MetricWise: A BI Tool with a Trusted Metrics Layer

MetricWise is a BI tool built around a trusted metrics layer.

It helps teams define important business metrics once, then use them consistently across dashboards, self-serve exploration, and AI analytics. Instead of every dashboard carrying its own version of revenue, active users, or churn, MetricWise gives your team one approved source of truth for the numbers that matter.

I’ve been building MetricWise over the last few months, and it is now ready for first customers to try.

Most BI tools make it too easy for metric logic to spread across reports, dashboards, SQL queries, and ad hoc analysis.

One team defines active users one way. Another team uses a slightly different definition. A third dashboard has its own SQL. Nobody notices until two reports show different numbers for the same metric.

That problem gets worse with AI analytics. If an AI tool is pointed directly at raw warehouse tables, it can generate SQL that looks correct but returns the wrong answer.

MetricWise is built around a simple idea: define trusted metrics once, govern them properly, and make every analytics workflow use those definitions.

MetricWise lets you build semantic models from your warehouse tables.

You can define tables, joins, dimensions, and measures in a visual editor or directly in a DSL. You can switch between both depending on how your team prefers to work.

The semantic layer describes how your data is structured. The metrics layer defines the approved business metrics your team can trust.

Before a model goes live, MetricWise validates it and shows diagnostics so you can catch issues early.

MetricWise includes AI analytics that works from approved semantic models, not raw tables.

Users can ask questions in plain English, choose the model scope, view history and settings, and save generated dashboard artifacts into the Library.

The goal is simple: AI answers should come from governed metric definitions, not guesses against your warehouse schema.

For deeper analysis, MetricWise includes Visual Explore.

Users can pick dimensions and measures, add filters and date controls, run queries, inspect the generated SQL, and review the results. From there, they can customize charts and save them as reusable charts, dashboard tiles, or metric views.

This gives teams self-serve analytics without losing control of metric definitions.

MetricWise supports dashboards, folders, private assets, and shared assets.

Teams can search, sort, and filter Library items, drag and resize dashboard tiles, add dashboard filters, duplicate dashboards, copy links, and open any tile back in Explore when they want to continue analysis.

MetricWise includes a catalog for shared and private metrics.

Users can browse metrics, view canonical definitions, save their own metric views, turn those views into charts, and add them to dashboards.

The point is to make the correct definition easy to find before a number becomes a meeting debate.

MetricWise supports GitHub source control for semantic model changes.

Teams can connect a repository and branch, review which model files were added, changed, or deleted, write a commit message, and push updates.

Private dashboards and charts are excluded from sync, so personal analysis does not accidentally end up in production.

MetricWise currently supports BigQuery service account connections, Google Sheets, and file-style connections.

You can test connections, refresh schemas, select visible tables, and control field visibility.

Snowflake is visible in the app and marked as “Soon” while support is being completed.

MetricWise supports Google sign-in, workspaces, workspace switching, workspace creation, member invites, role changes, and removals.

Admins can also manage plan details, query defaults, theme settings, and which models AI Search is allowed to use.

MetricWise is now open to first customers.

If your team wants a BI tool with a trusted metrics layer, self-serve exploration, dashboards, AI analytics, and source control for semantic models, get in touch.

Define metrics once. Use them everywhere.