How to Run a Pre-Pitch AI Visibility Audit, Live on the Sales Call

1. Why a Live Audit Beats a Pitch Deck

Most GEO pitches open with the market: AI assistants are becoming a starting point for buying research, answers replace the list of blue links, brands need to show up. All of that is true, and none of it moves the person across the table, because it is about everyone. A prospect nods at a trend and still files it under "later".

A live audit is about them. You ask an AI assistant the kind of question their customers ask, in their market, while they watch, and the answer either names their brand or it does not. When it names their competitors instead, the conversation changes on its own: the prospect starts asking you questions. Why them? What are they doing differently? Can we fix it?

Doing it live matters for a second reason: trust. A screenshot in a deck can be cherry-picked, and prospects know it. An answer that appears on a shared screen in front of them is harder to argue with, and the fact that you are willing to run it live, in front of them, says something about your confidence.

2. What a Pre-Pitch Audit Should Show

Keep the audit to one screen and three questions. Anything more turns a sales moment into a technical review, and the prospect stops listening.

• Can AI systems read the website? A readiness score with the two or three biggest technical gaps in plain language: crawlers blocked in robots.txt, missing structured data, no author information. Not the full technical checklist; the gaps that matter.
• What does the assistant answer? One question a real buyer would ask, asked live, with the full answer available if the prospect wants to read it.
• Who is named instead? The brands the answer recommends. This is the line prospects remember, because competitors are personal in a way market trends are not.

The order is deliberate. The technical gaps explain why; the answer and the competitors show what it costs. Together they set up the offer: fix the foundations, then measure and improve the answers over time.

3. Before the Call: Five Minutes of Preparation

The audit is live, but the agency should never be surprised by it. Five minutes beforehand:

• Pick the market. AI assistants answer differently per country. Run the check for the market the prospect actually sells in, not where your agency sits.
• Pick the assistant. ChatGPT is the default because most buyers use it; if the prospect's audience lives somewhere else, show that one.
• Run it once privately. Answers vary between runs, so know what you are walking into. If the brand is named, the story becomes "you are visible here, but not in these other questions", which is just as useful.
• Set up your screen. Share the browser tab, not the whole window, so the prospect sees the result and nothing else: no bookmarks, no other tabs, no address bar.

4. On the Call: The Ten-Minute Script

A structure that works on most first calls, after the introductions:

1. Ask before you show. "If one of your customers asked ChatGPT for a recommendation in your category, what would they type?" Their answer gives you something to compare: the check writes its own question from their website, and reading the two side by side makes the result theirs, not yours.
2. Run the check while they watch. Enter the website and the market. The technical part takes about a minute; use it to explain what is being checked.
3. Read the answer together. Is the brand named? Where? Which brands come first? Let the silence work for a moment before you interpret anything.
4. Connect the gaps. Point to two or three items from the readiness check that plausibly explain the answer, in plain language: "the assistants cannot read your product data", not "your JSON-LD fails validation".
5. Bridge to the work. "This was one question on one day. Your buyers ask hundreds of variations, in several assistants. The first thing we would do is measure all of them, then fix what is holding you back."

Ten minutes, and the rest of the call is about scope and price instead of whether AI search matters.

5. Handle the Snapshot Objection Honestly

Sooner or later a prospect says it: "That is just one answer" or "I asked ChatGPT this morning and it mentioned us." Both are correct, and agreeing is the strongest move you have.

A live audit is a snapshot: one answer, to one question, from one assistant, in one market, on one day. AI assistants do not give the same answer twice, they answer each phrasing differently, and they change from week to week as their sources change. That is not a weakness of the audit; it is the argument for the service. Nobody can judge their AI visibility from one chat, theirs or yours, which is exactly why it has to be measured across many questions, several assistants and time.

Overclaiming is the one way to lose this moment. Never present a single answer as their "AI ranking". Present it as the reason to find out.

6. From Snapshot to Retainer

The audit earns the right to propose; the proposal should follow its structure:

• Measurement first. A tracked set of buyer questions, typically 15 to 25 to start, in the prospect's market language, across the assistants their customers use, with the competitors from the audit as the benchmark.
• Foundations next. The technical gaps from the readiness check, fixed in the first weeks, because they are the fastest visible progress.
• Reporting that proves the work. A monthly report under your agency's brand that shows mention rate, position and the competitor ranking moving, so the retainer renews itself.

The full delivery workflow, from the first baseline to the monthly improvement loop, is in our guide on how agencies run GEO for their clients.

7. Running It in Ayzeo: The Prospect Scanner

Ayzeo's Prospect Scanner is built for exactly this call. It opens from the "Sales" section of the dashboard sidebar in its own browser tab, with nothing on it but the check, and the page carries your agency's logo and accent color, with no Ayzeo branding on the page.

• Input: the prospect's website, the market, and the AI assistant. ChatGPT is preselected; Gemini, Perplexity, Claude and others appear when your plan includes them. The question itself is written from the prospect's website.
• Result: the AI readiness score with the biggest gaps, the question that was asked, whether the brand is mentioned, the brands recommended instead, and the full answer.
• Timing: about a minute for the technical part; the live answer usually follows within one or two more.
• Limits: 20 checks per organization per day, for the organization's owner and admins. No contact details of the prospect are collected or stored.

The Prospect Scanner is part of the White-Label Dashboard add-on ($299/month on the Pro plan), together with your branding on the dashboard itself; Enterprise accounts get it on request. Step-by-step instructions are in the Prospect Scanner help article.

8. Mistakes to Avoid

• Running it cold. A private run five minutes before the call costs nothing and saves you from improvising around a surprising answer.
• Calling one answer a ranking. It is a snapshot. Treat it as evidence that measuring is worth it, not as the measurement.
• Drowning the prospect in checks. Two or three gaps in plain language beat a full technical audit on a first call.
• Sharing more than the result. Share the tab, not the screen; your other prospects and clients are nobody else's business.
• Promising positions. Nobody controls what an assistant says. Promise measurement, fixes and a reporting rhythm, and let the trend speak.

9. Frequently Asked Questions

Q: What is a pre-pitch AI visibility audit?

A: A short check an agency runs on a prospect's brand before or during a sales call: whether AI systems can read the prospect's website, what an AI assistant answers when asked a question their customers would ask, and which brands it recommends. It turns a general pitch about AI search into evidence about the prospect's own brand.

Q: Is one AI answer enough to judge a brand's AI visibility?

A: No. One answer is a snapshot: assistants vary their answers between runs, phrasings, markets and days. A single answer is good evidence that a gap may exist and is worth investigating, but judging visibility needs a tracked set of questions across several assistants over time.

Q: Should I run the audit before the call or live with the prospect?

A: Both. Run it privately a few minutes before so you know roughly what to expect, then run it live on the call. The live run is what convinces, because the prospect watches their own result appear instead of looking at a screenshot.

Q: Which AI assistant should I use for the audit?

A: The one the prospect's buyers use most. ChatGPT is the usual default; for some audiences Gemini or Perplexity matters more. Use the prospect's market, because assistants answer differently per country.

Q: Can I run a pre-pitch audit under my agency's brand?

A: Yes. Ayzeo's Prospect Scanner shows the check on a page with your agency's logo and accent color and no Ayzeo branding on the page, in its own browser tab for screen sharing. It is part of the White-Label Dashboard add-on on the Pro plan.

Q: What should I propose after the audit?

A: Measurement, foundations and reporting: a tracked set of buyer questions across the relevant assistants with the audit's competitors as the benchmark, fixes for the technical gaps the audit found, and a monthly report under your brand that shows the progress.

Show, Then Sell

The best GEO pitch is not an argument about the future of search. It is the prospect watching an AI assistant recommend their competitors, followed by a calm, honest explanation of what one answer means and what it would take to change the next thousand. Prepare for five minutes, run it live, agree that it is a snapshot, and propose the measurement that turns it into a plan.

Key Takeaways

• Make it about them: a live answer about the prospect's own brand moves a first call further than any market statistic.
• Three questions, one screen: can AI read the site, what does the assistant answer, and who is named instead.
• Prepare, then run it live: a private run beforehand, then the real one with the prospect watching.
• Agree that it is a snapshot: one answer proves nothing on its own, which is exactly the case for ongoing measurement.
• Propose in the audit's order: measure, fix the foundations, report the progress under your brand.