AI audits

What an AI audit should include, and what it should cost

A useful AI audit gives you four things: a map of your workflows with hours on each step, a ranked list of fixes with hours saved, dollars per year and cost to build, a build or skip call on every fix, and a written report. A fixed price with fixed deliverables and a fixed turnaround is the easiest to judge.

"AI audit" has become one of those phrases that can mean almost anything. For some firms it is a free call that ends in a proposal. For others it is a long slide deck about trends. Neither one tells you what you need to know, which is simple: where is my team losing hours, which of those hours can AI give back, and is it worth paying to fix?

This post lays out what a useful audit hands you, how to judge the price, and the questions to ask before you sign anything. We sell an audit, so read it with that in mind. The checklist holds up no matter who you hire.

Start with the work, not the technology

A good audit begins by watching real work. Someone sits with the people who do the job, on a call or in the room, and times each step. How long does it take to turn a new inquiry into a booked appointment? How many times is the same information typed into a different system? Where does a task sit waiting for someone to notice it?

If an audit starts with a list of tools instead, be careful. Tools are the answer to a question. An audit that skips the question tends to recommend whatever the auditor already sells.

Four things you should walk away with

1. A map of the workflows, with hours on it. Not a diagram of boxes and arrows for its own sake. A map that shows each step and how long it takes today. You should be able to point at it and say, "That step is costing us most."

2. A ranked list of fixes with numbers. Each fix should carry three figures: hours saved, dollars per year, and cost to build. Without all three you cannot compare one fix against another, and you cannot tell whether any of them pays for itself.

3. A build-or-skip call on every fix. Some fixes are not worth doing. The saving is too small, the data is too messy, or software you already own does the job. An honest audit says so in writing. If every item on the list comes back "build," you are reading a sales document.

4. A written report you own. You paid for the findings. They should be yours whether or not you hire the same firm to build anything, and they should be clear enough to hand to another developer.

What it should cost

There are three common ways audits are priced, and each has a catch.

Free. A free audit is paid for by the build that follows it. That does not make it dishonest, but it does mean nobody is being paid to tell you "skip this." Expect a shorter look and a longer proposal.

Hourly. Hourly pricing puts the risk on you. You do not know the total until the work is done, and a slower audit costs more.

Fixed price. A fixed price with a fixed list of deliverables and a fixed turnaround is the easiest to judge. You know what you pay, what you get and when. Ours is $2,500 for five business days, and we publish it so nobody has to ask.

Whatever the price, ask whether it is credited if you go on to build. A credit lines up everyone's interests: the auditor is paid for honest findings, and you are not charged twice for the discovery work when a build goes ahead.

Questions to ask before you sign

  • What exactly do I receive, and in how many days?
  • Will every fix come with hours saved, dollars per year and cost to build?
  • Will the report tell me which fixes to skip?
  • Do I keep the report if I do not hire you to build?
  • Is the audit fee credited toward a build, and on what terms?
  • Who does the work: the person I am talking to, or someone I have not met?
  • How will you handle my data, and will you sign a nondisclosure agreement?

Clear answers to those seven questions tell you most of what you need to know about the firm.

What an audit is not

An audit is not a strategy retreat, a training course or a technology roadmap for the next five years. Those have their place. An audit is narrower and more useful: a short, priced look at specific work, ending in a decision you can act on next week.

It is also not a commitment to build. The best outcome of some audits is a short list of things to stop worrying about. Knowing that AI is not worth it yet for a given task saves money just as surely as a good build does.

Where to start

If you are not sure your business is ready for an audit, start smaller. Our free AI readiness quiz is ten multiple-choice questions about your workflows, data and tools. It takes under five minutes, scores you out of 100 and shows you where to look first.