AI Readiness Scorecard
Why these questions
Most teams that struggle with AI don't have a model problem. They have an ownership problem, a data problem or a governance problem that the demo hid.
That is why the questions cover the whole product and not the technology alone. People, product, architecture and data, governance and risk, and delivery fail together. They also get fixed together.
The same pattern shows up outside AI. At RBC, growing a digital banking engineering function from 2 to more than 20 people in eight months, across 18 programs, only worked because the people, the architecture and the governance were designed at the same time. AI work is the same problem on a faster clock.
The scorecard
Answer each question for your team as it works today, not as it plans to work. If the honest answer is "sort of," mark it no.
People
1.Does every AI initiative have one named owner who is accountable for the outcome and can stop the work?
Why we ask: Shared ownership usually means nobody makes the hard call when results are mixed.
2.Has at least one product manager and one designer on your team shipped an AI feature to real users, not only a prototype?
Why we ask: AI changes how a product is specified, designed and tested, and that only becomes real after a launch.
Product
3.Can you state the customer problem behind each AI initiative in one sentence, without using the word "AI"?
Why we ask: If the problem needs the technology to describe it, the technology is probably the goal.
4.Did you agree what a "good enough" answer looks like, and who signs off on it, before building started?
Why we ask: Without that agreement, quality gets judged by whoever built the demo.
Architecture and data
5.For each use case, do you know where the data lives, who owns it, and whether you are allowed to use it for this purpose?
Why we ask: Data that exists is not the same as data you can use, and the difference shows up late.
6.Could you change the model or the vendor behind a feature in weeks rather than quarters?
Why we ask: Models, prices and terms change often. A product that can't move with them carries the risk of every change.
Governance and risk
7.Do you have a short, written rule for which AI uses are allowed, which need review and which are off limits, and do teams actually follow it?
Why we ask: A policy nobody reads protects nobody. Teams without one either freeze or improvise.
8.Were legal, security and risk involved before the build started, not at launch?
Why we ask: Late reviews are where pilots go to wait.
Delivery
9.Do you check AI output quality on every release against a fixed set of real examples, the way you test code?
Why we ask: AI features rarely fail loudly. They get worse quietly, and only a repeatable check catches it.
10.Do you know the cost per task, or per customer, of each AI feature that is live?
Why we ask: Usage-based pricing grows with success. Teams that don't track it find out from the invoice.
Your score
Count your yes answers.
0 of 10 yes
Answer all ten questions to see your band.
This scorecard is not a benchmark and the bands are not a maturity model. It is a set of questions drawn from two decades of product and delivery work, meant to show where to look first.
What this scorecard deliberately isn't
- A vendor or tool recommendation
- A maturity model with levels to buy your way up
- A build guide. It tells you where the gaps usually are, not how to close them
- A comparison with other companies. There is no industry average here, because there is no honest one to give
Your answers stay with you
Your answers stay in your browser. They are not sent, stored or tied to you. The page counts visits and button clicks, plus which score band appeared, and nothing else. The analytics are cookieless and respect Do Not Track.
FAQ
Yes. No sign-up and no email required.
A few minutes. It's ten yes/no questions.
Yes. The questions are about how a team makes product decisions and ships them, which applies to a bank, a retailer or a public agency as much as a software company.
The person accountable for the product or the AI portfolio. It's more useful when two or three leaders answer separately and compare. The disagreements are usually the finding.
No. A low score with honest answers is more useful than a high score with hopeful ones. Most teams start in the first two bands.
Twenty minutes on your score and your no answers. You leave with a clearer view of which gap to look at first. There's no pitch deck.
Want a second opinion on your gaps? Book a 20-minute call.
Book a 20-minute callAlso useful: the Pilot-to-production checklist, for when a pilot works and the next step is a live product.