Start with a key outcome metric you care about—like revenue—and break it down layer by layer into the inputs that drive it. For example: Revenue = Orders × Items per Order × Price per Item; each can be segmented further (new vs. existing customers, channel, region). That's a metric tree—a structure that connects outcomes to the inputs behind them.

Why does this matter now?

Data investment has compounded. Dashboard counts have multiplied. But analyst headcount hasn't kept pace—and neither have insights. Metric trees give your data a model, not just more charts, so analysis flows automatically to the people who need it.

How It Works: Metric Trees Live in 2 Weeks

Step 01 — Design

01 · Model your business.

Typical timeline ·

1–2 meetings

Templates for every business model — and every domain, from finance to growth, product, and ops.

Start with the KPIs that matter — revenue, retention, margin, engagement — and map the inputs.

Start small.

Begin with a single metric tree or business workflow. Most customers are live after modeling just one or two critical KPIs.

Start from a template.

25+ editable shapes spanning SaaS, marketplaces, commerce, fintech, healthcare, logistics, media, and gaming.

Fit your business.

Revenue, CAC, loss ratio, pooled revenue, contribution margin — if your business has a driver model, Trace can represent it.

Step 02 — IMPLEMENT

02 · Connect to your data.

Typical timeline ·

~1 week

LLMs dramatically accelerate setup while preserving governed metric calculations.

Trace's agent writes the spec — SQL, metrics, entities, attributes — from your schema or existing definitions.

100% accuracy. No hallucinated numbers.

AI never calculates your KPIs. Core metrics run on governed SQL — reviewed, version-controlled, exact.

Trace never creates metrics — only reads them.

Trace is a source of intelligence, not another source of truth.

Open and auditable.

Configurations live in git and ship through PRs.

Step 03 — OPERATE

03 · Automate the analysis.

Live by ·

Week 2

The tedious work of analysis runs automatically — so teams can focus on decisions, not dashboard hunts.

Metric trees provide the model. Encoded analytical skills investigate performance across every KPI.

A library of skills, to operate on the metric trees.

Anomaly detection, AutoScan segments, driver attribution, variance-to-plan — the analysis runs itself.

AI-ready by design.

Governed, structured metric trees and built-in analysis skills give LLMs the context to analyze reliably and deeply.

Always on, wherever you work.

Investigate directly in the app, or get robust automated briefings from the Trace AI agent in Slack.

FAQ — From THE demo

Common questions.

the details that matter

Short answers to frequently asked questions.

What exactly is a metric tree?

A metric tree is a computable model of how business metrics connect. It links KPIs to the drivers beneath them, making analysis repeatable, explainable, and automatable.

Do I need to model my entire business?

No. Most customers start with 1-2 metric trees around critical KPIs and expand over time.

How is this different from dashboards?

Dashboards show what happened. Trace explains why. Metric trees and built-in analytical skills automate the investigation behind KPI movement.

How is this different from AI copilots?

AI copilots answer questions. Trace automates analysis. Metric trees and encoded analytical skills give AI the context needed to investigate performance reliably.

How fast can we get started?

Most teams are live within two weeks of kickoff. Designing your trees and connecting to data takes about a week. By week two, Spotlight is running and alerts are flowing.

Do we have to move or restructure our data?

No. Trace reads directly from your warehouse — Snowflake, BigQuery, Databricks, Redshift, Postgres. No migration, no replatform, no new infrastructure.

What about dbt or our existing metrics layer?

Trace coexists with both. We're dbt-aware and read from emerging metrics layers if you have one. Specs sit in git, version-controlled — nothing locked into Trace.

Who owns the trees and specs?

You do. Configs are clean, read-only, version-controlled in git. Trace never creates metrics — it's a source of intelligence, not a source of truth.

What happens when our business model changes?

Add a node, edit an equation, or fork a template. Trees evolve with your business, and every change remains versioned and auditable.

From dashboards to automated analysis.