1. If someone in a key role left tomorrow with no handover, how much of what they know would still be findable somewhere?
Nearly all of it is written down somewhere more than one person could find.
Some of it is written down; the rest would need reconstructing.
Very little of it is written down anywhere.
We'd genuinely be stuck; it exists in their head only.
2. If you asked two people on the same team the same policy question, how likely is it you'd get two different answers, both said with confidence?
Very unlikely, there's one answer and people give it.
Unlikely for most things, though there are a few known grey areas.
Fairly likely for anything slightly unusual.
Very likely, this happens regularly.
3. When a written process or policy changes in practice, how reliably does the document describing it get updated?
There's a real review process and it's actually followed.
Some things get reviewed on a cycle; a lot don't.
Updates happen occasionally, usually only when someone notices.
Documents just go stale; nothing flags it.
4. The exceptions to how your team normally works, special cases, one-off approvals, workarounds, who owns keeping track of those?
There's a clear owner and a place they're recorded.
Some are owned and recorded; others aren't.
They mostly live in a couple of people's heads, informally.
Nobody owns this; exceptions just happen and get forgotten.
5. If someone asked an unusual question, something outside the day-to-day, how long would it typically take to get a reliable answer?
Minutes; it's findable by more than one person.
Under a day, usually by asking around.
Several days, involves tracking down the right person.
We might never get a fully reliable answer.
6. How comfortable would you be letting an AI system handle one of your team's real judgement calls without a person checking it first?
Comfortable for most calls; we've a good read on which need a person.
Comfortable for some, clearly not for others.
Not very comfortable for most of them.
Not comfortable at all; nothing here is clear enough yet.
7. When someone new joins a team, how do they learn the parts of the job that aren't in the onboarding docs?
Onboarding docs actually cover the parts that matter, including the exceptions.
Docs cover the basics; the rest comes from shadowing a colleague for a while.
Mostly by asking around and getting inconsistent answers.
By making mistakes until someone corrects them.
8. If you picked a random policy document or internal page right now, how confident are you it reflects how things actually work today?
Very confident, most things are current.
Confident about the well-used pages, less so about the rest.
Not very confident; plenty is probably out of date.
Not confident at all; I'd assume it's wrong until checked.
9. Has an AI tool, or a person working from incomplete information, already got something wrong in a way that traced back to missing or unclear information?
No, and we'd probably catch it early if it happened.
Not that we know of, but we're not fully sure we'd catch it.
Yes, at least once, and it was caught before it mattered.
Yes, and it caused a real problem before anyone noticed.