Wanting AI and being ready for it are different conditions.
I have watched capable companies commit to a build and then discover, in week two, that the customer data lives in three systems with three different definitions of what a customer is. That discovery costs more than the assessment would have.
Short answer: AI readiness is a function of seven things. Clear objectives, documented workflows, usable data, connectable systems, a team that understands what is changing, defined governance, and a named owner. Score each from one to five. Your lowest score, not your average, determines what you should do next.
Why the average is the wrong number
Most readiness frameworks give you a total and a tier. That is comforting and slightly misleading.
Readiness behaves more like a chain. A company with excellent data, strong systems, and no internal owner will fail. A company with modest data and a determined owner usually succeeds, because someone is there to fix things as they surface.
So run the score, then look at your weakest category and treat that as the actual finding.
The seven categories
1. Business objective clarity
The question: can you name the specific business outcome this is meant to change, in one sentence, with a number attached.
- 1 We know AI matters and want to do something
- 3 We have a list of ideas and no ranking
- 5 We can name the workflow, its current cost, and the target
Most companies sit at 2. This is the cheapest category to improve and the one with the largest downstream effect.
2. Workflow maturity
The question: could two people in your company describe this process the same way.
- 1 It varies by who is doing it
- 3 Documented, but the documentation is out of date
- 5 Documented, current, and the exceptions are known and listed
The exceptions matter more than the main path. Every process has a set of edge cases handled by one experienced person’s judgment. If those are not surfaced, the automation will fail on them and lose the team’s trust in the first month.
3. Data condition
The question: is the information the system needs accurate, findable, current, and consistently structured.
- 1 Scattered, duplicated, or living in inboxes
- 3 Mostly in systems, with known quality problems
- 5 Clean, structured, single definition per field, accessible
This is the category where projects die without anyone announcing it. They just take three times longer and produce mediocre output.
4. Systems and connectivity
The question: can your CRM, accounting, document storage, and communication tools be read from and written to.
- 1 Legacy or closed systems with no meaningful access
- 3 Modern systems, but no one has confirmed the connections work
- 5 Documented access, permissions understood, sandbox available
Legacy integration is one of the persistent barriers Canadian businesses report. Worth checking before scoping, not during.
5. Team readiness
The question: do the people whose work changes understand why, and do they trust the change.
- 1 Nobody has been told
- 3 Announced, not discussed
- 5 The team helped scope it and knows what stays under their control
Adoption failure is not a technology problem. When staff believe an automation exists to remove them, they will find reasons it does not work, and some of those reasons will be correct.
6. Governance and risk
The question: are approval thresholds, access limits, logging, and incident response defined before deployment rather than after.
- 1 No AI policy of any kind
- 3 A policy exists, mostly about which tools are allowed
- 5 Defined per system. What it can read, what it can change, what needs approval, how activity is logged
7. Ownership
The question: is there a named person accountable for this system after the consultant leaves.
- 1 No one
- 3 Assigned to someone with no time allocated
- 5 Named, resourced, and reviewing the system monthly
Scoring
Add your seven scores for a total out of 35.
| Total | What it means | What to do next |
|---|---|---|
| 7 to 14 | Foundation work needed | Document one process, fix one data source, name an owner |
| 15 to 22 | Ready for a contained pilot | One low-risk workflow, draft-and-approve, thirty-day measurement |
| 23 to 29 | Ready for workflow automation | Build, measure, then sequence a second and third |
| 30 to 35 | Ready for connected agent work | Multi-step systems with defined autonomy boundaries |
Then override the tier if any single category scored a 1. A single 1 in ownership or governance should hold you at pilot regardless of the total.
The three findings I see most often
Strong systems, weak process documentation. Common in companies that invested in software but never wrote down how they work. The fix is a whiteboard session, not a purchase.
Strong intent, no owner. The executive is enthusiastic. Nobody has been given hours. This is the most reliable predictor of a stalled project.
Good data, no governance. Usually in regulated or client-sensitive industries where the technical foundation is fine and the approval model has never been discussed. Cheap to fix if done before deployment, expensive after.
What readiness does not mean
It does not mean waiting until every category hits 5. That company does not exist and it never will.
It means knowing your weakest link before you build on top of it, and scoping the first project so that weakness cannot sink it. A company scoring low on data can still run a successful intake automation, because intake generates its own clean input.
Readiness is about matching project ambition to organizational condition. Nothing more complicated than that.
FAQ
How long does this assessment take?
Ninety minutes with the right three or four people in the room. Longer if you have to go find out how a process actually works, which is itself a useful finding.
Who should be in the room?
Whoever owns the workflow, whoever owns the data, and whoever will be accountable for the result. Bring at least one person who does the work daily.
What if we score low everywhere?
Then your first project is not AI. It is documenting one process and cleaning one data source. That work has value regardless of what you build afterward.
Should we reassess later?
Yes. Score again after your first pilot. The categories move faster than people expect once a project forces clarity.
Does a high score guarantee success?
No. It reduces the chance of the failure modes that are predictable. The unpredictable ones remain.
Where to go next: Run the seven categories with your leadership team this month. Bring the scores and we will tell you what your first project should be and what it should avoid.
