A small business is ready for AI when it can name one repetitive task, point to data clean enough to run it on, name a person who owns the result, and set a number that proves it worked. Score yourself against the 12 questions below: 9 or more checked means start a pilot this month.
Why "AI readiness" gets treated as a bigger question than it is
Most of what gets sold as an AI readiness assessment is a $2,000 to $8,000 consulting engagement that takes two to four weeks and ends in a slide deck. That is the right move for a company picking AI strategy across a dozen departments. It is the wrong first move for a five-person shop trying to figure out whether a chatbot or an automated intake flow is worth building this quarter.
The actual gap between a business that gets value from AI and one that does not has little to do with which model or vendor it picks. Goldman Sachs surveyed 1,256 small business owners across all 50 states in early 2026 through its 10,000 Small Businesses Voices program: 76% are already using AI, 93% of those call the impact positive, but only 14% have it built into core operations (Goldman Sachs, 2026). The gap between "using AI somewhere" and "AI is part of how we run" is a readiness gap, not a technology one, and it is closeable with an afternoon and this list.
The 12-question checklist
Work through these in order. Each one is a yes or no, and "sort of" counts as no.
Process
- Can you write down the exact steps of the task you want AI to handle, in the order they happen, without asking anyone else?
- Does that task happen the same way at least 80% of the time, with only a handful of predictable exceptions?
- Do you know how many hours per week the task currently costs, measured, not estimated?
Data
- Can the tool reach the data it needs (a CRM, an inbox, a spreadsheet, a document folder) without you exporting files by hand every time?
- Is that data reasonably current and free of duplicate or contradictory records?
- Have you checked whether the data contains anything a customer would not want fed into a third-party AI tool: payment details, health information, anything under an NDA?
Ownership and budget
- Is there one named person, not "the team," who will look at the results every week for the first two months?
- Does that person have the authority to turn the pilot off if it is not working, without a committee vote?
- Do you have a number, even a rough one, budgeted for setup and the first three months of usage?
Success and governance
- Have you written down the one metric that decides if this worked, before you build anything: hours saved, response time, tickets deflected, leads answered?
- Is there a way for a customer or employee to tell when they are talking to AI instead of a person, and a path to a human when they ask for one?
- Have you decided who checks the tool's output for accuracy before it reaches a customer, at least for the first month?
That structure follows the same shape NIST uses in its AI Risk Management Framework: govern, map, measure, manage, scaled down to fit a business with no dedicated AI or compliance staff (NIST AI RMF).
What your score actually means
| Score | What it means | What to do next |
|---|---|---|
| 9 to 12 | Ready | Pick the single task with the highest weekly hour cost and scope a pilot this month |
| 5 to 8 | Close | Fix the specific gaps below before spending on a build, usually one to two weeks of process work |
| 0 to 4 | Not yet | The bottleneck is undocumented process or messy data, not AI. Fix that first, revisit this list in a month |
A score in the middle is normal, not a failure. Clutch surveyed 600 small businesses using AI in mid-2026 and found 53% now have a dedicated AI budget, but a much smaller share can answer questions 1, 3, and 10 above without guessing (Clutch, "Small Businesses Are Moving Beyond AI Experimentation," 2026). Budget exists well ahead of process clarity for most businesses on this list, which is exactly backwards from what makes a pilot work.
The three questions that trip up the most businesses
Across engagements, three questions on this list account for most of the "close but not ready" scores.
Question 3, the measured baseline. Businesses know a task is slow but have never timed it. Without that number, there is no way to prove a pilot saved anything, and no way to catch a pilot that is quietly not working. This is a 20-minute fix: time the task for a week before building anything.
Question 7, a named owner. "The team" owns nothing. A pilot with no single person checking results weekly tends to keep running, unmeasured, for months. Assign the name before the kickoff meeting, not after.
Question 11, disclosure and human handoff. This one gets skipped because it feels like a legal question rather than a build question, but it is a design decision that has to happen before launch, not after a customer complains. A written path to a human, even a simple "type 'agent' to talk to a person," resolves it in most small business contexts.
What it costs to close the gaps versus hire it out
A consulting-led AI readiness assessment runs $2,000 to $8,000 and takes two to four weeks to produce a scored report and a prioritized use-case list. For a business with fewer than twenty employees, working through the 12 questions above with the actual task owner in the room usually surfaces the same gaps in under two hours, at no cost beyond that time.
Where outside help earns its cost is not the scoring, it is closing gap 4 through 6: connecting a tool to existing systems cleanly and checking the data for anything that should not leave the building. That work runs alongside the first pilot build rather than as a separate assessment phase. Our AI setup and deployment service runs this as one combined step: score the business, fix the two or three gaps that matter, and scope the first pilot in the same engagement instead of billing for a report first and a build later.
After you score 9 or higher: what to actually build first
Passing this checklist answers whether you are ready, not what to build. That is a separate, smaller decision: pick the one task eating the most hours per week that follows a repeatable pattern, not the one that sounds most impressive to describe. A custom AI agent is worth the extra setup cost when the task touches several internal systems or proprietary data; an off-the-shelf tool configured to the task is the right first move when a decent one already exists. Either way, the same 90-day structure applies: one pilot, one metric, a real decision point instead of a project that quietly runs forever.
Frequently asked questions
How long does an AI readiness check actually take?
Working through the 12 questions with the task owner in the room takes 60 to 90 minutes for most small businesses. The slower part is usually question 3, timing the actual task, which needs about a week of measurement before you have a real baseline instead of a guess.
Do I need a data team to pass this checklist?
No. Questions 4 through 6 ask whether a tool can reach your data cleanly and whether it contains anything sensitive, not whether you have a formal data strategy. Most small businesses can answer these by opening the actual spreadsheet, CRM, or inbox and looking.
What if we score under 5?
Spend a month on process, not AI. Document the task in question 1, time it for question 3, and name an owner for question 7. Those three alone move most businesses from under 5 to 8 or higher, at which point a narrow pilot is worth scoping.
Is a free checklist as reliable as a paid readiness assessment?
For a business under twenty employees choosing between building a first pilot or not, yes. Paid assessments earn their cost at larger scale, across many departments and use cases, where prioritizing between a dozen candidate projects is the actual hard problem. A five-person shop usually has one obvious candidate task, not twelve.
Does this checklist apply to AI voice agents and chatbots the same way?
Yes, the same 12 questions apply regardless of what the pilot looks like on the front end. A voice agent adds one practical wrinkle to question 11: callers need to hear the disclosure, not just see it, and the human handoff needs to work mid-call, not just at the start.
What is the single biggest reason businesses that pass this checklist still fail their first pilot?
Skipping the day-90 decision point. A pilot with a real baseline and a named owner can still drift indefinitely if nobody sits down at the three-month mark and decides to scale, fix, or stop it. The checklist gets you to a good start; a scheduled decision point is what keeps the pilot from running unmeasured for a year.
Score under 9 and not sure which gap to close first? Our AI setup and deployment service runs this checklist with your team, fixes the two or three gaps that actually block a pilot, and scopes the first build in the same conversation. See the full AI Studio lineup or start a conversation about what's eating the most hours in your week right now.