AI Readiness Assessment: 10 Questions to Answer Before You Spend Money on AI

An AI readiness assessment answers one question before you spend money: can this business get a real return from AI right now, and if not, what has to change first? For a small business, readiness comes down to three things: a first use case worth doing, data the tool can reach and trust, and people with rules, training and an owner. Answer the 10 questions below, score each one 0 to 2, and add them up. A total of 0 to 7 means do the groundwork first, 8 to 14 means run one small pilot, and 15 to 20 means you are ready to standardize AI on your first tasks and plan the next step.

If your score comes out low, you have company. In a Census Bureau analysis of its Business Trends and Outlook Survey (opens in a new tab), between 17% and 20% of U.S. businesses said they had used AI in the previous two weeks, across surveys taken from December 2025 to May 2026. The same analysis found AI use grew among firms with at least 20 employees but did not change significantly among firms with fewer than 20.

Money spent on AI gets wasted when a business buys licenses or hires a developer before it knows which job the tool is for.

How to score this AI readiness checklist#

Score every question the same way:

• 0: No, or nobody knows.
• 1: Partly, or the answer lives only in one person's head.
• 2: Yes, and it is written down.

Score it with two or three of the people who do the work. They know the honest answers. It takes about 10 minutes.

First use case: questions 1 to 3#

1. Can you name the first task you want AI to help with?#

Not "marketing" or "efficiency." A task: drafting quote follow-up emails, summarizing service call notes, turning a weekly spreadsheet into a report, answering the same customer questions every day. Good first tasks are frequent, tedious and low risk if a draft is wrong, because a person will review it anyway.

Score 2 if you can name one task and the person who does it today. Score 1 if you have a general area. Score 0 if the plan is "we should be doing something with AI."

2. Could someone write that task down as steps?#

AI works best on work that already has a shape. If the task changes every time depending on who does it, AI copies the inconsistency. The AI Business Maturity Model (AIBMM), the framework our coaches use, makes this a checkpoint: before a business moves past its second level, it should be able to describe at least one workflow as steps (opens in a new tab).

Score 2 if the steps are written down, 1 if one person could explain them, 0 if nobody could.

3. Do you know what that task costs you today?#

This is the question that protects your money. Write down how many hours a week the task takes, how long customers wait, or how often it goes wrong. Without that baseline, you cannot tell whether a subscription or a consultant paid for itself. The AIBMM's criteria for leaving its first level include tracking at least one measurable improvement (opens in a new tab).

Score 2 for a measured number, 1 for a reasonable estimate, 0 for no idea.

People: questions 4 to 6#

4. Does one person own AI decisions, with time to do it?#

Someone has to pick the tools, answer staff questions and say no to risky requests. In a 15-person firm that is often the owner or an operations lead. Ownership with no time on the calendar is not ownership yet.

Score 2 if a named person owns it and has time set aside, 1 if it is understood but informal, 0 if nobody does.

5. Do you know which AI tools your staff already use?#

The AIBMM's description of Level 1 (opens in a new tab) includes employees trying tools privately and quietly using personal accounts, which it calls shadow AI. Ask your staff without blame and write down the answers. Our guide to shadow AI and staff pasting company data into ChatGPT covers how to find out what is already in use.

Score 2 if you have a current list, 1 if you have a rough idea, 0 if you have never asked.

6. Have your people practiced on the tools you want them to use?#

A license is not training. People need hands-on time on their own tasks, with someone who can show them what a useful prompt, a wrong answer and a safe habit look like.

Score 2 if the people who will use AI have practiced on real work, 1 if a few have taught themselves, 0 if nobody has.

Data: questions 7 and 8#

7. Do you know where the data for your first task lives, and who can open it?#

AI tools connected to your files work through the access of the person signed in. Microsoft, for example, says Copilot respects your identity model and permissions (opens in a new tab). That is good news if your permissions are tidy and bad news if an old HR folder was opened to everyone years ago. If you run Microsoft 365, our 12-point Copilot readiness check walks through the permission cleanup.

Score 2 if you can name the system and the people with access, 1 if you know the system but not the access, 0 if the data is scattered across inboxes, desktops and personal drives.

8. Is that data clean enough that a new hire could trust it?#

Three versions of the price list. A customer spreadsheet full of duplicates. A job history that lives in one estimator's memory. AI does not fix any of that. It repeats it, quickly.

Score 2 if the data is current and there is one version of it, 1 if it is usable after some cleanup, 0 if nobody trusts it.

Guardrails: questions 9 and 10#

9. Do you have approved tools and written rules for what never goes into AI?#

Pick the tools you allow, put them on company accounts with multifactor sign-in, and list what never gets pasted in: customer financial data, health information, passwords, anything under a confidentiality agreement. Business accounts put the tool under your company's contract. Microsoft, for one, covers Copilot Chat for organizations under its data protection terms and says prompts and responses aren't used to train foundation models (opens in a new tab). Our free AI acceptable use policy template gives you a first draft. The AIBMM's criteria for leaving Level 1 include an approved toolset, even a small one, and basic do and don't rules.

Score 2 for an approved list and written rules your staff have seen, 1 for rules that are only spoken, 0 for none.

10. Does a named person review AI output before it reaches a customer?#

AI drafts can be wrong in confident, specific ways. Decide who signs off on anything that goes to a customer, a bank, a regulator or a job site. The AIBMM counts explicit review ownership among its criteria for leaving Level 1.

Score 2 if a reviewer is named for each use, 1 if review happens but nobody owns it, 0 if output goes straight out.

What your AI readiness score means#

Add up your 10 scores. Then look for zeros on questions 7, 9 and 10. A zero on data access, rules or review is a stop sign at any total, because those are the gaps that turn a helpful tool into a data leak or a bad customer email.

How your score maps to the AI Business Maturity Model#

The AIBMM has five levels (opens in a new tab): AI Drifter, Prompted Operator, Agent Builder, Agent Manager and Physical Integrator. These 10 questions are a quick screen for the first three, where small businesses make their first AI spending decisions. They are not the official AIBMM scoring.

• 0 to 7: Level 1, AI Drifter. AI use is ad hoc and inconsistent, with no shared system. The AIBMM describes this level as where almost every business starts.
• 8 to 14: working toward Level 2, Prompted Operator. At Level 2 (opens in a new tab), AI becomes the standard first-draft engine for specific tasks, people use shared templates, and a human reviews the output. A supervised pilot is how you build those habits.
• 15 to 20: ready to work at Level 2 and plan for Level 3, Agent Builder. Your answers cover much of what the AIBMM asks before a business leaves Level 1: approved tools, basic rules, chosen workflows and a named reviewer. At Level 3 (opens in a new tab), AI connects to approved tools and data and runs repeatable steps. The AIBMM's criteria for reaching it include standardized prompts that are widely adopted, which these questions do not test, and being ready to connect AI to your systems of record, even read-only at first.

Levels 4 and 5 involve several coordinated AI agents and, at Level 5, machines and physical work. The AIBMM calls Level 5 aspirational (opens in a new tab) for most businesses and sets a practical goal instead: get your three to five most important workflows to Level 3 or 4.

The levels also apply to individual workflows. Your proposals can sit at Level 2 while invoicing is still at Level 1, which is normal. A sensible plan moves one workflow at a time.

Self-check or facilitated AI readiness assessment?#

A self-check tells you where to look, not what you will find there. For a second free read, try our six-dimension AI maturity self-check or the AIBMM's free Discovery assessment (opens in a new tab), which takes 15 to 20 minutes and needs no signup.

A facilitated assessment is worth paying for when you score 8 or higher, when the owner and the team scored very differently, or before you commit to a platform, a developer or company-wide licenses. Coulee Tech's facilitated AI readiness assessment is run by Certified AIBMM Coaches and typically takes 2 to 4 weeks, including stakeholder interviews. It covers six dimensions (strategy, data, technology, talent, culture and governance) and ends with a maturity scorecard, a gap analysis, a prioritized action plan and a presentation to your leadership team.

If you would rather build the judgment to make these calls yourself, AI business coaching works alongside the owner or leadership team, typically over three to six months. Both are part of our AI services for small and mid-size businesses.

Frequently asked questions#

Book a facilitated AI readiness assessment#

If you scored 8 or higher, or scored a zero you are not sure how to fix, book a facilitated AI readiness assessment. We will interview your team, review your documentation and data, and hand you a prioritized action plan before you spend money on tools.