What-If and Scenario Analysis
Turn levers into outcomes, honestly
What-if analysis drives levers and recomputes outcomes: scenarios as named combos of inputs, sensitivity as one-lever sweeps, and honest comparisons that brace for the 'what if X' question.
What you will learn
- Model scenario inputs as levers and formulas
- Run sensitivity sweeps on single levers
- Compare scenarios with stated assumptions
What is this?
What-if and scenario analysis is the dashboard that answers 'and then what?': a lever model (price, volume, cost) recomputes outcomes as inputs move; scenarios are named combinations (base, aggressive, conservative); sensitivity sweeps flip one lever at a time to show which input the outcome adores. Honest scenario work labels assumptions, brackets ranges, and never presents one path as the path.
Why do you need it?
- Execs ask 'what if we raise price 10%?' — the lever model answers live.
- Sensitivity names the real risk-driver, so teams watch it.
- Scenario framing beats single-point planning on honesty.
- What-if modeling is a staple of every finance-adjacent BI role.
Mental model
A flight simulator's sliders
What-if levers
A simulator doesn't fly a route once; it lets you yank wind, load and throttle and watch the aircraft respond. Scenario analysis is the same lever-cabin: each control is a business input, the gauges are the outcomes, and the discipline is labeling which combinations you actually believe.
How it works
- Build the formula chain: inputs → model → outcomes (transparent arithmetic).
- Name scenarios: base + optimists + pessimists as coherent input bundles.
- Sensitivity: sweep each lever ±20% while others hold; chart the deltas.
- Interactions: note where two levers compound (price × volume curves).
- Present honestly: all scenarios shown, assumptions dated, ranges kept.
What-if: levers in, outcomes out, honest frame.
Examples
A pricing lever model
MODEL revenue = price × qty qty = q0 × elasticity^ln(price/p0) BASE price 100, qty 1000 -> revenue 100k+10% price, elast -1.2 -> qty 912 -> 100.3kSCENARIOS (bundles, labeled) aggressive: price +15%, elast -1.0 -> 103.6k conservative: price +5%, elast -1.8 -> 96.1kSENSITIVITY sweep: outcome rides elast hardest — the team watches elasticity every fortnight, not price.Levers chain into outcomes; scenarios are named bundles; sensitivity names the lever that actually matters.
In production
- A pricing review: price × elasticity model answers the '+10%?' question live.
- A capacity plan sweeps staffing levels against SLA outcomes.
- A budget cycle runs 9 scenarios with date-stamped assumptions, tabled side by side.
Best practices
- Keep the model arithmetic transparent — executives should be able to pound on it.
- Show every scenario, especially the losers — honesty is the report.
- Sweep one lever at a time; interactions get their own labeled section.
- Date and version assumptions; re-run the cycle on material changes.
Common mistakes
- Publishing only the rosy scenario — advocacy disguised as analysis.
- Scenarios that change two levers and claim single-var insight.
- Elasticity band guessed without a sensitivity check.
- Model as black box — no one can audit which formula said what.
Debugging tips
- Outcome barely moves with levers → formula chain wrong or weightless; audit the chain.
- Scenarios diverge wildly → cross-lever interactions unlabeled; structure the sensitivity.
- Exec challenge 'your numbers are wrong' → assumptions stale or unstated; version + date them.
Interview checkpoint
Hands-on lab
- Build a transparent lever model
- Run a one-lever-at-a-time sensitivity sweep
- Read the sweep to name the real risk-driver
Sweep Lever Sensitivity in SQL
45 min · advanced
Quick quiz
Sensitivity analysis reveals:
Scenario analysis should always show the losing scenarios.