What an AI audit actually looks like (with sample pages)
Most people have never seen an AI audit deliverable. Here's what's in the document, why it's structured the way it is, and what to expect when you commission one.
I get asked variations of this question every week.
“What does an AI audit actually look like? What do I get for the money? Is it just a slide deck?”
Fair question. Most business owners have never seen one, because most AI advisory work happens behind closed doors and the deliverables don’t end up published anywhere. So here’s the inside view, structured around what’s actually in the document I send clients.
The audit isn’t a slide deck
The Strategy AI Audit deliverable is a 15 to 20 page written document. PDF, branded, in your inbox. Not a slide deck. Not a Notion link. Not a Google Doc that I keep editing after I send it.
Two reasons for that.
First, leadership teams need to read it together and pull it apart. A document supports that. Slide decks don’t. Slide decks are designed to be presented at, not read through.
Second, the document needs to be a thing you own. Print it, mark it up, hand it to your internal team, take it to your accountant. Tools like Notion are great for collaboration but bad for handover. The document is yours, it doesn’t expire, and it doesn’t depend on me to make sense of it.
What’s in it
The structure is the same on every audit, partly because it works, partly because consistency makes the document easier to read at a leadership table.
1. How the business runs today
Two or three pages summarising what I learned from the stakeholder interviews and workflow review. Not a transcript of what people said. A clean view of how the business actually operates today, where the friction sits, and what stood out as unusual (good or bad).
The point of this section isn’t to tell you something you don’t know. It’s to confirm that I understand what’s going on. If a client reads this section and disagrees with my read of their business, the rest of the document is going to land wrong, so we sort it out before we get to recommendations.
2. The five revenue areas, scored
Every audit is structured around five areas where AI typically moves the numbers in an established business: customer follow-up and retention, sales pipeline conversion, internal admin and operations drag, customer service, and lead generation.
Each area gets a score and a paragraph. The score reflects how much opportunity I see in that area for this specific business right now. The paragraph explains the scoring and flags whether AI is the right answer or whether the issue is operational.
This is the section that most clients are surprised by. They expect lead generation to score highest because that’s what every AI agency talks about. In most established businesses, customer follow-up scores higher. Sometimes much higher.
3. Specific use cases
10 to 20 specific AI use cases, each scored on revenue impact, implementation effort, timeline, and team adoption risk.
Each use case is a one-paragraph description of what we’d actually build, what tool we’d use, what it would cost, and what we’d expect to see in the numbers within the first 90 days.
The use cases are ranked. Not by my preference. By the score. So the ones near the top are the ones with the best return for the effort.
4. Tooling: keep, fill, remove
For every business with more than a few staff, there’s a stack of existing tools. Some of them are doing useful work, some are duplicating each other, some haven’t been logged into in months but are still being paid for monthly.
This section walks through what I’d keep, what gaps need filling, and what should be cancelled. It includes specific dollar costs where I have visibility, and a recommendation on which tools to deepen versus replace.
This section often pays for the audit on its own, just by stopping a couple of unused subscriptions.
5. The 90 day roadmap
Three 30 day phases. Each phase has specific projects, named owners (roles, not people), tools, costs, and expected outcomes.
This is what makes the document operational. Without this section, an audit is just a list of things that might work. With it, the leadership team can sit down on Monday morning and start the first phase.
The roadmap is sequenced. The order matters. Item 1 in phase 1 is usually a small win, partly because it builds team confidence and partly because the data from the win informs the bigger items in phase 2.
6. What not to do
This is the section most clients tell me is the most useful, and it’s the one I have to fight management consultants over. Most consultants are paid to find work, so they don’t tell you what to skip. I’m paid for the document, not the implementation, so I can.
This section flags AI use cases that look good on paper but won’t pay off in your specific business, plus tools you might be tempted by that I’d avoid for reasons specific to you.
7. Risk register
Data risks (customer information, IP), vendor lock-in, change management considerations. Three to five pages.
This is the section that separates a real strategic audit from a tool shopping list. AI projects fail more often because of team adoption and data handling than because the technology was wrong.
8. Team adoption plan
A short section covering training, communication, and rollout pacing. Specific to your team, not generic.
If you’ve been through a system rollout that died because the team didn’t adopt it, you know why this section exists.
What it’s not
The audit is not:
- A sales document for further work. I write the same audit whether the client engages Dynamo Wheel for implementation or runs it themselves.
- A list of every AI tool on the market. There are too many. The audit covers the ones I’d actually recommend for this business.
- A guarantee. It’s a structured assessment of where AI is most likely to pay off, based on over 30 years inside wholesale and distribution businesses. Implementation still needs to happen.
- An ongoing engagement. Once the document is delivered and the walkthrough is done, the engagement is finished. Clients can come back, but there’s no retainer kicking in by default.
Sample pages
I’m putting together a sanitised sample of the document, with one client’s permission, that I’ll publish here once it’s ready. In the meantime, if you’d like to see how a specific section looks, the right-fit call is the easiest way to walk through it.
The audit is what most clients tell me they wished they’d had two years ago. If you’ve been spending money on AI tools and not feeling like you’re getting traction, the audit is usually where to start.