A portfolio operator dashboard for an AI-first software roll-up

The brief

Trilogy operates dozens of acquired software businesses simultaneously. Each one had its own metrics dashboard, its own internal vocabulary, its own definition of "active customer." Leadership reviews happened from sixty parallel status decks, none of which were comparable across brands.

The bet: a federated dashboard with one definition of every metric, sourcing from each brand's existing data warehouse, would replace the deck culture and let leadership operate on the portfolio as a single unit.

What we built

A federated data layer that pulls from each portfolio company's existing warehouse (mostly Snowflake, occasionally BigQuery or Postgres) on a nightly cadence, normalizes against a portfolio-wide schema, and stores the result in a single mart with row-level access control for operator-specific views.

On top of that, an operator dashboard that surfaces ARR, NRR, churn, support-load and an "AI-ops score" — Trilogy's internal measure of how deeply each brand has integrated AI into its operations — for the entire portfolio in one view. Drill-down to a single brand in two clicks; drill-down to a specific customer cohort inside that brand in two more.

The unglamorous part: a 47-page metric dictionary that defines exactly what every term means, signed off by each operating partner. The dashboard is the surface; the dictionary is what makes the numbers trustworthy.

Outcome

Time to produce the weekly portfolio read dropped from 45 minutes to 4 — and the read is now done once, by leadership, in front of the dashboard, instead of compiled across sixty parallel decks. 94% of brand operators log in weekly without prompting. The federated layer pulls from 12 different source systems without those teams having to change their pipelines.

The deck culture has, mercifully, died.

"Before this we had sixty status decks. Now we have one dashboard and the decks are gone."