Choose a model for the work you need to doMe
← Use casesData analysis

Business metrics & growth diagnosis

Break changes in conversion, retention, revenue, or cost into contributing factors and testable business questions.

What to prepare

Inputs

Metric definitions, events or aggregates, historical baseline, and grouping dimensions.

What to produce

Deliverables

Metric specifications, funnel or retention analysis, driver breakdown, and recommended actions.

Step-by-step workflow

Break the work into verifiable stages, then choose models and supporting tools.

  1. 01

    Align metric definitions

    Specify denominators, deduplication units, time windows, currencies, and late-arriving data.

  2. 02

    Check whether the change is real

    Check instrumentation, definitions, traffic composition, and delays before comparing with the baseline.

  3. 03

    Identify major contributions

    Run SQL or Python to segment by channel, region, user cohort, and product version.

  4. 04

    Propose testable actions

    Separate confirmed problems from hypotheses. Assign owners and observation metrics to recommendations.

Required model capabilities

Combine models for the actual stages. The directory includes candidates matching one or more of these capabilities.

View matching models

Supporting tools

Read-only database connectionSQL / PythonMetrics dashboard

Selection & delivery checks

  • Choose strong SQL, calculation, and structured explanation capabilities.
  • Inspect totals and segments to avoid composition changes hiding real trends.
  • Model-generated numbers are not actual database results.

An example request

Use SQL and Python to analyze conversion and revenue changes, segment channels and users, calculate contributions, and summarize growth hypotheses to test.
Find models for this request

Related scenarios

All scenarios