AI Readiness Assessment: A DIY Framework for SMEs
· 9 min read
Key takeaways
- An SME AI readiness assessment scores six dimensions (data, processes, people, budget, governance, execution capacity) from 0 to 3, for a total out of 18.
- Vendor tools from Microsoft, AWS, and ServiceNow assume dedicated IT roles and six-figure budgets that 5-50 person businesses do not have.
- Scores under 6 mean fix foundations first; 6-12 means run a 60-day pilot; above 12 means move to a 90-day rollout.
- Force tradeoffs in your plan: one ICP, one channel, one core offer, and explicit don't-do items so tasks compound.
- Run the assessment in one sitting under two hours; refining scores for weeks is procrastination, not assessment.
An AI readiness assessment tells you whether your business can actually use AI tools well, and you can run one yourself in under two hours. Score six dimensions (data, processes, people, budget, governance, execution capacity) from 0-3, then use the totals to pick a 30-, 60-, or 90-day plan instead of buying an enterprise-branded report.
What Is an AI Readiness Assessment for an SME?
An AI readiness assessment measures whether your business can actually run AI-assisted work: is your data findable, are your processes documented, do your people have capacity, and can a plan move into execution without you carrying every step. For a 5-50 person service business, that means six practical dimensions: data, processes, people, budget, governance, and execution capacity.
Enterprise tools measure something else. Microsoft's adoption and readiness guidance targets organizations rolling out Copilot across tenant-wide IT estates (Microsoft Learn), not an owner-operator with undocumented processes and weak KPIs. The stakes differ too: an enterprise gets a report; you need a 30-, 60-, or 90-day decision. Our beginner's guide to AI readiness covers the same ground in more depth.
Why Most AI Readiness Assessment Tools Are Built for Enterprises, Not You
Microsoft, AWS, and ServiceNow all publish AI readiness frameworks, and they are useful as thinking prompts. But they assume you have the roles and budget the questions are written for.
What the vendor tools do well
They are thorough, free to read, and cover governance and security questions an SME owner often skips. If you want a checklist of what a mature AI adopter looks like, they deliver one.
Where they break down at 5-50 people
The mismatch is structural, not cosmetic:
| Assumption baked into vendor tools | Reality in a 5-50 person service SME |
|---|---|
| Dedicated CIO or data team | You carry strategy, ops and delivery personally |
| Cloud or platform budget in six figures | Price sensitivity measured in CAD per month |
| Multi-quarter governance cycles | Decisions made this week, executed next |
| IT change management process | Undocumented processes, no SOPs |
A framework that asks who chairs your AI steering committee cannot score your business honestly. An SME assessment needs six dimensions you can score yourself in an afternoon, then turn into a 30/60/90-day decision, not a maturity badge. Our free strategy check is built for exactly that scale, and the next section gives you the full framework so you can run it with any tool, including raw ChatGPT.
A Six-Dimension AI Readiness Assessment Framework for SMEs
Enterprise tools score a dozen dimensions with weighted matrices; an SME owner-operator needs six questions they can answer honestly in an afternoon. AWS's own generative AI readiness guidance breaks readiness into business, people, governance, and technology lenses. Here is the same idea compressed for a 5-50 person service business.
Data: is anything documented and findable?
If client records, contracts, and delivery notes live in someone's inbox, AI has nothing to work with. Score yourself on whether a new hire could find your key documents in under ten minutes.
Processes: SOPs before automation
Undocumented processes cannot be automated, only guessed at. Write the SOP first, then ask which step AI could take over.
People: who owns the change?
Founder-dependent work is the default failure mode. Name one person accountable for each AI experiment, even if that person is you.
Budget: what can you actually spend?
A CAD $19/month tool and a $2,500 sprint are different commitments. Decide the ceiling before you evaluate anything.
Governance: board, tone, and approvals
Non-profits and multi-site operators hit approvals, tone, and context first. If leadership sign-off is needed, build it into the plan, not after it.
Execution capacity: can a plan move into delivery?
A 90-day roadmap nobody executes is procrastination with formatting. Structured tools like OGPP Strategy Advisor force tradeoffs, one ICP and one channel, then pair the draft with expert review. Raw ChatGPT gives you a starting draft with no forcing function.
The AI Readiness Assessment Template: Score Each Dimension 0-3
Scoring guidance and what each level means
Score each of the six dimensions from 0 to 3: 0 means absent, 1 means ad hoc, 2 means documented but uneven, 3 means documented and consistently used. Score honestly against evidence, not intent. "We have KPIs" is a 1 if nobody checks them weekly; it is a 3 if a dashboard exists and someone owns it. If you argue about a score for more than two minutes, take the lower number.
Interpreting your total: three decision thresholds
The template produces a total out of 18, and three thresholds drive the decision. Below 6, fix foundations first: pick the two lowest dimensions and spend 30 days there. Between 6 and 12, run a 60-day pilot on one workflow with a clear owner and a before/after measure. Above 12, move to a 90-day rollout with SOPs and review cadence. The thresholds force tradeoffs, which is the point: one ICP-level problem, one channel of effort, explicit "don't do this" items so tasks compound instead of getting cherry-picked.
Worked example: a 12-person property services company
A 12-person property services firm might score: data 1 (jobs in spreadsheets, no CRM), processes 1 (undocumented, founder-dependent), people 2 (stable team, no AI experience), budget 1 (no line item for tooling), governance 1 (no review cadence), execution capacity 2 (one ops lead with slack). Total 8: a 60-day pilot on one workflow, likely quoting or scheduling, with the ops lead as owner. That is a starting draft, not the final plan; budget and capacity need a human pass before anything moves into execution. If you want to see how other owner-operators scoped theirs, the OGPP Strategy Advisor use cases page shows the pattern, and a $149 expert review adds judgement where the score alone cannot.
How to Turn Your Scores Into a 30-, 60-, or 90-Day Plan
Scores only matter if they force a decision. Enterprise tools stop at the report: ServiceNow's assessment outputs a readiness rating and recommendations, then hands off to an implementation team you do not have (ServiceNow documentation). Your version should end with a dated plan instead.
Forcing constraints: one ICP, one channel, one offer
The failure mode is cherry-picking tasks that never compound. So the plan gets constraints, not a task dump: one ICP, one main channel, one core offer, and explicit "don't do this" items. A 30-day plan with three priorities beats a 90-day sheet with thirty. If a proposed action touches a second ICP or channel, it waits for the next cycle.
Where AI drafts and where a human pass is needed
AI drafts the 30/60/90 sequence; a human pass edits it around budget, team capacity, and real distribution. That pass matters most where approvals, board tone, and context live, especially for non-profits. Treat the output as a starting draft, then map which processes are worth automating first: how to identify business processes worth automating. OGPP Strategy Advisor builds plans this way, with an advisor review when the decision needs professional judgement, at CAD $19/month or $149 per review. Honestly, it will not fix weak execution capacity on its own; a score of 0 on that dimension means hiring or delegating before automating anything.
Running the Assessment With Raw ChatGPT vs a Structured Tool
Where a raw prompt falls short
You can paste the six dimensions into ChatGPT and get a plausible-looking scorecard in two minutes. The problem is repeatability and honesty: a raw prompt has no memory of your constraints, will happily score you a 2 on data quality based on vibes, and won't force the tradeoffs that make the exercise useful. As one owner put it, the value is in "forcing constraints: one ICP, one main channel, one core offer," not generating a task list you cherry-pick from. A structured tool encodes the 0-3 rubric so the same inputs produce the same scores, and the output lands as a 30-, 60-, or 90-day decision rather than a generic report.
Running OGPP Strategy Advisor through the same criteria honestly
OGPP Strategy Advisor is built for exactly this flow: score the six dimensions, then turn the scores into an owned plan, with human advisors available when budget, capacity, or a board needs a professional pass. Against the criteria above, it scores well on forcing tradeoffs and execution handoff. It falls short on breadth: it does not do deep data-infrastructure auditing the way an enterprise tool might, and at $19/month membership it assumes you will do the assessment work yourself. If you want a facilitator in the room, a boutique consultant still fits better.
Common Ways SMEs Get the Assessment Wrong
Using the assessment to procrastinate
A score you refine for three weeks is a delay, not a decision. Run the six dimensions once, in one sitting, and act on the band you land in. If you score 0-1 on execution capacity, the honest output is "fix one process first," not a bigger research project.
Cherry-picking tasks so nothing compounds
Picking three easy items off the template while ignoring the constraint you scored lowest defeats the purpose. The framework forces tradeoffs: one ICP, one channel, one offer, and explicit "don't do this" items. Work the weakest dimension, or nothing compounds.
Skipping the governance dimension when a board is involved
If you report to a board or leadership team, an AI-drafted plan from raw prompting in ChatGPT still needs a human pass for approvals, tone, and context. Score governance honestly, or your 30/60/90-day plan stalls at the approval step.
What to Do After Your Assessment: Next Steps by Score Band
Under 8: fix the foundation before buying anything. Pick your weakest dimension, usually data or undocumented processes, and spend 30 days writing SOPs for your two most repeated workflows. Nothing else compounds until this is done.
8-14: run a 90-day roadmap on one core offer and one main channel. Force the tradeoffs yourself: one ICP, one offer, clear "don't do this" items. Treat the output as a starting draft, then have someone who knows your budget and team capacity edit it.
15-18: you have execution capacity, so move the plan into delivery. Assign owners, set weekly KPIs, and review at day 30. If a decision needs professional judgement, an expert review or a sprint beats another solo planning cycle.
FAQ
Is there a free AI readiness assessment template I can download or copy?
Yes. The six-dimension framework in this article works as a template: score data, processes, people, budget, governance, and execution capacity from 0 to 3, add the total, and match it to a 30-, 60-, or 90-day plan. You can copy it into a spreadsheet or paste the dimensions into ChatGPT and run it yourself.
How long does an AI readiness assessment take for a small business?
Under two hours if you run it in one sitting. Scoring six dimensions against evidence takes most owner-operators 60 to 90 minutes, and arguing about any single score for more than two minutes means taking the lower number. The real time sink is refining scores for weeks, which is procrastination, not assessment.
Can I use the Microsoft or AWS AI readiness assessment for my SME anyway?
You can read them, and they are useful as thinking prompts, especially for governance and security questions an SME owner often skips. But they assume dedicated IT roles, six-figure platform budgets, and multi-quarter governance cycles. Questions like who chairs your AI steering committee cannot score a 12-person service business honestly.
What score means I'm not ready for AI at all?
Below 6 out of 18 means fix foundations first: pick your two lowest dimensions, usually data or undocumented processes, and spend 30 days there before buying any tool. A 0 on execution capacity specifically means hiring or delegating before automating anything. No score means never, though; it means the sequencing changes.
Do I need a consultant to interpret my assessment results?
No. The thresholds are mechanical: under 6 fix foundations, 6 to 12 run a pilot, above 12 move to rollout. A human pass still matters for editing the plan around budget, team capacity, and board approvals, and a $149 expert review adds judgement where a score alone cannot, but interpretation itself needs no consultant.