How and which questions
Short, practical answers for common modelling questions.
How can I simulate my startup idea?
Start with a saved idea so your assumptions stay attached to a workspace. Then run Monte Carlo to stress test outcomes across uncertainty.
- Create an idea in /ideas.
- Fill the Lean Canvas so qualitative context is captured.
- Quantify key KPIs (customers, ARPU, CAC, churn, costs).
- Run a simulation and iterate one assumption at a time.
How do I read a tornado chart?
A tornado chart ranks inputs by impact on an output metric (like EBITDA). The longest bars at the top are your “big levers”.
- Output: What metric are you optimising?
- Range: What uncertainty bounds did you assume?
- Top 3 drivers: Which variables dominate variance?
- Action: What experiment reduces uncertainty fastest?
How should I choose Triangular vs. PERT for startup estimates?
Use Triangular when you want a simple “min / most likely / max”. Use PERT when you want smoother tails and a tunable confidence around the mode.
- Triangular: fast, intuitive, good for early assumptions.
- PERT: better when you believe the mode is more credible than extremes.
Which model should I use (Monte Carlo vs. System Dynamics)?
If you’re validating uncertainty in KPIs quickly, Monte Carlo is the default. If feedback loops and stock/flow dynamics dominate (retention loops, supply constraints, adoption delays), System Dynamics is the right next step.
- Monte Carlo: best for sensitivity + ranges + fast scenario iteration.
- System Dynamics: best for structure, causality, and time-delay effects.
Which variables should I refine first?
Start with whatever shows up as the top sensitivity driver, then refine: customers, ARPU, churn, CAC, and fixed costs. Reduce uncertainty by replacing guesses with real experiments (pricing tests, channel tests, retention cohorts).