We charge $750 for a fixed-scope AI-Readiness Audit: real data, a 48-hour turnaround, a prioritized 90-day plan. Before we let a single paying stranger buy it, we ran the report itself through six separate reviews — does it sell the next step, does the buyer know what to do, does it look and read like us, can we legally stand behind every claim in it, what happens to the client's data, and does every number trace back to something real. The rule: a failed review blocks the launch, no shipping around it. It failed four of the six on the first pass. In our last post we published the ugly baseline from running this exact audit on our own site — 15 search impressions, 0 clicks, zero analytics events configured. This post covers the review process we put the product through before selling it to anyone else, the four failures it caught, and what we fixed.
Why we gated our own product before selling it
The plan was staged on purpose: run the audit on ourselves first, catch what's broken while the only customer is us, then open it up. Anyone can write a service description that sounds good; the harder test is whether the actual deliverable — the document a customer opens, the emails that follow it, the promises printed on the page — holds up when someone whose job is to find the gap goes looking for one. So before we sold this audit to anyone, we required six independent passes over the real, generated report.
The six checks, in plain English
- Does it sell the next step? Findings need to connect to the paid follow-on work we offer, pricing has to be stated plainly, and any upsell has to be there without pressure tactics.
- Does the buyer know what to do? A time-pressed owner needs a "read this part first" instruction, an owner-do-this-first checklist, a follow-up touch a few days later, and an offer to walk through it live.
- Does it look and read like us — everywhere? Brand colors and type applied correctly, plain-spoken copy, no unexplained jargon, and — because plenty of people read email in dark mode — the emails need to actually render correctly there, not just in a bright inbox on a test machine.
- Are we making any claim we can't stand behind? No disguised guarantees, a clear statement of what data we accessed and that access can be revoked, and no borrowing another company's results to sell ours.
- What happens to the client's data, and can we prove it? Where it lives, how long we keep it, and whether what we promise about deleting it is backed by a mechanism, not just a sentence.
- Does every number in the report trace back to something real? An automated check that pulls every figure in the deliverable and confirms it against the underlying data — not "does this sound plausible," but "does this specific number exist in the source."
Two of the six passed clean: the review of whether the report sells the next step, and the review of whether a buyer knows what to do with it. Four did not.
The four that failed — and what they caught
The brand-and-copy check failed on two counts. One acronym in the report — GA4, the analytics platform — was used seven times before it was ever spelled out, buried in a disclaimer at the very end. A reader following our own "read this part first" instruction would never reach it, so we moved the explanation to the term's first appearance. Separately, the emails that go out with the report were missing the technical markup that tells email clients "this message declares its own colors, don't override them" — without it, a message that reads fine in a normal inbox can render as unreadable text-on-text in a dark-mode one. We added that markup and made every block of both emails declare its own colors explicitly, not just the outer container.
The claims check failed on one count. A draft of the report referenced results from a different engagement as proof that our method works. Even though that engagement is one we've talked about publicly elsewhere, using someone else's outcome as evidence inside a paid deliverable sold to a third party isn't something we have standing to do without their explicit sign-off — and we didn't have it for this context. We rewrote the section to make the same point using only our own audit of our own site.
The data-security check failed on one count. The report promises that raw data we pull from a client's analytics and search console is deleted 30 days after delivery. When this review ran, that sentence was true in intent but not in fact — there was no actual mechanism that would delete anything on day 30. We built and tested a script that does the deletion, keyed off the delivery date rather than the day we happened to pull the data, so the promise and the code now match.
The automated fact-check failed on one count. Two numbers in the report's baseline table — referring domains and AI-answer citations — sat in the same table, same format, as ten other numbers genuinely measured from real data. But we don't currently pull backlink or citation data at all, so those two were an estimate dressed up as a measurement, indistinguishable from the real ones at a glance. We relabeled both as "not tracked by this pipeline yet" instead of asserting a number we couldn't back up.
What this means if you're the one buying it
None of this changes the target numbers or softens the disclaimer already in the report: the 90-day targets are goals we're accountable to, not guarantees, on your report exactly as they are on ours. What it does mean is that before we sell you a document with your data in it, we've already made someone try to find the sentence that oversells, the number that doesn't trace, and the data promise that isn't actually enforced — and fixed what they found. That's a higher bar than "does the report look professional," and one a buyer usually has no way to check from the outside.
The pitch is still the transparency
We're not asking you to take a review process on faith any more than we asked you to take a case study on faith in the last post. The report you'd get is the same one that went through these six checks — same template, same disclaimer, same 48-hour turnaround. And the same verify-the-real-thing discipline runs one level deeper in how we catch our own AI agents reporting "done" when they aren't.
Frequently Asked Questions
Why review your own audit before selling it? Because a service description is easy to write and a real deliverable is hard to stand behind. Running the finished report through six independent checks — sales, clarity, brand, claims, data security, and fact-tracing — surfaces the oversell, the unenforced promise, and the untraceable number before a paying customer ever sees them.
What did the six reviews check? Whether the report sells the next step, whether the buyer knows what to do with it, whether it looks and reads like us, whether every claim is one we can legally stand behind, what happens to the client's data, and whether every number traces back to real source data. Four of the six failed on the first pass.
What does the $750 AI-Readiness Audit include? Real data on how search and AI engines currently see your site, a prioritized 90-day plan, and a fixed 48-hour turnaround. The 90-day targets are goals we hold ourselves accountable to, stated plainly as targets rather than guarantees.
Get the Same Report — Gated the Same Way
Get the $750 AI-Readiness Audit → real data, a plan you keep, 48 hours — or start with a free assessment.
John Colaluca is the founder of Kubernyx, a software and AI automation firm based in Sheridan, Wyoming — building production AI and AI-visibility systems for founder-led firms, including ReceiptStream.