AI Visibility Audit™
When customers ask AI who to buy from, does it say your name?
We test your business the way real prospects search — live buyer questions across ChatGPT, Gemini and Perplexity — then tell you exactly what’s helping, what’s hurting and what to fix first.
100
points, every one backed by evidence
50
objective tests across 10 categories
4
AI platforms tested with real buyer prompts
The 100-point framework
Ten categories. Fifty tests. One score.
Every category is worth 10 points, built from five objective tests worth 2 points each. Same rules, every client, every industry.
AI Answer Visibility
Can AI identify, describe, cite and recommend you accurately?
Brand & Entity Clarity
One consistent, verifiable company identity everywhere.
Technical SEO & Metadata
Titles, descriptions, headings, indexability and internal links.
Structured Data & Machine Readability
Your facts in valid schema AI can parse cleanly.
Content Quality & Authority
Specific, original, current and answer-ready content.
Local Presence & NAP Consistency
Name, address, phone and hours agree across every source.
Reputation & Third-Party Authority
Reviews, mentions, credentials and real case studies.
UX, Accessibility & Conversion
Can visitors understand the offer and act, on any device?
Competitive Positioning
You against three real competitors on message, proof and presence.
Crawlability, Governance & Measurement
AI crawler access, sitemaps, speed and conversion tracking.
90–100
Leading
75–89
Strong
60–74
Needs improvement
40–59
At risk
0–39
Critical
Evidence for every point
Each test is scored only from saved evidence — a URL, screenshot, markup or validator result. Nothing is guessed.
Reproducible AI tests
Every prompt is date-stamped with the platform, exact wording, answer and citations, so we can re-run it later.
Critical flags, called out
Serious risks — a blocked site, wrong contact details, broken forms — are flagged separately, even when the score looks fine.
How it works
We ask AI what your customers ask.
A panel of real shopper questions, built from your services, your towns and what people actually search. Then we record who AI names — and whether the facts are right.
| Customer question | You named? | Who showed up instead | Accuracy |
|---|---|---|---|
| ★ Where should I buy a used car in town? | No | Four competing dealers | Not cited |
| ★ Any lease deals near me this month? | No | National deal websites | Not cited |
| What are their hours on Saturday? | Yes | Directory listing with the old name | Outdated |
| ¿Hay alguien que hable español? | Yes | Brand’s Spanish dealer page | Accurate |
Sample rows from a real audit panel. ★ = closest to a purchase.
Then we hand you the fix
{"@type": "Question",
"name": "Where should I buy a used car in town?",
"acceptedAnswer": { ... }}
Answers written in your customers’ own words, with the code that tells AI to use them.
And we grade every answer
3×
Every prompt runs at least three times, in fresh sessions, with the agreement rate reported.
15–50
Prompts per audit, built from a matrix of intent, category and location.
Logged
Platform, model version, browsing setting and date recorded for every run.
Stranger
Tests run logged out, with and without location, including adversarial prompts.
One report. Every answer AI gets wrong about you.
01
Executive Scorecard
Your score out of 100, every category score, critical flags and the five priority actions.
02
Name, Address & Phone Check
Your details side by side across the directories and listings AI reads.
03
Customer Prompt Visibility
The real questions your customers ask AI — and who it recommends instead.
04
Detailed Findings
Every issue ranked Critical to Pass, with the business impact and paste-ready code.
05
Priority Checklist
Week 1, Week 2, Month 1 — so your team knows exactly what to do first.
06
Summary & Methodology
The bottom line in plain English, plus exactly how we tested.
4 days
from kickoff to your finished report
Sample audit result
Here’s what a report looks like. Before and after.
Day one, this business scored 47 out of 100: needs major work. After its team shipped the fixes in the report, the re-test scored 77.
BEFORE · NEEDS MAJOR WORK
AFTER FIXES
FINDINGS AT BASELINE
4
Critical
5
Warnings
3
Info
2
Pass
CATEGORY BREAKDOWN · OUT OF 100
Before After
Illustrative example of an Amped Up AI report and re-test. Results vary by business and are not guaranteed.
Revenue Impact Report
Not every issue deserves the same urgency.
We translate technical findings into business priorities, so your team knows which gaps affect discovery, leads, appointments and sales.
HIGH IMPACT · LOWER EFFORT
Act first
Missing schema, weak service pages, broken calls to action.
HIGH IMPACT · HIGHER EFFORT
Plan and resource
Authority content, location architecture, major UX changes.
LOWER IMPACT · LOWER EFFORT
Quick cleanup
Metadata consistency, minor FAQ and internal-link gaps.
LOWER IMPACT · HIGHER EFFORT
Defer or validate
Large changes without a clear visibility or revenue case.
Revenue estimates are kept separate from the score, with every assumption disclosed. Estimates, not guarantees.
Ongoing Intelligence
Visibility is not a one-time event.
AI answers, competitor content and buyer behavior keep changing. Monitoring shows whether you’re gaining ground — and where a new gap is opening.
Track
Mentions, citations and share of answer.
Compare
Competitor movement by prompt and market.
Alert
Material gains, losses and emerging gaps.
Re-measure
Score progress after improvements launch.
Beyond visibility
Get found. Then convert.
Visibility without conversion is a vanity metric. Every audit checks your conversion readiness — headline clarity, calls to action, proof points and visitor routing.
Ready for more? Our AI website agents engage visitors around the clock, answer questions, qualify leads, book meetings and route your hottest prospects straight to sales.
Ask about AI website agentsIllustrative conversation.
Methodology FAQ.
The questions sophisticated buyers ask before trusting the numbers.
What is an AI Visibility Audit?
It measures whether AI platforms like ChatGPT, Gemini, Perplexity and Google AI Overviews mention, recommend and cite your business when customers ask buying questions — and gives you a ranked, paste-ready plan to improve.
How many prompts do you test?
Typically 15–20 for a single location and 30–50 for a multi-location business. The count comes from a matrix — intent type, how your brand and category are named, and each location in scope — so it scales predictably as you add locations or services.
How do you choose which prompts to test?
We start from real search behavior: your Search Console queries when you can share them, plus “People also ask” and related searches. We cover the whole funnel, from discovery to ready-to-buy, and deliberately include adversarial prompts designed to expose weak spots. An audit that only asks flattering questions is a highlight reel, not an audit.
AI answers change every time. How do you account for that?
Every prompt runs at least three times, each in a fresh session, and we report the agreement rate — for example, “cited the correct address in 2 of 3 runs” — instead of a single pass or fail.
Which AI platforms and models do you test?
ChatGPT, Perplexity, Gemini and Google AI Overviews. Every run is logged with the exact product, model version, whether live browsing was on, and the test date, because each of those changes the answer.
Does the AI know where the shopper is?
We test both ways: with a location in the prompt, and with none at all, which reveals what the AI assumes on its own. Tests run logged out with no account history, so you see what a stranger to your brand would see.
How do you know if an AI’s answer is right?
Every stated fact — address, phone, hours, offers — is checked against your Google Business Profile, your website and an independent listing, then graded Accurate, Outdated, Fabricated or Not cited. When an AI names a source, we also check that the source actually says it.
How is the score calculated?
Ten categories worth 10 points each, built from 50 objective tests scored 0, 1 or 2 from saved evidence. Anything we can’t verify is marked “not verifiable” rather than guessed, and critical risks are flagged separately from the score.
How often should testing be repeated?
Quarterly at minimum, right after any fix to confirm AI picked up the change — that can take weeks to months — and after a major model release or search update. AI visibility is a snapshot, not a permanent grade.
How long does it take, and do you need access?
Four days from kickoff to your finished report, with no website access needed. We re-run the same questions at day 60–90 to show what changed.
How much does it cost?
Pricing is quoted for your business and number of locations. Email info@workplacewave.com or request a demo for a quote.