Sell observation, not outcomes
The product is not “get us into ChatGPT.” No one controls what a probabilistic answer engine generates, and any pitch that implies otherwise converts well and collapses at the first quarterly review. What you can sell honestly, and repeatedly, is uncertainty reduction: a controlled view of how answer engines currently use the client’s sources and describe the brand, plus the eligibility, factual, and source fixes that evidence justifies. What actually changes between classic search work and this engagement is laid out in GEO vs SEO; the sales conversation goes better when you can name those four changes without hype.
The rest of this page is the practical machinery: honest answers to the real objections, a qualification tree for whom to refuse, and a worked pricing anchor.
The objection table
These are the objections that actually come up, with answers that stay true after the contract is signed.
| Objection | Honest answer |
|---|---|
| “Can you get us to number one in ChatGPT?” | No, and no one can. There is no stable ranked page; answers vary by user, phrasing, and day. What I can do is measure how often you appear and are cited across a fixed prompt panel, find the barriers and factual errors we can fix, and re-measure with the same method. |
| “Isn’t this just SEO with a new name?” | Most of the foundation is shared, and I will not bill you twice for it. The new work is real but narrow: crawler policy with separate training and search decisions, answer sampling, and checking what assistants say about you against approved facts. You pay for that method, not for an acronym. |
| “AI referral traffic is tiny. Why now?” | Correct, for most sites it is small today, and I will show you your own numbers rather than a scare chart. The case for a baseline now is that it is cheap, it catches factual misrepresentation early, and it makes every future measurement comparable. If your numbers do not justify monitoring, the audit will say so. |
| “Can’t we just ask ChatGPT ourselves?” | Yes, and you should. One prompt is an anecdote. The audit runs a frozen panel repeatedly under recorded conditions, classifies every answer against facts you approved, and leaves an evidence trail another practitioner could check. That is the difference between a screenshot and a baseline. |
| “Will this get us cited more?” | I cannot guarantee citations, and you should walk away from anyone who does. The audit finds removable barriers and correctable representation risks. Fixing those is worthwhile on its own, and it improves the conditions citation depends on without controlling the outcome. |
| “Why does it cost thousands to read AI answers?” | Because reading is the smallest line item. A standard scope produces hundreds of classified observations plus technical review, fact reconciliation, source mapping, and a prioritized roadmap. I will show you the hour model line by line. |
| “Our developer says block all AI bots and be done.” | That is a legitimate policy, but it is a business decision, not a hygiene default, because one of those bots controls whether you appear in ChatGPT search at all. The decision splits into search discovery and training consent, and I will document both choices for your sign-off. |
| “Which tool will you use?” | Method first, tool second. For your scope I may use a tracker or a controlled manual protocol; either way you get the raw answers, citations, and timestamps, not a black-box score. |
The third and seventh answers do the most work in real conversations, because they concede ground a hype seller cannot afford to concede. Conceding it is exactly what makes the rest credible. The crawler-policy detail behind the seventh sits in OAI-SearchBot vs GPTBot, and the tooling stance behind the eighth in the method-first tool comparison.
Refuse these clients: a qualification tree
A new service category cannot absorb bad-fit engagements. Walk each prospect through five gates, in order, and stop at the first failure.
- Is the site crawlable and indexed on its priority pages? If no: refuse the AI visibility audit and propose SEO remediation first. Answer engines select from accessible, indexable sources; auditing visibility a site is not eligible for wastes everyone’s money.
- Do AI assistants plausibly mediate a buyer decision for this business? If no, for example a purely local walk-in trade with no consideration phase: refuse, or narrow to a small readiness diagnostic. Do not manufacture a threat.
- Is there an owner who can approve brand facts and act on findings? If no: pause until a sponsor exists. An audit whose findings have no owner becomes a PDF, and unowned PDFs do not renew.
- Does the client accept an observation deliverable rather than an outcome guarantee? If no, after you have explained it once, clearly: walk away. This gate protects you from the exact client who will demand the ranking you never promised.
- Does the budget cover the minimum viable scope? If no: offer the smaller readiness diagnostic instead of quietly shrinking the method until the evidence is meaningless.
Prospects who clear all five gates convert well and renew, because the engagement was designed to succeed before it was sold.
The pricing anchor, worked
Anchor the conversation in an hour model, not a competitor’s retainer. The full model, with per-workstream hour ranges by scope tier, is in how to price an AI visibility audit. The standard worked case from that model:
Worked example
Standard audit price floor
Declared assumptions: standard scope of 30 prompts across 3 platforms with 3 runs (270 observations), 52 delivery hours, loaded cost $75 per hour for a senior practitioner, $250 direct tool and transcription cost, 55 percent target gross margin.
Delivery cost: 52 x $75 + $250 = $4,150.
Price floor: $4,150 / (1 - 0.55) = $9,222, rounded to $9,200.
Quote the scope at or above the floor, then round to a commercially sensible number. In the sales conversation, the useful line is the observation count: 30 prompts x 3 platforms x 3 runs is 270 classified observations before any technical, factual, or source work. Clients who see that arithmetic stop comparing the audit to “reading some AI answers.”
Present one number for a defined scope with named exclusions: implementation, ongoing monitoring, and new markets are separate proposals. Scope creep in a new category reads as improvisation, and improvisation kills trust exactly where you need it most.
Sell it inside accounts you already have
Cold-pitching AI visibility invites the hype comparison. Existing clients skip it, because the trust and the site knowledge already exist. The motion that works:
- Raise it inside a scheduled review, not a special meeting. One slide: what answer engines are, what you can and cannot observe, what a baseline costs.
- Bring a three-prompt sample for their own brand, run twice, screenshots attached, clearly labeled as an anecdote rather than a baseline. Its job is to make the category concrete, not to alarm. If the sample looks fine, say so; that honesty is the pitch.
- Scope from their real buyer decisions, using the panel construction method in the prompt set guide.
- Send a fixed proposal with method, deliverables, exclusions, and acceptance criteria, structured like the audit proposal template.
- Deliver the audit, then let the roadmap sell the next stage. Monitoring and implementation renew from evidence, not from urgency.
Written exclusions, consent on record, a frozen panel and findings that trace to saved evidence are all scored checks. Know your own weak sections before a prospect finds them.
Score before you pitchFear-based selling is the shortcut to avoid at every step: no “your brand is invisible in AI” cold emails, no invented revenue-loss figures, no urgency theater. Beyond being commercially fragile, unsubstantiated performance claims are the kind of advertising the FTC’s guidance for small businesses exists to police: claims need substantiation, and an observation engagement is the rare service whose claims are fully substantiatable. If you want the complete delivery system behind the sale, method, workbook, report, and proposal, that is what the AI Search Visibility Audit System packages.