPT or Perplexity a question your customers ask, watched a competitor get cited, and wondered why your name was not in the answer. The audit converts that anecdote into a measured query set with a number you can act on.
  • ii You are scoping an AI search investment and need a baselineBefore committing to a multi-month engagement, you want a clear-eyed picture of where you stand today. The audit becomes the zero-state benchmark against which every subsequent quarter of work is measured.
  • iii You have started AI search work and cannot tell if it is workingYou have invested in schema, content, or third-party citations and want an independent measurement of the result. We run the audit cleanly against the current state of the models — no internal-team bias, no agency-of-record conflict.
  • iv You are entering a new market or categoryA pre-launch audit shows whether the models recognise you in the new territory at all, what the existing competitive citation landscape looks like, and where the gaps are most easily closed in the first six months.
  • v You are reporting to a board or executive teamA written audit with a Visibility Score, competitor benchmarks, and a remediation plan is something a non-technical audience can read and act on. It turns a fuzzy “we should do something about AI” agenda item into a measured business case.

The audit is also the natural pre-engagement step before any of the active AI optimisation disciplines — citation building, schema work, per-LLM optimisation. We will not start fixing what we have not measured, because the measurement itself is what tells us where the leverage is. If you are uncertain whether AI search is yet a meaningful channel for your business, the audit is the cheapest, fastest way to know for sure.

04 — A complimentary report

Curious how Google sees your site?

Send us your URL. We’ll send back a Premium SEO Report, prepared by hand, within 48 hours — domain authority, keyword rankings, backlinks, competitor gap, and the quick wins worth chasing first. The same report doubles as the foundational layer for AI search visibility, because the entity clarity, structured data, and authority signals search engines reward are also what AI models read.

No sales call required.

You cannot improve what you have not measured. The audit is the new baseline for search visibility — and the honest first sentence of every conversation about AI.
— The Aureole Practice —
05 — Frequently Asked

Questions we get
about the audit.

If a question is missing here, the contact link at the foot of the page goes straight to the person who would answer it. No ticket queues, no funnels.

i How long does an AI visibility audit take to deliver?
A standard audit is delivered within one to two weeks of kickoff. Week one is query-set design, model runs, and entity-clarity inspection. The early part of week two is competitor benchmarking, source analysis, and writing up findings. Larger query sets, multi-region audits, or audits that require Chinese-language model coverage can extend to three weeks. We scope the timeline upfront based on the breadth of the query set, the number of competitors benchmarked, and the languages required — and we hold to it.
ii Can I run this audit myself with a tool I already pay for?
You can certainly run queries yourself — and we encourage owners to spot-check their own visibility regularly. What a structured audit adds is the methodology: a representative buyer-intent query set rather than a few favourite phrases, multiple runs across each model to filter signal from noise, competitor benchmarking with the same queries, entity-clarity inspection across the sources models pull from, and a written deliverable that survives the meeting it is presented in. The third-party “AI rank tracker” tools currently on the market measure something narrower than this, and most lag live model behaviour by weeks because they rely on cached scraping. Anyone selling you a fully automated AI visibility dashboard is overselling their tooling — the field is too young and too non-deterministic for that to be honest yet.
iii Which AI models do you actually query?
The standard audit covers ChatGPT (with web browsing enabled), Claude, Perplexity, Google AI Overviews, and Microsoft Copilot. These are the five surfaces where your buyers are most likely to encounter you in 2026. Where it is relevant — for example, businesses targeting the Chinese market — we extend the audit to include Doubao, Kimi, DeepSeek, and Baidu’s AI search results. New models are added to the audit set when their adoption reaches a level that meaningfully changes a buyer’s information journey, not the moment they are launched.
iv What is a Visibility Score and how is it calculated?
The Visibility Score is a normalised measure, calculated per model, that combines three things: how often your business is cited across the audit query set, how accurately the model describes you when it does cite you, and how your share-of-voice compares to your direct competitors. Each component is scored against a clearly defined rubric and weighted so that the overall number is meaningful rather than arbitrary. Critically, the score is not the deliverable — the underlying findings are. The score gives you a single comparable number to track quarter to quarter; the findings tell you why the number is what it is and what to do about it.
v Do we have to commit to ongoing AI search work afterwards?
No. The audit is intentionally structured as a self-contained engagement. Many clients commission only the audit, take the remediation plan in-house, and execute with their existing team or agency. Others use the audit findings to scope a larger AI search optimisation engagement with us. Both paths are equally welcome. We deliberately price and structure the audit so that the value is in the audit itself — not in pulling you into a longer commitment you have not asked for. The remediation plan is yours either way.
vi How often should we re-run the audit?
For most businesses, every six months is the right cadence. AI models update their training data, refresh their retrieval indices, and evolve their citation behaviour gradually but continuously — and the competitive landscape shifts as your competitors invest in their own visibility. A six-month re-audit shows movement against the previous baseline, validates that the remediation work is producing measurable gains, and surfaces new gaps before they become entrenched. Businesses in fast-moving categories — SaaS, AI tooling, e-commerce in trending verticals — sometimes prefer a quarterly cadence. Slower-moving regulated industries can typically run annually. The full AI Search Optimisation service bundles audit cadence into the retainer.
The Invitation

Find out where AI
search sees you today.

An AI visibility audit across ChatGPT, Claude, Perplexity, Google AI Overviews, and Microsoft Copilot — measured by hand, delivered as a written document, no obligation to continue. You will hear back from the team that does the work, not a sales department.

Mon–Fri · 9–6 PT support@aureoleintelligence.com Reply within 1 business day