AI readiness checklist for a small business: 12 questions before you spend a dollar

Most of what goes wrong with AI in a small business goes wrong before any tool is bought. The tool gets blamed, but the real problem was that nobody knew which task it was for, who would run it, or what it was supposed to save. These twelve questions are the ones we ask in the first hour of an assessment. You can ask them yourself for free, and the answers will tell you whether to spend anything at all.

They are the same twelve that sit on our Opportunity Assessment page, in the same order, because the order matters. The early ones are about knowing your own work. The later ones are about whether a change would survive contact with your team.

The twelve questions

1. Can you name the three tasks that eat the most hours each week? Not the three most annoying ones. The three biggest. Most owners can name one without thinking and then guess at the other two, and the guesses are often wrong. If you cannot name them, no tool can be pointed at them, and the first useful step is a week of rough tallying, not a purchase.

2. Does each of those tasks have one written way it is done? Ask two people to describe the task and see if you get the same answer. If you get two, automation has nothing to automate yet. A written process can be three lines on a sticky note. It just has to be the one version everyone agrees is correct. Standardizing first is the cheapest fix there is.

3. Do you know who does each one and how often? “The office handles it” is not an answer. A name and a frequency is. This is what turns a vague sense of wasted time into a number, and the number is what decides whether a fix is worth the cost. It is also how you find out that the task is done by someone who is leaving in March.

4. Is the information those tasks need already in a system, or in someone’s head? Automation runs on data it can reach. If the pricing lives in a spreadsheet on one laptop, or the customer history lives in a veteran’s memory, the first job is getting it into a shared place. That is unglamorous work and it is usually the real bottleneck.

5. Do your core tools (invoicing, scheduling, email) talk to each other at all? If a job booked in one tool has to be re-typed into another, you have found the cheapest automation in the building. Most mainstream tools can be connected for a subscription fee and a few hours. If you do not know whether yours can, that is a question a half day answers.

6. Is there a task where a rough first draft would save real time? Quotes, follow-up emails, job summaries, proposals, reports. AI is good at first drafts and bad at final answers. If someone on your team spends an hour a day starting from a blank page, that is the kind of place where AI reliably saves time. If nobody does, the case for a writing tool is weaker than the sales pitch suggests.

7. Is there a task where a small mistake would be expensive? Wrong price on a bid, wrong dose, wrong address on a delivery, wrong number on a tax filing. These tasks can still be helped, but they need a human checkpoint built in, and that checkpoint is part of the cost. Knowing which tasks are in this group keeps you from automating the wrong thing first.

8. Has anyone on the team already tried an AI tool on real work, and what happened? Somebody has. They may not have told you. Ask, without judgment, and listen for two things: what worked and why they stopped. The reason they stopped is usually the reason a bigger rollout would also stall, and it is worth knowing before you pay for one.

9. Is there one person who would own a new tool after it is set up? Not the owner. Someone who would notice when it breaks, fix the easy problems, and tell the rest of the team how to use it. If that person does not exist, the tool tends to go quiet within a few months. That is the pattern we have watched teams fall into most often. Naming the owner before buying is the best predictor we know of whether a tool is still in use a year later.

10. Could you spare a half day of owner time and short sit-downs with two or three staff? This is the real cost of an assessment, and it is higher than the fee. If the answer is no right now, wait until it is yes. An assessment done around the owner, rather than with them, produces a memo nobody can act on.

11. Do you have a number in mind for what an hour of the team’s time is worth? It does not have to be precise. Loaded wage plus a margin is fine. Without it, every decision about automation is a guess, because you cannot compare a $4,000 build against “some hours a week.” With it, the arithmetic is a minute’s work, and it is the arithmetic that should make the decision. Our cost-to-automate post walks through exactly that calculation on a hypothetical.

12. If the honest answer is “fix the process first,” would you do that before buying anything? This is the one most owners answer too quickly. Fixing a process is slower and less exciting than buying a tool. It also costs nothing and makes every later tool cheaper. If you would rather buy than fix, say so, because the assessment will sometimes tell you to fix, and the advice is only worth the fee if you would take it.

How to score it

Count the no answers. One or two means you are close, and the gaps are probably specific enough to close yourselves. The worksheet is built for exactly that: fill in the tasks, the people, the frequency, and the hourly number, and you have most of the evidence an assessment would gather.

Three or more no answers means the gaps compound. A tool bought at that point gets aimed at the wrong task, owned by nobody, fed by data that is not in a system yet, and quietly abandoned. That is the pattern we have watched play out most often, and it is the pattern the questions are designed to catch early.

If you said no to three or more, the half-day assessment is where we start. It is on site along the Wasatch Front, from $750, and it ends with a ranked findings memo and one quick win already implemented. You can book it here.

Common questions

Do we need to answer all twelve before we do anything?

No. The list is there to show you where the gaps are, not to block you. A business that can answer the first five is usually ready to connect two tools today. The rest decide whether a bigger build would hold up.

What if we said no to most of them?

Then you are in good company, and the right first step is cheap. Most of the no answers are process and ownership questions, not technology questions. Fixing those costs time, not money, and it makes every later quote smaller.

Is this checklist different from the worksheet?

The worksheet is the thing you fill in: which tasks, who does them, how often, what an hour is worth. This post explains why each question is on it. Do the worksheet after reading this and you will have most of an assessment's first hour done already.

Why does owner time matter so much?

Because the person who can say which version of a process is correct, and who can approve a change, is usually the owner. Without a half day of that person's attention, an assessment produces a list nobody is empowered to act on.