Is Your Business Ready for AI? A 12-Point Checklist

A 12-point AI readiness checklist covering data, process docs, people, and governance — with a 0-12 scoring rubric to fix before spending on AI.

What Does "AI-Ready" Actually Mean?

Is Your Data Ready for AI? (Points 1–3)

Are Your Processes Documented Enough to Automate? (Points 4–6)

Is Your Team Ready for AI? (Points 7–9)

Do You Have Governance Guardrails? (Points 10–12)

How Do You Score Your AI Readiness?

What Are the Most Common Readiness Failure Patterns?

The Bottom Line

1. Your core business data lives in systems, not in heads or inboxes

2. Your key systems can talk to other software

3. You actually trust your numbers

4. Your top three workflows are written down, step by step

5. You measure volume and time on at least one process

6. You've identified a high-frequency, rule-based first candidate

7. One named person owns the initiative — with hours, not vibes

8. Leadership has committed a realistic budget

9. Your team sees automation as relief, not replacement

10. You have a written AI usage policy

11. Your security basics are in place

12. A human stays in the loop for consequential decisions

Scored 10–12: pick your first project this week

Scored 8–9: build, but patch your gap in parallel

Scored 4–7: run a 30–60 day fix sprint

Scored 0–3: fix operations before touching AI

Your business is ready for AI when four things are true: your core data lives in systems you trust (not inboxes and heads), your key processes are documented and measured, someone owns the initiative with real budgeted hours, and basic governance — an AI policy, security controls, human oversight — is in place. Score yourself on the 12 points below; 8 or more out of 12 means you're ready to build.

I run a version of this assessment at the start of every fractional CTO engagement, and it predicts project outcomes better than any technology choice does. Businesses that score 8+ almost always get payback on their first automation. Businesses that score below 4 and build anyway are the ones funding the statistic that a third to half of SMB AI initiatives get abandoned in year one. The good news: every item on this list is fixable in weeks, not years, and most fixes cost more discipline than dollars.

Forget the enterprise definition — data lakes, ML teams, transformation committees. For a 5-to-100-person business, AI readiness means the conditions exist for a $3,000–$6,000 automation project to succeed and prove it succeeded. That's it. Each item below is worth 1 point if it's genuinely true today (not "mostly" or "we're planning to"). Be honest; the score only helps if it's real. Grab a pen — this takes ten minutes.

Customers in a CRM, financials in QuickBooks or Xero, jobs in a project or field-service tool. If your source of truth for "what did we quote the Hendersons?" is scrolling someone's email, every automation you build will be starved at the source. You don't need fancy systems — a consistently used spreadsheet scores the point. Inconsistently used anything doesn't.

Check whether your CRM, accounting tool, and scheduling system have APIs, Zapier/Make connectors, or at minimum clean CSV export. Modern cloud tools almost all do; the danger zone is the 2009-era desktop app or the niche industry system with no export path. One legacy system doesn't fail you — but integrating around it costs 2–3x more, so know before you budget.

Would you bet a paycheck that your CRM pipeline reflects reality and your books are current within two weeks? AI automation amplifies whatever it's fed — a lead-scoring workflow built on a CRM full of dead deals just scores garbage faster. If your team keeps "shadow spreadsheets" because they don't trust the official system, score this a zero and fix it first; it's usually a 2–4 week cleanup, not a rebuild.

Not a manual — a one-page numbered list per process: trigger, steps, decisions, exceptions, output. In my assessments, roughly 7 in 10 SMBs fail this point, and it's the single most expensive gap on the checklist because undocumented processes turn into billable discovery time. The fix costs 2–4 hours per process: have the person who does it narrate while someone types.

"How many invoices a month, and how long does each take?" If nobody can answer with a number, you can't calculate payback — and projects without payback math get defunded at the first hiccup. Two weeks of tally-sheet tracking is enough. The ROI framework lists exactly what to capture.

You should be able to name a process that runs 20+ times a month and follows the same steps at least 80% of the time — lead response, invoice entry, appointment reminders. If every candidate on your list is rare or judgment-heavy, you've got an idea problem, not a readiness problem; the priority framework solves it in an afternoon.

Not a committee, not "the office manager when she has time." One person with 2–4 budgeted hours per month to supervise workflows, triage exceptions, and report results. In my experience this single point swings failure rates more than any technology decision — automations without an owner decay silently until someone notices customers stopped getting confirmations.

Year one realistically costs $3,000–$15,000: implementation plus $100–$400 a month to run (full breakdown in the pricing guide). If the mandate is "explore AI but spend nothing," you're not ready — you're window shopping. Equally disqualifying: a $100k "transformation" budget with no single measured process, which just fails bigger.

Have you told the team what's being automated and what happens to the freed hours? Staff who fear replacement quietly sabotage rollouts — they keep shadow processes, don't report edge cases, and let errors slide. The businesses that succeed frame it explicitly: "the robot gets the data entry, you get the customers." One honest team meeting scores this point.

One page: which tools are approved, what data may never be pasted into them, and which outputs need human review. Your employees are already using AI — surveys consistently find a majority of workers use unapproved tools at work — so the policy isn't about permission, it's about steering existing behavior away from the free chatbot with your customer list in it. Template and walkthrough in how to write an AI policy.

MFA on email and financial systems, unique credentials per person, and an offboarding step that revokes access. Automation multiplies security exposure — a workflow with API keys to your CRM and your bank is a juicier target than any single inbox, which is why automation security deserves its own checklist. If you'd fail the SMB cybersecurity checklist badly, fix that before wiring systems together.

Money leaving the building, contracts, hiring/firing, anything compliance-touching: AI drafts, a human approves. This isn't caution theater — it's what keeps a misfired workflow at "embarrassing" instead of "expensive," and if you're in a regulated space (healthcare, finance, legal), it's table stakes. Score the point if you can name where the human checkpoints will sit.

One point per item, honestly assessed. Here's how to read your total:

Your bottleneck is selection, not readiness. Run the scoring exercise in what to automate first, ship one workflow in 30 days, publish the savings internally, and set a one-automation-per-quarter cadence on the SMB roadmap model.

Start the first project, but assign your missing point a deadline. The one exception: if either point 1 or point 3 (data in systems, data you trust) is your zero, fix it before building — data gaps are the only ones that poison a project outright rather than just slowing it.

Don't buy anything yet. Document your top three processes (one afternoon each), start a two-week volume tally, name an owner, and draft the one-page policy. Total cost is typically under $2,000 — mostly internal time — and it converts a coin-flip project into a near-sure one. Re-score at day 45.

You'd be automating chaos, and automated chaos is just faster chaos. Get a system of record in place, get the books current, and get MFA turned on. This is exactly the stage where a few hours of fractional CTO time saves five figures of misdirected spend — the strategy problem has to be solved before the tooling problem exists.

AI readiness for an SMB isn't a data warehouse and a hired ML engineer — it's twelve mundane, checkable conditions across data, process, people, and governance. Score yourself honestly: 8+ means pick a project and build; 4–7 means a cheap 30–60 day fix sprint first; below 4 means the highest-ROI "AI investment" you can make is basic operational hygiene.

Want the assessment run for you? Book a free strategy call — we'll walk the 12 points against your actual systems in 30 minutes, and you'll leave with a score, your two biggest gaps, and what fixing them costs.

• The enthusiastic zero. Owner sees a demo, buys tools the same week, scores a 3 in reality. Six months later the subscriptions are canceled and "we tried AI" becomes the company's official position for two years. The tools were never the problem.
• The perpetual preparer. The opposite failure: scores an 8, keeps "getting ready" anyway — more research, more vendor calls, no build. Readiness above 8 has diminishing returns; at some point the only remaining test is shipping something.
• The hero dependency. Everything rides on one tech-curious employee with no budgeted hours. It works until they get busy or quit, then every workflow decays at once. Points 7 and 8 exist precisely to prevent this.
• The trust gap nobody mentions. The CRM says one thing, the shadow spreadsheet says another, and everyone builds on the CRM anyway because admitting the gap is awkward. The automation faithfully broadcasts wrong data at scale.
• Governance as an afterthought. Workflows ship with shared logins and API keys in a Google Doc, and the AI policy gets written after the first incident instead of before. Retrofitting guardrails costs 3–5x more than installing them on day one.