Free AI readiness assessment for small business: what it should ask
Most AI readiness assessments were written for enterprises and ask about model governance and cloud infrastructure. A 20-person company needs different questions. These are the seven things a small-business assessment should measure, and the free one we run answers them in about two minutes.
An AI readiness assessment for a small business should measure whether there is a repeated process to hand off, whether the data it needs exists somewhere a system can reach, whether the rules are written down, and whether the business has a reason to change now. Enterprise assessments such as Microsoft’s cover seven pillars including model management and infrastructure over 45 minutes. Ours asks ten yes or no questions across seven dimensions, takes about two minutes, is free, and ends in a readiness spec sheet.
What readiness means for a small business
Readiness is not a technology score. It is whether a specific piece of work can be handed to a system without the business first being reorganized around it. The failure research is the cleanest definition. RAND interviewed 65 data scientists and engineers about why more than 80 percent of AI projects fail, and the causes it names are all things an assessment can check in advance: the problem was never agreed, the data the project needed did not exist, the team chased the technology instead of the problem, and nothing was in place to run the result.
Flip each of those and you have the readiness test. Is there a named problem? Does the data exist somewhere a system can read it? Is the goal a business result rather than a tool? Is there someone to own the result? A business that can answer yes to those four for one process is ready for that process. That is the whole unit. Nobody is ready for AI in general.
The survey data says most small businesses are somewhere in the middle of this. The Federal Reserve's 2026 Report on Employer Firms found 46 percent of small employer firms using AI, and among them about half still experimenting, 44 percent with AI partially integrated into a process, and 7 percent fully integrated. An assessment is how a business finds out which of its processes could move from the first group to the third.
What the enterprise assessments ask
Search for an AI readiness assessment and the first results are from Microsoft, AWS, and ServiceNow. They are well built and they are built for a different animal. Microsoft's AI Readiness Assessment measures seven pillars: business strategy, AI governance and security, data foundations, AI strategy and experience, organization and culture, infrastructure for AI, and model management. It takes about 45 minutes of multiple-choice questions and returns curated guidance to improve the score over time. Microsoft's accompanying guidance sorts organizations into five maturity stages: exploring, planning, implementing, scaling, and realizing.
For a company with a data team and a cloud contract, that is the right instrument. For a dental practice, a contractor, or a 15-person agency, four of the seven pillars have no referent. There is no model to manage, no infrastructure to provision, and the AI strategy is the owner's opinion. Answering 45 minutes of questions about them produces a score that measures the distance from an enterprise, which the business already knew.
The useful part of the enterprise frameworks is the shape: start from the business, check the data, then decide. A small-business assessment keeps that shape and swaps the questions.
The seven dimensions a small-business assessment should check
Our ten-question diagnostic checks seven dimensions with yes or no questions about a normal week. Each one maps to something a system needs in order to work, or to a reason the business should bother.
| Dimension | What it asks | Why it matters |
|---|---|---|
| Repeatable work | Does anyone copy information from one system into another? | Repetition in the same shape is what a system can take over. |
| Document flow | Do you rebuild the same document over and over? | A proposal or report that changes only in the details is a template with a process around it. |
| Scattered data | Would "how is the business doing" take more than two systems to answer? | Data spread across systems is the gap most first projects close. |
| Writable rules | Do daily decisions follow rules you could write down? Does your highest-paid person do routine work every week? | A rule you can write is a rule a system can apply, and stop on when it does not fit. |
| Trapped knowledge | Would knowledge walk out the door if your best person left? | Knowledge in one head cannot be handed to a system until it is written down. |
| Ground truth | Does important information live outside any system? Is anyone already using AI on their own? | Paper and group texts are invisible to software. Unguided AI use means the boundary question is already live. |
| Growth ceiling | Have you turned away work because the back office could not keep up? Do leads go cold because nobody had time? | The reason to change now, and the number the project gets measured against. |
The ten questions behind the seven dimensions, as asked on the diagnostic. Every question is answered about a normal week, not about plans.
Nothing in that table asks about models, infrastructure, or strategy, because none of those is where a small business fails. It fails on an undefined process, data nobody can reach, and a rule that lives in one head. Those are checkable in two minutes.
How to read the result
The diagnostic ends in a readiness spec sheet: which dimensions lit up and what that points to. Three readings cover most businesses.
- Several dimensions lit, growth ceiling included. There is a process to hand off and a reason to do it now. Name the process, price what it costs today, and check whether an existing tool covers it before anyone builds. This is the reading that leads to a first project.
- Repetition and documents lit, nothing else. A tool probably covers it. The AI Finder returns a shortlist of existing products with the catch on each, and the first project is adoption, not construction.
- Trapped knowledge and ground truth lit, little else. The work is not defined yet. Write the rule down, move the schedule out of the group text, and run the assessment again in a month. Adding software before that makes the mess faster.
Ten questions, about two minutes, free.
Yes or no questions about a normal week across the seven dimensions above. It ends in a readiness spec sheet, and there is nothing to install or hand over.
The three signs a process is not ready
An assessment is worth the two minutes mostly for what it rules out. These three mean the answer is process work, not AI, and each is fixable without hiring anyone.
- It changes shape every time. If no two instances of the work look alike, there is no pattern to hand off. Define the common case first.
- Nobody owns the outcome. With no owner, nobody can define correct and nobody catches a wrong result. Name the person before naming the tool.
- Correct is an opinion. If two experienced people disagree about the right output, write the rule down. The written rule is the deliverable, and it is worth more than the software.
What to do after the assessment
The assessment answers one question: is there a process here that a system could take, and is there a reason to take it now. The next steps are the same regardless of who does them. Pick the one process. Put a number on what it costs today in hours, fees, errors, or delay. Test whether an existing tool covers it. Write down what data may leave your building and what the system is never allowed to decide. Then ship that one process and measure it. The six decisions that come before software walk through each step and what it costs.
If the reading is the first one above and you want the tool step done for you, Buy Your Time Back is $995, payable after we confirm fit, and refunded if the team does not get five hours back in a normal week. If the reading is the third, the best thing we can tell you is to come back in a month, and the assessment will say so.
What to hold on to
- Readiness is per process, not per business. One named problem, data a system can reach, a business result as the goal, and an owner. That is ready.
- Enterprise assessments measure the wrong distance. Model management and infrastructure pillars score how far a small business is from an enterprise, which it already knows.
- Seven dimensions fit a small business. Repeatable work, document flow, scattered data, writable rules, trapped knowledge, ground truth, growth ceiling.
- The result sorts you into three moves. Start a first project, adopt an existing tool, or define the process and come back in a month.
- Two minutes, free, nothing to install. Ten yes or no questions about a normal week, ending in a readiness spec sheet.
Questions owners ask us
What is an AI readiness assessment?
A structured set of questions that tells a business whether a piece of work can be handed to an AI system without reorganizing the business first. Enterprise versions score strategy, governance, data, infrastructure, and model management. A small-business version checks for a repeated process, reachable data, written rules, an owner, and a reason to change now.
Is there a free AI readiness assessment for small business?
Yes. Our ten-question diagnostic is free, takes about two minutes, asks yes or no questions about a normal week across seven dimensions, and ends in a readiness spec sheet. There is nothing to install and no data to hand over. Microsoft also offers a free assessment, built for enterprises, that takes about 45 minutes.
How long does an AI readiness assessment take?
Ours takes about two minutes: ten yes or no questions. Enterprise assessments such as Microsoft's run about 45 minutes of multiple-choice questions across seven pillars. The difference is scope, not depth. A small business has fewer systems to ask about and no models to manage.
What are the signs a small business is not ready for AI?
The work changes shape every time, nobody owns the outcome, or correct is a matter of opinion. Each one means the process is undefined, and software added on top makes the mess faster. All three are fixable in-house: define the common case, name an owner, write the rule down.
What is an AI readiness framework?
A framework is the set of dimensions an assessment scores. Microsoft uses seven pillars from business strategy to model management and five maturity stages from exploring to realizing. Ours uses seven dimensions built around a normal week in a small business: repeatable work, document flow, scattered data, writable rules, trapped knowledge, ground truth, and growth ceiling.
What happens after the assessment?
You get a readiness spec sheet showing which dimensions lit up. From there: name the one process, price what it costs today, test whether an existing tool covers it, and write down the data boundary and the decisions the system may not make. If you want the tool step done for you, Buy Your Time Back is $995 and refunded if the team does not get five hours back each week.
Sources
The enterprise frameworks are described from their own pages; the survey and failure figures are quoted from the reports. Checked September 5, 2026.
- Microsoft Learn, "AI Readiness Assessment". The seven pillars, the 45-minute format, and the guidance output, as described on the assessment page.
- Microsoft Cloud Blog, "A strategic approach to assessing your AI readiness". November 6, 2024. The five maturity stages.
- Federal Reserve Banks, "2026 Report on Employer Firms". Findings from the 2025 Small Business Credit Survey, 6,525 responses. AI use and integration depth.
- RAND Corporation, "The Root Causes of Failure for Artificial Intelligence Projects and How They Can Succeed". The 80 percent failure rate and its root causes, from 65 interviews.
- The ten questions. Our diagnostic, the seven dimensions, and the ten questions quoted in the table.