The AI readiness checklist for small business: score yourself.

This is the scoring model behind our free AI Readiness Check, published in full so you can run it on paper. Eighteen questions, six areas, each answered on a 1 to 5 scale, for a score out of 90. The point is to identify an area worth checking against real work, before deciding where to spend time or money.

Why readiness, not adoption, is the question

Adoption alone does not tell you whether AI is helping. KPMG Canada found 93 percent of the Canadian businesses it surveyed use AI in some form, and 2 percent say they are seeing a return on it. Fifty-seven percent said they struggle to capture value at all, up from 40 percent the year before. Those are large firms with budgets. Cisco’s 2025 AI Readiness Index, across 8,000 companies in 30 markets, rates 13 percent as fully ready, a number that has not moved in three years; 64 percent struggle to centralize their data and only 18 percent have finalized which use cases they are pursuing.

Statistics Canada’s small-business survey points to a related question. Its third-quarter 2026 survey found that of businesses with no plans to use AI, 79 percent said it was not relevant and 10 percent said they lacked the knowledge to judge. Those answers do not explain the underlying reason for every firm. A readiness check gives an owner a quick way to examine the workflows before committing to a project.

How to score this

Answer each question on a 1 to 5 scale, where 1 is the left-hand option and 5 is the right. Total the three questions in each area for a score out of 15, then add the six areas for a score out of 90. Be honest rather than generous; this is a self-assessment, and a higher score does not verify the systems behind it. The bands and their limits are at the end.

1. Website clarity

Your website is the first workflow a customer touches. If it cannot explain what you do, every automation downstream is routing confused people.

• Can a visitor understand what you do within 10 seconds? Unclear (1) to Very clear (5)
• Do your service pages explain who you help and what action to take? Not yet (1) to Consistently (5)
• Do you have clear trust signals: proof, examples, reviews, outcomes? Not yet (1) to Consistently (5)

A low score here is a reason to inspect a service page with a real customer task. If its offer, proof, or calls to action are unclear, address that before adding tools behind it.

2. Workflow friction

This is the area that decides whether AI has anything to plug into. Cisco found only 13 percent of companies have repeatable processes for turning ideas into production. A workflow that is not written down is harder to automate reliably, because the team may not share the same understanding of the steps.

• How often does work depend on manual copying, reminders, or inbox follow-up? Constantly (1) to Almost never (5)
• Do you have documented steps for key workflows? Not yet (1) to Consistently (5)
• Can your team quickly identify where work most often gets delayed or lost? Unclear (1) to Very clear (5)

A low score here is a reason to trace one real workflow. Map the handoffs, delays, repeated admin, and owner gaps before deciding whether automation is warranted.

3. Lead capture and follow-up

Leads are the workflow with the most obvious dollar value, which makes it the best place to measure a first automation.

• Are forms, calls, bookings, and inquiries tracked in one place? Not yet (1) to Consistently (5)
• Do new inquiries receive consistent follow-up? Not yet (1) to Consistently (5)
• Do you know which channels produce your best leads? Not yet (1) to Consistently (5)

A low score here is a reason to trace one recent inquiry. Check whether it has a clear owner, next action, and response time before choosing a CRM or automation. If routing is unclear, fix that first; AI can wait.

4. Tool sprawl

More tools feels like progress. Deloitte Canada’s 2026 survey found 59 percent of organizations cite data quality as the constraint on their AI plans and 51 percent cite integrating legacy systems. Extra tools can create duplicate or conflicting customer records, which makes any automation or AI workflow less reliable.

• How many tools are involved in managing a typical request? 6 or more (1) to 1 clear system (5)
• Do your tools share information automatically? Not yet (1) to Consistently (5)
• Do teams know where the source of truth lives? Not yet (1) to Consistently (5)

A low score here is a reason to inspect one typical request across the tools your team uses. Decide which system should hold the current record before connecting more apps.

5. AI usefulness

The CFIB survey reports generative AI use at 45 percent of small firms. I would look at how it is being used before adding more tools. When everyone prompts a chatbot differently, without agreed rules or a review step, it is harder to check quality and protect customer information. The Future Skills Centre found fewer than one in ten small and medium businesses offer formal AI training and about half of employees using AI at work received none.

• Are people using AI in repeatable workflows or only ad hoc? Only ad hoc (1) to Built into workflow (5)
• Are there clear rules for what AI should and should not be used for? Not yet (1) to Consistently (5)
• Are AI outputs reviewed, tracked, or integrated into work? Not yet (1) to Consistently (5)

A low score here is a reason to inspect one current AI-assisted task. Write down the input, the human check, what cannot be delegated, and where the final result goes. Our automation vs AI guide covers how to decide which steps get rules, which get AI, and which get a person.

6. Data and reporting

A baseline number makes it easier to tell whether an automation helped. Reporting gives you something concrete to check a month later, rather than relying only on a general impression.

• Do you have a simple dashboard for important activity? Not yet (1) to Consistently (5)
• Can you see where bottlenecks happen? Unclear (1) to Very clear (5)
• Do you review metrics monthly and make improvements? Not yet (1) to Consistently (5)

A low score here is a reason to choose one monthly decision and list the facts needed to make it. A lightweight dashboard may be useful if the important number is missing or unreliable.

What your score means

Add the six areas. The bands below are percentages of 90, the same thresholds the online check applies. Three of the four are described as cards; the fourth, 72 and up, is "ready to scale." It means the answers describe strong foundations, not that a purchase or retainer is required. Validate the score with a recurring task before investing further.

Under 36 / 90

Needs foundation

The systems are not clear enough for AI to help yet.

Work moves by memory. Inquiries arrive in three places and get answered from whichever one someone opens first. Nobody could draw the process on a whiteboard the same way twice. Buying technology at this stage tends to copy confusion into software. Start with the decisions that define the work; those changes often come before a build.

What it looks likeUnder 40 percent. Typically low in workflow friction and tool sprawl at the same time.

First buildMap one workflow on paper. Pick the source of truth for customer records. Write the follow-up rule. No spend required.

36 to 53 / 90

Scattered but fixable

The pieces exist. They do not connect.

There is a CRM, or half of one. Some workflows are documented. People use AI, each in their own way. The business works because a few people keep the gaps from showing. The score is uneven: strong in one area, weak in another. This is a good place for one well-chosen build because enough of the foundation exists to measure whether it helped.

What it looks like40 to 59 percent. One or two areas under 8 out of 15, the rest middling.

First buildWhichever area scored lowest. Usually a CRM and follow-up system or a workflow mapping sprint, then one automation on the mapped workflow.

54 to 71 / 90

Systemizing

Clear workflows, ready for AI inside them.

Work is tracked in one place, follow-up happens without reminders, and the team could tell you where delays occur. Here, AI has a defined step to sit inside and a number to check afterwards. BCG’s 10-20-70 rule is a useful reminder here: the model accounts for 10 percent of the value in the rule of thumb, while people and process account for 70 percent.

What it looks like60 to 79 percent. The total score is strong enough to test one focused improvement, but it can still hide a weaker area.

Next checkChoose one high-volume repetitive step, such as summarizing, drafting, or classifying. Test it with a human review and a measure of time, errors, or follow-up before expanding it.

Use the lowest area to choose what to inspect first. It is a self-reported signal, so trace one real request or task before treating it as a priority.

The three pillars, and why small business needs six areas

Enterprise frameworks usually reduce readiness to three pillars: data, people, and process, sometimes with technology or governance added. Microsoft’s 2026 Work Trend Index found organizational factors explain about two-thirds of the difference in AI impact between companies, with individual skill explaining the rest, which is the same conclusion from the other direction. Small businesses need the six areas above because at this scale the website is the process, the lead flow is the data, and the owner is the governance. Six areas make the first build easier to identify.

What to do Monday

Score the 18 questions on this page, or take the interactive version, which does the math and sends you a report. Your name and email are required for the report; newsletters are optional. Circle the lowest area, then test it against one real workflow. The federal SME AI Adoption Blueprint published in late 2025 makes the same case: start with the workflow, not the model. A readiness sprint is for businesses that want a second opinion on what their score means.

Score it honestly.Then test one real workflow.

Common questions

What is a readiness checklist?

A readiness checklist is a structured set of questions that scores whether a business has the foundations for a change to succeed, before the change is attempted. For AI, that means scoring the clarity of your website, workflows, lead handling, tools, current AI use, and reporting, because AI plugs into those systems rather than replacing them. A good checklist produces a score, a band, and a first action, not just a list.

How to assess AI readiness?

Score 18 questions across six areas on a 1 to 5 scale, total each area out of 15, and add them for a score out of 90. Under 40 percent means fix foundations, 40 to 59 is scattered but fixable, 60 to 79 is systemizing, and 80 and up is ready to scale. Then identify the lowest-scoring area and check it against one real workflow. It is a starting point for investigation, not proof that you need to build or buy anything.

What are the three pillars of AI readiness?

Most enterprise frameworks name data, people, and process, sometimes adding technology or governance. Cisco’s index uses six: strategy, infrastructure, data, talent, governance, and culture. For a small business the same ideas map to six practical areas: website clarity, workflow friction, lead capture, tool sprawl, AI usefulness, and data and reporting, because at that scale the website is the process and the owner is the governance.

What is the 10-20-70 rule for AI?

BCG’s rule of thumb for where the value of an AI effort comes from: about 10 percent from the algorithms, 20 percent from data and technology, and 70 percent from people, process, and how work changes. It is why readiness matters more than model choice, and why a business scoring low on workflow friction should map its processes before buying any AI tool.

See how we turn an AI readiness review into a scoped plan for your business.