How to Run a Free GEO Audit on a Prospect Call

Run a free GEO audit on a call in about 15 minutes: agree on 8 to 10 buyer questions, ask them live in ChatGPT, Perplexity and Google, note whether the prospect and its two closest competitors are mentioned or cited, check the prospect's own pages for extractable answers, and finish by naming the three biggest gaps and a paid next step. Keep it a sample, not a full study.
| Minute | Activity | Output |
|---|---|---|
| 0 to 2 | Frame the audit and its limits | Shared expectations |
| 2 to 5 | Agree on 8 to 10 buyer questions and 2 competitors | Short prompt list |
| 5 to 10 | Ask them live in three engines and note who appears | Mention and citation tally |
| 10 to 13 | Inspect what the answers cite and what the prospect's pages say | Top three gaps |
| 13 to 15 | Recommend next step and send the follow-up | Paid audit or pilot proposal |
Why run a GEO audit live instead of sending a report?
Because the screen does the persuading. A prospect who watches ChatGPT recommend two competitors and skip their company understands the problem in ten seconds, while a PDF with the same information can sit unread. Live audits also let you learn what the prospect cares about, which questions they wish they showed up for, and who they consider a competitor.
There is a data reason too. Pew Research Center found that when Google shows an AI summary, users clicked a regular result link in 8% of visits, against 15% without one, and clicked links inside the summary in 1% of visits. If a prospect asks "does this matter?", that is a sourced way to say that AI answers change how people click. Ahrefs added in its December 2025 update that AI Overviews correlate with a 58% lower position one click-through rate across 300,000 keywords, a correlation and not proof of cause. Use these as one-line context, not a lecture.
The risk of a live audit is variance. AI answers change from run to run, so what you see on the call is a sample. Say so at the start, which makes you look careful instead of unprepared when an answer shifts.
What should you prepare before the call?
Ten minutes of prep make the call. Look up the prospect's site, main services, city or region if local, and two or three competitors they are likely to name. Draft 10 to 12 questions in three groups: category ("best accounting software for small restaurants"), comparison ("X vs Y for a 10-person team") and problem ("how do I reduce chargebacks on my online store"). You will drop a few live.
Open the tools in separate tabs, logged out or in a clean browser profile so personalization does not skew results: ChatGPT, Perplexity, Google Search with AI Overviews, and optionally Gemini and Copilot. Have Ahrefs' free AI visibility checker ready. According to Ahrefs, it needs no signup, takes a brand name and reports total AI mentions across ChatGPT, Gemini, Perplexity, Copilot and Google AI Overviews, with a mentions-by-platform breakdown and the top five topics, cited domains and cited pages. It is a limited preview; Ahrefs says full data requires Brand Radar's paid plans.
Prepare a one-page template (a spreadsheet works) with columns for the question, engine, whether the prospect was mentioned, whether its URL was cited, which competitors appeared, and notes.

How do you open the call and set expectations?
Say what the audit is, how long it takes and what it is not. A workable script: "I'd like to spend fifteen minutes checking how your company shows up when buyers ask AI tools about your category. It's a small sample, not a full study, and answers vary from run to run, so treat it as a directional look."
Then ask two questions before you touch a screen. What do your customers ask before they buy? Who do you lose deals to? The answers give you prompts that feel personal and competitors the prospect already worries about.
Do not promise results. Nobody controls an AI engine's output, and Google states that there are no additional requirements or special optimizations for appearing in its AI Overviews or AI Mode beyond standard Search eligibility (pages must be indexed and snippet-eligible). Promising citations would put you at odds with the very documentation the prospect could check.
Which questions should you ask the engines?
Pick eight to ten and choose ones a buyer would type in a full sentence. Pew found that searches of 10 or more words produced Google AI summaries 53% of the time and question-style searches 60% of the time, compared with 8% for one to two word searches. Natural, longer questions are where AI answers appear most, so they are the ones worth testing.
Mix branded and unbranded prompts. An unbranded prompt like "best commercial cleaning company in Denver" shows whether the prospect is in the consideration set. A branded prompt like "is [prospect] any good" shows what the engines say about them, including errors. Wrong hours, wrong services or outdated pricing are common findings and are easy to fix, which makes them an effective opening for a paid engagement.
Add two comparison prompts naming a direct competitor. Those answers often expose which third-party pages (review sites, directories, forums) the engines trust in the category.
How do you read the answers in real time?
For each answer, score four things quickly: is the prospect named, is a prospect URL cited, which competitors are named, and what sources are cited. Do this out loud so the prospect follows along.
Sources deserve the most attention. When ChatGPT or Perplexity cites a review site, a listicle from a trade publication or a Reddit thread instead of any company's own site, that tells you where the influence sits. Ahrefs' analysis of 15,000 long-tail queries in July 2025 found that only about 12% of URLs cited by AI assistants ranked in Google's top 10 for the original prompt (Perplexity 28.6%, ChatGPT in-text 8.6%, Gemini 8.2%, Copilot 8.0%), while 76% of Google AI Overview citations came from top 10 pages. In plain terms, a prospect ranking well in Google may still be missing from ChatGPT answers, and you can show that in real time.
If the prospect appears, note how: as a recommended option, in a list, or in passing. If they do not, note who does and what those competitors have in common. That pattern is your diagnosis.
What should you check on the prospect's own site?
Spend three minutes on a quick page review, ideally on the page that should answer the prospect's highest-priority question. Look for four things.
Does the page state a direct answer near the top, or does it start with a company story? The Liu et al. 2023 "Lost in the Middle" paper found that language model performance is often highest when relevant information sits at the beginning or end of the input and degrades in the middle, which is a research-based reason to lead with the answer.
Does it contain facts a model can quote: specific numbers, prices, service areas, dates, named sources? The Aggarwal et al. 2023 GEO paper tested methods on 10,000 queries across 25 domains and found that adding quotations, statistics and cited sources produced the strongest visibility gains, up to 40% in some settings, while keyword stuffing did little or hurt. Those are research results on specific engines, so present them as a reason to try, not a guarantee.
Is it crawlable and indexable? Check robots.txt and whether the content appears without JavaScript. Google says content must be crawlable and eligible for a snippet to support its AI features.
Is business information consistent across the site, Google Business Profile and major directories? Google lists accurate business profile information among its recommendations.
How do you turn findings into a pitch without overselling?
Name the top three gaps in one sentence each, in the prospect's language. For example: "You are not named in any of the six unbranded questions, two competitors are named in four, and the sources ChatGPT trusts here are two review sites where you have no profile." Those numbers come from your live tally, so record them accurately and read them from your sheet.
Then separate what is fixable quickly from what takes time. Fixable quickly: correcting business information, adding a direct answer and sourced facts to a priority page, claiming missing profiles. Slower: earning third-party coverage, building original data, creating content for questions the prospect never addressed.
Offer a paid next step that fits the finding. A one-time GEO audit, priced in published agency guides between roughly $1,500 and $5,000 (Demand Local's 2026 guidance, which it presents as benchmarking and opinion), gives a fuller prompt baseline and a prioritized plan. A short pilot on one service line is another option. Our article on how much an agency should charge for GEO explains how to compute your own price instead of copying a range.
If you are an agency handing this call to a delivery partner, say so plainly. EliteLeadGrow provides white-label SEO and GEO fulfillment, and agencies often use a live audit on the sales side and hand the paid work to a partner behind their brand.
What follow-up should you send after the call?
Within a few hours, send a one-page summary: the prompts asked, a table of who appeared where, screenshots of two or three telling answers, the three gaps, and the recommended next step with a price or a scoping call. Screenshots matter because answers will differ the next time the prospect checks, and the summary establishes what was observed on the day.
Keep the tone factual. Include a line that the sample was small and that a full audit would run each prompt several times across engines. Attach the prompt list so the prospect can rerun it.
Track your own conversion. After ten audits, count how many became paid engagements and which findings led to them. That is a better guide to your pitch than any industry advice.
What mistakes make a live GEO audit backfire?
Running it cold, without prep, is the most common. The second is treating one answer as truth, then being embarrassed when the prospect reruns it and sees something else. The third is overclaiming, such as saying that fixing the site will make ChatGPT recommend them. The fourth is confusing mention and traffic: AuthorityTech reported in August 2026 that AI referrals are roughly 1% of total website visits, so a prospect who hears "AI is where your customers are" may be right about influence but wrong about referral volume. Semrush has reported that the average AI search visitor is worth 4.4 times the average organic visitor, as covered by MarTech (Semrush owns the publisher, which the article discloses), but treat that as a vendor study.
A final mistake is auditing a prospect who has no SEO base. If the site is not indexed properly or has thin content, GEO advice is premature, and the honest recommendation is to fix the basics first.
How long should a free GEO audit take on a call? Fifteen minutes is enough for a directional look: about eight to ten questions across three engines plus a quick page check. Longer than that starts to become unpaid consulting.
Can I automate the audit before the call? Partly. Ahrefs' free AI visibility checker gives a limited preview of mentions, topics and cited pages without signup, and paid trackers can run prompt sets in bulk. Still run a few prompts live so the prospect sees real answers.
What if the prospect shows up everywhere? Then the finding is different: check accuracy, sentiment and which competitors follow closely, and pitch monitoring or defense of share of voice. Not every prospect needs a big program.
Should the audit be free? A short sample audit is a reasonable sales tool, and a full audit is a paid product. Set the line by time and depth: a free 15-minute sample, a paid multi-run baseline with recommendations.
The bottom line
A free GEO audit on a call works when it is short, prepared and honest about its limits. Ask real buyer questions, show the prospect who appears and where, read the sources the engines rely on, check one important page for extractable answers, and finish with a next step sized to what you found. The audit sells your judgment, and a small accurate sample sells it better than a big claim.
Keep reading
- What Tools Do Agencies Need to Track AI Search Visibility: tooling.
- How Much Should an Agency Charge for GEO Services: turning the audit into a quote.
- What Is Share of Voice in AI Search and How Do You Track It: metric to show.
- How to Build a GEO Case Study With No Clients Yet: proof.
- GEO Audit Checklist for a New Client Pitch
Sources
- Ahrefs: AI Visibility Checker - free tool inputs, outputs, platforms and limits
- Pew Research Center: AI summary and click behavior - 8% vs 15% click rates, summary frequency by query length
- Ahrefs: AI Overviews reduce clicks (update) - 58% lower position 1 CTR
- Ahrefs: Only 12% of AI cited URLs rank in Google's top 10 - overlap by platform, 76% for AI Overviews
- Google Search Central: AI features and your website - no special requirements, indexing and snippet eligibility
- Google Search Central: Optimizing for generative AI features - crawlability, business profile guidance
- arXiv: GEO: Generative Engine Optimization - methods and up to 40% gain
- arXiv: Lost in the Middle - position effects in long contexts
- Demand Local: How to price GEO services - audit price range
- AuthorityTech: AI referral traffic is 1 percent of total visits - roughly 1% of visits
- MarTech: Average LLM visitor worth 4.4x organic search visitors - Semrush 4.4x





