AI Visibility Partners

Service · Baseline Audit · $750 · One time

What do you get in the Baseline Audit?

A written record of what ChatGPT, Perplexity, Gemini, and Google AI Overviews say when a customer in your market asks who to hire, measured under rules registered before the first prompt runs. 25 to 50 prompts, three passes per cell, three coverage measures reported separately, a stated noise floor, a competitor and source-gap map, a prioritized 90-day backlog, and a 60-minute readout. $750, one time. The fee credits toward your first sprint month if you start within 30 days. Everything in it is yours whether or not you hire us again.

By Austin Brewer, founder of AI Visibility Partners · Updated 10 September 2026

What is in the deliverable?

Eight items, delivered as one document with an appendix you can check line by line. The document is written for the owner; the appendix is written for whoever will do the work, so the backlog can go straight to your web contractor without translation. Nothing in it depends on continuing with us. The registration form and the scoring sheet are the same ones the monthly programs run on, which is what makes a later re-measurement comparable to this one.

Measurement registration

The standard form with your appendix filled in: the exact prompt strings, the engines and their default settings on the run date, your market, your competitor set, and the run dates. Signed off by both parties at the readout.

Standard form on the Method page

Scoring sheet

Every cell (one prompt on one engine), three passes each, scored named in k of 3 for each of the three measures. 100 to 200 cells for a Baseline Audit. Single-pass hits are logged and never counted as coverage.

Verbatim answers

The full text and a dated screenshot of every pass, in the appendix. You can check any score in the sheet against the answer that produced it.

Three coverage measures

Mention coverage, owned-citation coverage, and third-party-source presence, each as a share of cells, each by engine and for the panel. They are never blended into a single score.

Competitor and source-gap map

For every prompt cluster and engine, who is named and cited where you are not: the directories, the local publications, and the competing firms, with the pages the engines drew on.

Crawl and extraction checks

Whether the engines can fetch your pages, whether each page states what you do and where in a form an engine can quote, and what structured data is present or missing.

Prioritized 90-day backlog

The fixes, ranked by expected effect on the three measures and by effort, written so any competent web contractor can execute them without us.

60-minute readout

Walked through with Austin Brewer, recorded on request. The appendix is signed off here and the backlog is ordered against what your firm can actually get done.

How are the prompts chosen?

From the questions a customer asks before they ever reach a website, phrased the way a customer would ask them rather than the way a marketer would search. They fall into four clusters, and every panel carries all four so that a firm strong in one is not flattered by a panel built only from it. Your service lines, your city, and the competitors you name fill in the specifics. Once registered, the panel is fixed; prompts can be added later and are reported separately, and none is ever swapped or removed.

Who should I hire?

Named-recommendation prompts for your services and your city. These are the answers where a name turns into a call, and the ones directories hold most firmly.

"Who are the best family law attorneys in Charlotte, NC?"

What will it cost?

Budget and scope questions. Customers ask these first, and the business whose cost guide the engine quotes is the one they remember.

"What does a divorce attorney cost in Charlotte in 2026?"

How do I choose?

Comparison and process prompts, including the category questions a competitor may already own.

"Flat-fee or hourly attorney for an uncontested divorce?"

Is this firm any good?

Direct brand prompts. What an engine says when a customer types your name after a referral, and which sources it draws that on.

"Reviews of your firm name in Charlotte"

What does the scorecard look like?

Three rows, one per measure; five columns, one per engine and one for the panel. Each cell reads as a count of covered cells over cells run, so you can see the sample behind every percentage. Beneath it: how many prompts triggered an AI Overview, the noise floor set from two baseline runs, and who holds the answers you are absent from. The sample below is illustrative and contains no client data.

Scorecard · Illustrative · No client data 32 prompts · 4 engines · 128 cells · 3 passes each
Measure · cells covered (named in 2 or 3 of 3) ChatGPT Perplexity Gemini AI Overviews Panel
Mention coverage 3 of 326 of 321 of 320 of 29 8%10 of 125
Owned-citation coverage 0 of 324 of 320 of 320 of 29 3%4 of 125
Third-party-source presence 9 of 3212 of 325 of 327 of 29 26%33 of 125

AI Overviews triggered on 29 of 32 prompts in every pass; 3 prompts are reported as untriggered, not as zero.

Noise floor from two baseline runs one week apart: 2 cells per measure. Later movement of 2 cells or fewer is reported as within noise.

Holding the answers this firm is absent from: two national directories, one regional competitor, one city magazine. Named in the report.

Illustrative figures shaped like a typical first audit. No client data. Your scorecard carries your prompt strings, your competitor set, and every underlying pass in the appendix.

What does the audit not tell you?

It does not tell you how many customers asked these questions, what any one of them saw, or whether a lead came from an AI answer. It measures what a clean, signed-out session in your market is told on the run dates, and it sets the baseline a later run can be compared to. It is a photograph with a stated exposure, not a forecast. Any vendor who claims their audit removes these limits should be asked how.

  • No engine publishes query volume for conversational prompts. The panel measures presence in answers, not how often the question is asked.
  • Signed-in buyers see personalized answers this protocol cannot reproduce. The clean session is a shared floor that two measurements can share.
  • A single audit shows one point in time. Movement, and whether it is larger than noise, needs a second run under the same registration.
  • The full list of limits, and how each is stated in the report, is on the Method page.

Should you start with the Snapshot or the Baseline Audit?

The Snapshot ($495) runs 3 to 5 prompts with screenshots, a one-page directional readout, and a 30-minute call. It answers whether you have a problem. The Baseline Audit ($750) runs the full registered panel and is the only one of the two a later month can be measured against. If you are deciding whether to care, take the Snapshot. If you already intend to act, take the Baseline; the credit toward a first sprint month applies to the Baseline Audit. Both are priced flat and listed on the Pricing page.

What do firms ask before booking the audit?

What do you need from us to start?

Your firm name, website, and city, and the names of up to three firms you consider peers in your market. The prompt panel, the engine list, and the session protocol are built from that and shown to you at the readout before anything is signed off. If your site has a staging environment or a web contractor we should coordinate with, say so in the first email and we will include them in the readout.

Can we skip the readout and just get the report?

No. The readout is where the client-specific appendix to the registration is walked through and signed off, and where the backlog is ordered against what your firm can actually execute. It runs 60 minutes and is recorded on request so anyone who could not attend can watch it.

Is anything from our audit published or shared?

No. Nothing from a client audit leaves the engagement without a separate written agreement. The Pilot Sprint requires one because its discounted rate is what a public case study buys; the Baseline Audit carries no such requirement.

Ready to see your baseline?

Send your business name, website, and city. Audits run in the order they arrive, and the readout is scheduled when the write-up is done.

Email us your business name

hello@aivispartners.com · Reply within two business days