Discover which modeled drivers matter for the selected output

Keep the target context visible

The synthetic fixture has Line A price EUR 2,000 and volume 100, Line B price EUR 1,800 and volume 100. Line A COGS are 60% of its revenue; Line B has no COGS in this deliberately isolated example. Personnel and other opex are EUR 500 each. Base 2027 EBITDA is EUR 259,000 and EBITDA margin 68.157895%.

Choose Model, annual 2027, Base, reported EBITDA. The discovery pass tests supported writable sources in that context. It excludes read-only outputs and does not treat unrelated future-year assumptions as current-year drivers.

Read a group test before drilling into leaves

A Business Line group moves its eligible assumptions together to their low endpoints, then together to their high endpoints. The ranking uses the largest absolute deviation from Base. A revenue branch can group its linked COGS source; the same input identity is not counted twice.

Joint tests include interactions and offsets. Their effect need not equal the sum of individual bars. Drill into the Business Line and then the price/volume leaf to see the actual range and endpoint result instead of interpreting a group as an accounting contribution.

A margin target asks a different economic question

EBITDA is an absolute amount. EBITDA margin divides that amount by revenue, so a revenue change affects both numerator and denominator. Line B can dominate absolute EBITDA while Line A has a larger tested margin effect. The measured rankings above come from separate native discovery runs on the identical fixture.

Margin impacts are percentage points, not euro amounts and not relative percentages. Default monetary/quantity ranges generally test ±10%; percentage assumptions test ±2 percentage points, subject to input bounds. These choices influence the ordering.

Use discovery as a starting point for investigation

The native checks verify distinct input identities, measured group/leaf values and an unchanged workspace. Discovery may miss an input with a flat response at its tested endpoints, or one not exposed by the current source contract. Absence from the ranking is not proof of irrelevance.

Use Sensitivity to map a chosen response and Tornado to compare a curated shortlist with explicit ranges. Key Drivers supplies discovery and drill-down, not statistical inference, an additive attribution waterfall or a probability of the forecast.

1. Select EBITDA → Analyze → Key Drivers; review the leading groups.
2. Drill into Business Line A and its price/volume or linked cost sources.
3. Repeat for EBITDA margin and compare units, ranges and rankings.

Related decisions and model reference

For adjacent questions: Financial Model Analysis — Drivers, Margins and Cash; Bottom-Up Revenue Forecast: Build Three Business Lines; Profitability Analysis: Trace Profit Margins and Capital Returns.

For the exact controls and source behavior, use Key Drivers: discover tested impact and Eligible inputs, sources and unavailable results.