The email landed on a Sunday night from a dentist I've never met. Subject line: "quick question." Body: could I hop on for twenty minutes this week, take a look at the practice, and tell them what AI to buy. The coordinator was copied. There was a folder of vendor PDFs attached. Hygiene still had holes on Thursday. Insurance follow-up still ate the first hour of the day. They didn't need another unpaid opinion. They needed a decision.
I've gotten the same ask from HVAC owners, managing partners, agency leads, and people who run more than one retail location. A free AI audit of the business. A complimentary look. A "pick your brain" hold that dies the first time a truck breaks. I'm going to say no.
Not because I'm precious. Because a free audit trains you to collect advice you will not use. A paid diagnostic forces a decision. You already know the difference. You've sat through free vendor walkthroughs. You have the folder. The leaks are still there. I sell a free Mini Margin Check if you need the filter. I sell a $1,500 assessment if you want that filter applied to your shop in a 45–60 minute owner interview and seven-business-day analysis. I do not sell a complimentary look. There is no free discovery.
What a free audit actually is
It's a sales call wearing a clipboard. The person on the other end has fifteen minutes to impress you. They cannot reconstruct how a job moves. They cannot sit with yesterday's schedule, last week's unclosed quotes, or the intake pile. So they prescribe tools from a deck. "You should try this platform. A lot of dentists use that. HVAC loves this voice agent." You nod. You get a recap email. You do nothing, because nothing was specific enough to install on Tuesday.
You also learn the wrong lesson: a real diagnosis is supposed to be free until you commit. A Mini Margin Check can identify whether a deeper look is worthwhile, but it is intentionally not a report. People who pay for a diagnosis show up, answer the ugly questions, and make a decision from the result. That is the point of the paid Margin Scan, not a trick.
If you still want someone to "audit" you, make them charge you. Make them write down how they score ideas. Make them give you the page even if you never hire them to implement. Make them say which jobs they would not touch. Make them label every dollar figure as a model. If they offer a free look, you are the product of that look. The output will be a pitch. You already have enough of those.
What I mean by diagnosis
I'm not auditing you for gotchas. I'm not scoring your "AI maturity" so I can sell you fear. I'm asking how work actually ran, then scoring opportunities on three doors only: Effectiveness (money), Efficiency (time), Quality (the customer's experience).
If a recommendation doesn't hit one of those, it doesn't make the page. I will cut stretch ideas. A longer list is not a better diagnosis. A dentist does not need seventeen tools. An HVAC owner does not need a science project. A law firm does not need a bot that talks like a lawyer. An agency does not need a custom product before its own proposals are under control. A multi-location retailer does not need a model trained on anything.
The assessment is a map of your shop. A 45–60 minute owner interview and seven-business-day analysis. Notes. A short report you can hand to a spouse, a partner, an office manager, or a CPA. You can implement it yourself. You can hire anyone. You can hire Applied Margin later on a monthly Embedded Margin Partner retainer if you want the map executed. The document is complete either way. That is what makes it a product instead of a hook.
What we actually do in the paid assessment
We reconstruct a real stretch of time. Not a hypothetical "typical week." I want the day that already happened. Where did the hours go. Which jobs repeated. Where do people copy-paste between systems. What you already tried. If one job disappeared next month, which one would make you exhale.
I record with consent. The conversation is raw material. The deliverable is the scored map, not the vibe of a nice call. I will not prescribe software while the evidence is still arriving. If you show me a 15-step estimate process, the first move might be delete four steps, not add a bot. If your CRM is a graveyard, the first move is a definition of "next action," not another field. Judgment is the service. A model will happily recommend a giant platform to a six-person field shop. A person who has to stand behind the page will not.
The report stays short on purpose. Cover and scope. What I heard. Opportunities scored on money, time, and customer experience. Effort versus impact, in plain English. A card per recommendation: the problem, the evidence from your shop, the change (tool or process), cost band, who owns it, what we would measure. Conservative math with the assumptions labeled as a model, not a promise. Hours become dollars only if you can redeploy the time, cut a cost, or sell more. I will not invent a case-study number and put it on your practice.
Hypothetical (not a client, not a promise): an HVAC owner wants a "free look" because four after-hours calls a week die in voicemail, and even one would have been a $280 diagnostic. The free version of that conversation is "you should try a voice agent." The paid version writes the leak, the assumptions, the human checkpoint, and what not to buy. Same shop. Different product. Run your missed-call count. I will not pretend I already saved you the year.
Then we walk the page. Which item is actually urgent. Whether you want to do it in-house or want help. What timeline is real, not heroic. If it's useful, I can also record a short video of your report so the person who wasn't on the call can see it. The assessment does not depend on it.
Bring this filled in so the 90 minutes isn't a tour of adjectives:
Shop type: [dental / HVAC / law / agency / multi-location retail]
What we sell, in one sentence:
How demand shows up (phone, form, walk-in, referral):
What we already tried, and what died:
The job that ate last week (be specific):
If one loop disappeared next month, which one:
Who can change Tuesday (name and role):
What I will not automate (patient email, filings, auto-send, fake reviews):
What this looks like in a shop, not a deck
Dental: we sit with the schedule that already happened. Openings, confirmations, the insurance pile, the texts that went out from someone's personal phone. We do not start with a website chatbot because a vendor PDF said dentists need one.
HVAC: we sit with the missed-call list and the quotes that didn't leave. We do not start with route-optimization theater. If the owner is still the Saturday dispatcher, that goes on the page. Law: we sit with inquiries that went quiet after the consult. We do not start with a research toy for a firm whose bottleneck is follow-up. Agencies: we sit with the proposal that took a week. We do not start by building a product you can resell to clients. Multi-location retail: we sit with the complaint that got two different answers. We do not start by training a model on anything.
In every case the output is the same shape. Three doors. Ranked loops. What to skip. A page a spouse or partner can read without you translating it.
"Can't you just tell me on a call?"
I can tell you generic things on a call. So can YouTube. Generic: recall matters. Missed calls matter. Intake follow-up matters. Proposals should go out faster. Reviews should be requested after a good job, never invented.
Specific: your coordinator's Tuesday morning, your unclosed quote pile, your consults that went quiet, your store-three complaint that got a different answer than store one. That takes a structured interview and a scored page. I do not do that for free because the free version would be the generic version, and you'd be right to ignore it.
There's another reason. When you pay, you show up. You find the schedule. You don't cancel because a truck broke or a patient ran over. The work happens. That is not a personality judgment. It is what I have seen of unpaid "pick your brain" holds. They die.
I will not put vendor decks in a diagnosis. I will not promise that AI will transform the practice in a quarter. I will not borrow a number from someone else's client. I will not recommend anything that sends to patients, clients, or homeowners without a human. I will not fabricate reviews. I will not open with a custom-build quote before the map exists. "No new tool, fix the path" is a valid card.
The free Mini Margin Check exists because some owners are not ready to have their shop mapped. It is 15 minutes and identifies one workflow that may be worth investigating. It is not an audit, report, or tool list. The paid Margin Scan comes next only when the first look supports it; monthly ownership comes later only when the map is real.
FAQ
Should I get a free AI audit for my small business? No. A free audit is a sales call. It cannot reconstruct how a job moves, so it prescribes tools from a deck. Pay for a diagnosis, or keep the folder of PDFs. Those are the honest options.
What do I actually get from a paid diagnostic? A 45–60 minute owner interview followed by seven business days of analysis: opportunities scored on money, time, customer experience, effort, and risk, plus what not to touch. You keep the result even if Applied Margin never implements it.
Why $1,500 instead of a complimentary look? Price is a seriousness filter. People who pay show up with last week's real mess. Unpaid holds die the first time a truck breaks. I would rather you skip me than collect another unused recap.
If you're the owner who has a folder of free advice and the same Monday, pay for a diagnosis. The assessment is $1,500 for a 45–60 minute owner interview and seven-business-day analysis of your shop — scored, walkable, yours. Then decide. That's the whole point.