Inside a Real AI Opportunity Audit: A Worked Example With the Numbers Showing

Every AI consultancy sells an audit. The deliverables all sound the same: ranked opportunities, a readiness score, a roadmap. What none of them show you is the actual work product — the document with numbers on it that you'd be paying for.

So here's ours, opened up. What follows is a worked example of our AI Opportunity Audit: a realistic scenario built from public records and published industry benchmarks, the same way we'd start on your business before your real numbers replace the defaults. It's not a client case study, and we won't pretend it is — it's the method, shown instead of described.

The scenario

A contract beverage manufacturer. Around thirty employees, one automated production line carrying most of the volume, a deliberately capped client roster to protect quality. Every incoming request for quote gets priced by hand: batch size, materials, line time, liquid loss. Quoting takes days, and the people doing it are the same people scheduling the line.

If you run a business with repeatable operations, some version of this is familiar: skilled people spending hours on work a system should do, and nobody sure what fixing it is worth.

Part 1 of the audit: facts vs. models, labeled

The first thing a real audit does is separate what's known from what's estimated. Ours does it in a table, and every number in the document carries one of three labels:

Why this matters to you as a buyer: when a consultant asserts "you'll save $400,000," ask which label each input carries. If they can't answer, the number is decoration. When an input is a benchmark, the honest move is a range plus an invitation: you'll correct us on the call, and the model updates live.

Part 2: the sensitivity table — including the row where it loses money

Here's the part no one publishes. A serious ROI model doesn't produce one number; it produces a grid of outcomes across the assumptions, and it includes the ugly corner:

That first row is deliberate. If a proposal only shows you scenarios where buying wins, you're reading a brochure, not an analysis. The negative cell tells you exactly what has to be true for the project to be worth it — and hands you the question to settle in discovery: how many people actually touch an inbound quote, and how long does one take?

Five minutes with your real numbers replaces the benchmarks, and the grid becomes yours instead of the industry's.

Part 3: the verdict, even when it's "don't build"

The audit ends in one of three findings, in writing:

1. Build — a fixed-price proposal with the ROI math attached, so the quote is accountable to the analysis
2. Buy — an off-the-shelf tool covers it; here's which one and what it costs (we don't sell those, which is why you can trust the recommendation)
3. Don't — the numbers don't clear the bar at your scale; revisit when the trigger conditions we list are met

Roughly speaking, the third finding is the one you're really paying for. Spending $2,500 to avoid a $50,000 mistake is the best ROI in this entire industry — and a firm that never issues verdict #3 isn't auditing, it's upselling.

What this costs, against the market

2026 market pricing for a small-business AI assessment runs $2,000–$8,000, with several firms clustering around $2,500 for a comprehensive tier and at least one charging $6,000 with the fee credited toward a retainer.

Ours is $2,500, fully credited toward any build or retainer within 90 days — so if the audit finds something worth building and you build it, the audit was free. Two to three weeks, and you keep everything: the opportunity map, the labeled model, the sensitivity grid, the verdict.

One honest note on free assessments, since we offer a free step too: a free "audit" is usually a sales call wearing a costume, scoped to find the problem the vendor already sells. Our free scan is deliberately smaller than the audit — a 30-minute call and a written brief naming the top three opportunities we see. Identification is free. Quantification — the labeled model and the grid above — is the paid part. That line is drawn on purpose, and we tell you where it is.

Frequently asked questions

What's included in an AI readiness assessment?

In ours: a process and data review, an opportunity map ranked by value, an ROI model with every input labeled fact/benchmark/estimate, a sensitivity table showing outcomes across assumptions, and a written build/buy/don't verdict with a fixed-price proposal when the verdict is build.

What does an AI audit cost for a small business?

Market range in 2026 is roughly $2,000–$8,000 for SMB scope. Ours is $2,500, fully credited toward an engagement within 90 days. Enterprise assessments run five to six figures and answer different questions.

How long does it take?

Two to three weeks for a focused small-business audit. Anything quoted in months is enterprise methodology billed at your expense; anything quoted in minutes is a quiz, not an audit.

What if the audit says AI isn't worth it for my business?

Then that's the finding, in writing, with the conditions that would change it. You keep the analysis. We think a $2,500 answer that prevents a $50,000 mistake is the strongest outcome an audit can produce.

Is a free AI assessment worth anything?

As a conversation, sometimes. As analysis, rarely — the incentive runs the wrong way, because the free deliverable is the pitch. Ask any provider, including us, exactly where the free part ends and the paid part begins. If they can't draw the line, that's the answer.

---

Want the top three opportunities in your business named, free, in thirty minutes? Book the scan. For how we price the follow-on work, see whether a bot or a hire is cheaper and real retainer numbers for this market. If visibility rather than operations is your gap, start with the ChatGPT check — same method, different lens.