A short assessment for a business that wants to know whether it can adopt AI safely — and what has to be fixed first. It is designed to be completed in about fifteen minutes, honestly, for one workflow at a time. The result is a picture of where the business is strong and where it is exposed.

This checklist is the practical entry point to How to Adopt AI in a Small Business. Use it before choosing a use case, so the improvements target the real gaps.

Readiness is not enthusiasm. It is whether the data, the process, the people and the rules can support a workflow this quarter.

How to use it

  • Pick one workflow, not the whole business. Assessing everything produces a vague answer.
  • Answer with facts, not opinions. “We think our data is fine” is not an answer.
  • Score each item yes or no, with no partial credit.
  • Repeat quarterly, and track whether the gaps close.

Fifteen minutes per workflow, once a quarter, is enough to keep the position honest.

The checklist

Data

  • Can you name the source of each figure the output depends on?
  • Is there one agreed definition per metric, recorded somewhere?
  • Are there two versions of any key figure in circulation?
  • Could you explain a change in last period’s figure without investigating from scratch?

Process

  • Could a new team member follow the current process from a written description?
  • Do you know the cycle time end to end, including review?
  • Do you know where it most often goes wrong?
  • Is the output produced from a standard structure, rather than rebuilt each time?

People

  • Is there one named owner for this output today?
  • Is there a reviewer other than the preparer?
  • Does anyone own tool selection?
  • Would the people doing the work support a change to how it is produced?

Policy

  • Is there a written list of approved AI tools?
  • Is there a data rule stating what may not be entered into a tool?
  • Have staff been told the rule, with examples relevant to their work?
  • Is there a named person to ask when the category is unclear?

Measurement

  • Do you know the hours the workflow consumes per cycle?
  • Do you know what an error costs in rework or reputation?
  • Have you defined what success looks like for the next ninety days?
  • Can you report the four measures — usage, time, quality, risk — monthly?

Scoring and what it means

Count the number of “yes” answers out of twenty.

  • 17-20: proceed with a contained workflow immediately.
  • 12-16: proceed, and close the weakest group in parallel with the pilot.
  • 7-11: fix ownership and data first; the pilot would fail for reasons unrelated to AI.
  • Below 7: the business has a process and data problem that AI will expose rather than solve.

The groupings matter more than the total. Strengths and weaknesses cluster: a business with good data and no policy fails differently from one with a policy and no measurements, and the two require different first actions.

The four most common gaps

Across the businesses we assess, four gaps recur.

  • No named owner. The output belongs to everyone, so nobody schedules the work or signs it off.
  • No data rule. Tools are in use without a written position on what may be entered.
  • No baseline. The cycle time is unknown, so no improvement can be demonstrated.
  • No agreed definitions. The same metric means slightly different things in different documents.

All four are cheap to close and expensive to leave. The second is urgent; the third determines whether the program can be defended.

A worked assessment

A twelve-person firm assesses its monthly client report.

  • Data: 2. Two versions of the revenue figure are in circulation, and nobody could say last quarter why the margin moved.
  • Process: 4. The report workflow is documented; a new joiner can follow it.
  • People: 4. The client lead owns the report, and a second person reviews it.
  • Policy: 1. No approved-tool list, no data rule, and staff are using consumer tools on client material.
  • Measurement: 3. Cycle time is known; the cost of an error is not.

The total is 14 — proceed with caution. The pattern is more useful than the number: the policy gap is urgent and cheap, and the data gap blocks any use case that depends on the disputed revenue figure.

The resulting plan is small. Publish a one-page policy in week one. Pick a workflow that does not depend on the disputed metric — drafting the narrative from frozen figures, for example — and fix the revenue source in parallel. No strategy document is required; two decisions and a rule are.

What to do with the result

The output should be a short list, not a strategy document.

  • The gap to close first, with an owner and a date.
  • The workflow to attempt now, chosen from those the gaps do not block.
  • The baseline to capture, before anything changes.
  • The rules to publish, so the pilot starts inside a policy rather than outside one, and so the data rule exists before any real work is processed.

That output is deliberately small. Its purpose is to make the next ninety days produce evidence rather than a deck.

Frequently asked questions

How long does the checklist take?

About fifteen minutes per workflow once you know the process; the first run takes longer because you are establishing the facts.

Who should complete it?

The person accountable for the workflow, with input from finance or operations on the measurement section. A committee produces a softened answer.

What if the answer to an item is “sometimes”?

That counts as no. A control that operates intermittently is not a control.

Is a low score a problem?

Not necessarily — it is a baseline. What matters is the direction it moves, and whether the weakest group improves.

Do we need to fix everything before starting?

No. Close the gaps the chosen workflow depends on, and work on the rest in parallel. The exception is the data rule, which should exist before any real work is processed.

How often should we re-run it?

Quarterly, and before each new workflow. Readiness changes as data improves, the team learns and the tools develop — so a score from two quarters ago describes a business that no longer exists.

Should the assessment be shared with the whole team?

Share the result and the plan, not the scoring. The value is the change the assessment produces, and a score circulated without a plan invites debate about the score.

Next step

Score one workflow honestly, close the weakest gap, and pick a use case the gaps do not block. Download the AI Use-Case Library to choose from, or book an AI adoption call and we will run the assessment with you.

Sources

  • Checklist items reflect standard readiness-assessment practice applied to AI adoption: data, process, people, policy and measurement, with a baseline captured before change.

No statistic in this asset is invented; where figures appear in the linked guides, they are cited there with their source and date.