Key Metric Driver Report

Overview

The Key Metric Driver report helps you understand which connected metrics are most likely driving movement in your KPI. It ranks drivers by percent contribution, shows whether they move your KPI up or down, and flags where results are noisy or redundant.

Use it to replace “we think it’s X” with a clearer, shared view of where leverage probably lives.

Report Types & Settings

When you generate a report from a metric node, you can choose a report type:

• Key Metric Driver (this report)
• Marketing Mix Model
• Growth Loop

You can also tailor the written summary:

• Audience: executive, mixed, or technical
• Depth: high-level or deep-dive
• Intent: analysis, strategy, or story

These settings change the narrative and framing, not the underlying numbers.

When to Use This Report

Diagnose KPI Movement

• Problem: Your KPI changed and it’s unclear why.
• Use this report: Identify which drivers moved with the KPI and which ones explain the most variance.

Decide What to Improve Next

• Problem: There are too many “possible levers” and no clear priority.
• Use this report: Focus on the highest-leverage drivers and de-emphasise the rest.

Align Teams on What Matters

• Problem: Different teams tell different stories about what drives outcomes.
• Use this report: Create a shared, evidence-based model of “what moves the KPI.”

Validate Assumptions

• Problem: You have a strong belief about drivers but need evidence.
• Use this report: Confirm whether the data supports the relationship (and whether it’s stable).

What You Get

1. Driver ranking (tiered)

   • A ranked list of connected drivers, grouped into tiers based on contribution quartiles.

2. Direction of impact

   • Whether each driver tends to move your KPI up or down (direction comes from regression coefficients).

3. Contribution breakdown

   • Percent contributions (absolute impact) plus baseline share so you can see how much movement is driver-led.

4. Confidence signals and warnings

   • Significance labels and VIF help you spot noisy results, redundant metrics, or low explanatory power.

5. Recommended actions (when available)

   • Suggestions based on the results (for example: focus on a top driver, review redundancy, or improve data quality).

6. A shareable report

   • A narrative summary (tailored by audience/depth/intent), followed by the statistical analysis, then a notes section for your team.

How to Create This Report

1. Add metrics to your board

   • Add the KPI you want to analyze.
   • Add the candidate driver metrics you want to test.

2. Connect drivers to the KPI

   • Draw connections from each driver metric into the KPI metric.

3. Choose report type and settings

   • Select the KPI metric and click Report.
   • Pick Key Metric Driver.
   • Set audience, depth, and intent to shape the summary.

4. Generate the report

   • Click Generate and open it from the success prompt.

How to Use the Results

• Start with tier 1 drivers: Pick 1–2 drivers to focus on first.
• Check direction before acting: “Important” isn’t automatically “good” — make sure you’re pushing in the right direction.
• Use the recommendations section: Treat it as a prioritised checklist, not a final verdict.
• Turn drivers into bets: Use Driver Analysis to identify levers, then use Bet Impact Analysis to compare initiatives targeting those levers.

Notes & Limits

• Results are directional, not guaranteed causation.
• The report focuses on relationships in the same time period; if you expect delays, interpret carefully and look for lead/lag behavior elsewhere.
• Lagged effects, rolling stability, value chain analysis, and forecasting are not included in this report type.
• Redundant drivers can dilute clarity. If the report flags redundancy, simplify the driver set and re-run.
• If your data is thin, inconsistent, or recently changed (tracking, definitions, cadence), results can be misleading.

Data Requirements

• Your KPI must have at least one connected driver metric.
• Each metric needs at least 20 events. The report uses the most recent 20 per metric.
• Those 20 events must align by date and order across the KPI and drivers; the report does not fill missing dates.
• Each event must include a numeric value and a date.
• Metric names should be unique to avoid ambiguous results.