25 questions. One honest answer.

Is there a single, documented definition for each of your core business metrics (revenue, churn, active users, conversion)?

Would sales, finance, and product each give the same number if asked the same question on the same day?

Does every key metric have a named owner accountable for its definition?

Are metric definitions stored somewhere versioned and reviewable — not just in a BI tool or someone's head?

When a definition changes, is there a process that tells everyone downstream?

Do you know how many dashboards you have in production — and which ones are actually used?

Could a new analyst find the authoritative dashboard for a given business question without asking around?

Are your most-used dashboards documented — logic, sources, refresh schedule, caveats?

Have you retired or archived stale dashboards in the last two quarters?

Is there a meaningful difference between certified and ad-hoc content in your BI environment?

Do executives act on your dashboards directly — or do they ask an analyst to double-check first?

Are data-quality issues caught by monitoring before stakeholders notice them?

Can you trace a number on an executive dashboard back to its source systems (lineage)?

When two reports disagree, is there a defined way to resolve which is right?

Do you measure or score data trust anywhere — even informally?

Is critical business logic written down somewhere searchable — or does it live in a few analysts' heads?

Could your team answer stakeholder questions if your most senior analyst left tomorrow?

Are recurring stakeholder questions captured anywhere — or re-answered from scratch each time?

Do analysts spend less than a quarter of their week on repetitive, mechanical requests?

Is tribal knowledge (Slack answers, meeting decisions, exceptions) ever consolidated into documentation?

Have your AI experiments been grounded in approved definitions — rather than raw tables or ungoverned docs?

Do you have a way to evaluate whether an AI answer about your business is actually correct?

Is there a human-review path for AI-generated analysis before it reaches decision-makers?

Are there defined guardrails for what AI tools can and cannot answer in your environment?

If an executive asked “why did this metric move?”, could your AI tooling cite its sources?