Forget features first. Start with your actual data and who needs to see it.
Define your non-negotiables:
* Source systems (Snowflake, Postgres, a SaaS API?)
* Must-have outputs (daily email PDF, live dashboard, embedded in an app)
* Viewer skill level (analysts vs. executives)
* Your stack (Already on AWS? Using Kubernetes?)
Then look at tools. Score them against your list, not a generic review.
My rough scoring weight for a mid-size team:
* 30%: Connectors and data freshness
* 25%: Total cost (user licenses, compute, ingestion)
* 20%: Learning curve for the report builders
* 15%: Governance (row-level security, audit logs)
* 10%: Look and feel / customization
Example: If you need low-cost, ad-hoc exploration for analysts, Metabase. If you need pixel-perfect, scheduled reports for clients, Looker. If you're already in the Microsoft ecosystem, Power BI.
The tool that wins is the one that disappears and just lets your data flow.
You skipped the most important non-negotiable: the operational burden. "Total cost" includes your team's nights and weekends keeping the thing alive.
Looker's a great example. Pixel perfect reports, sure. Until your Git pipeline breaks or you need to scale the IAP. That's not in the sales brochure. Scoring weight for "who has to fix it at 3 AM" should be at least 20%.
Don't panic, have a rollback plan.