In production
What-if analysis: simulate the decision before you make it
A what-if is a second forecast under a change you control — price up 8%, a promo pulled, a plant offline. You compare paths, not opinions. The point is not a perfect simulation; it is catching the decision that only looks cheap this week.
Updated Aug 4, 2026·7 min read
A second path
A what-if is a second forecast under a change you control. Price up 8%. Promo pulled. Plant offline. You compare paths from the same model, not two opinions in a Google Sheet that cannot hold an interval.
Base path vs promo-on
Levers that are actually yours
The lever must be an input the model has. “What if demand were higher” is a wish. “What if we cut price 8% and keep the learned price response” is a scenario. Capacity, assortment, marketing spend — same test: can you timestamp it, and did the model see that kind of move before?
If the model never saw a 30% discount, do not treat the path as a lab result. Treat it as an extrapolation, and say so.
How to compare without cheating
Hold everything else still. Change one lever, or a package you would actually ship together. Keep the interval — a point-versus-point comparison pretends you know more than you do. Look at the same horizon you would buy on, not a prettier one.
What a scenario cannot do
It cannot see a competitor’s surprise. It cannot invent a relationship that was not in the training window. It cannot replace a test market. What it can do is catch the decision that only looks cheap this week — the promo that steals next week, the price cut that fills the truck and empties the margin.
FAQ
- What is what-if analysis in forecasting?
- You take the live model and change an input you actually control (price, promo, capacity), then generate a new path. The difference is the estimated effect, with the same uncertainty style as the base forecast.
- Is this the same as a sensitivity analysis?
- Sensitivity usually nudges one input to see how jumpy the output is. A what-if is a decision-shaped change ('cancel the holiday promo'). You want both: one for trust, one for planning.
- Why not just build another spreadsheet?
- You can. You will lose the model's learned relationships and the interval. Spreadsheets are great at arithmetic. They are weak at 'what usually happens after a 10% discount on this category.'
Keep going
Guide
Forecasting with covariates
A covariate is an extra series the model is allowed to see — a promo flag, a temperature, a price. Used well, it explains swings the target's past cannot. Used as a dump of every column you have, it just overfits last quarter.
Guide
Prediction intervals
A prediction interval is a range the future value is expected to fall in, not a promise. p50 is the middle path; p10–p90 is a band for planning stock, staff, and cash without pretending the future is a single line.
Guide
What is demand forecasting?
Demand forecasting estimates how much of a product or service people will want in a coming period, so you can buy, staff, and price before the rush — not after it has already walked in.
Use case
Travel pricing forecasting
Forecast market prices and willingness to pay by route, property, departure, segment, and booking window.
Use case
Retail demand forecasting
Anticipate demand by product, store, channel, and region before buying or allocation decisions are locked.