"Is this actually changing my prices?"

No. Simulation is read-only. It predicts the outcome of a price change so the customer can decide; nothing goes live until they act on it through the normal pricing flow.

"Why does it sometimes give me a price but no forecast?"

Two reasons. Either the product has a fixed rule in place (so its price is already decided and there is no curve to forecast against), or the product never had enough sales history for the system to learn a demand pattern. In both cases the system is being honest: it can hand back a price, but it will not invent a forecast it has no basis for.

"Why can it show demand and revenue but not profit for some products?"

Profit needs cost data. If a product's costs were never provided, the system cannot compute margin, so it leaves profit blank rather than guessing - but demand and revenue, which do not need costs, are still shown.

"How does it pick the 'optimal' price?"

It reads the product's demand curve and finds the price where profit (or revenue, if that is the goal) peaks. It keeps the search within sensible limits - never below cost, and never far outside the prices the product has actually sold at, because predictions get unreliable the further they stray from real data.

"Why did my small t-shirt get priced higher than my large?"

This almost always comes down to how the products are grouped. The system simulates and optimizes each product against its own demand curve, one product at a time. It only keeps variants priced consistently with each other when they have been grouped together in the data (under a shared group). If your small, medium, and large shirts are not grouped, the system treats each size as its own standalone product - it has no way of knowing they are the same shirt in different sizes. Each size then gets the price that is individually optimal for its own sales history, and nothing forces those prices into a sensible size order. The result can be a small priced above a large. The fix is to make sure variants that should be reasoned about together are grouped together in the data; once they are, the system optimizes them as a set rather than as unrelated products.

By default, three weeks. The customer can set any start and end date they want. (Note this is shorter than the seasonality forecast horizon, which looks 90 days ahead - simulation defaults to a nearer window because it is meant for immediate pricing decisions.)

"Can the forecast be wrong?"

Like any forecast, yes. It is built from patterns in past sales and cannot anticipate genuinely new events - a viral moment, a new competitor, a supply shock. That is exactly why it reports a confidence range rather than a single number, and why it stays inside the price range it has real data for. The narrower the range, the more the customer can lean on the central estimate.

"What if I have limited stock?"

Turn on stock awareness and the forecast becomes a realistic sell-through projection: it caps each day's sales at the stock expected to remain and tells you the date the product is projected to run out. Left off (the default), it shows pure demand at the chosen price, as if supply were unlimited.

"Does it handle multiple stores?"

Yes - but how the numbers read depends on whether a store is selected. When a specific store is chosen, the forecast is scoped to that store and the result notes which store it applies to. When no store is selected, the system falls back to a combined sales history across all of the customer's stores, and the forecast then describes what a single average store looks like across the portfolio - not the total across every store added up. This trips people up: a customer with ten stores might expect the no-store forecast to show ten stores' worth of volume, but it shows roughly one store's worth, averaged. To see a particular store's numbers, select that store. Account-wide cumulative simulation (summing across all stores) is a known gap and is on the way.

Price simulation - A prediction of what demand, revenue, and profit would do if a proposed price went live. Read-only; it changes nothing.

Demand curve - A product's learned relationship between price and sales volume. The simulator reads a point off this curve for the proposed price. Built by a separate model from the product's sales history.

Optimal price - The price that maximizes a chosen objective (profit or revenue), found by searching the demand curve. Kept within sensible cost and data-range limits.

Manual / Dynamic / Optimal - The three ways a price reaches the simulator: typed in by hand, produced by the customer's dynamic-pricing rules, or computed by the system.

Confidence range - The optimistic-to-pessimistic span around a forecast number. Narrow means the system is confident; wide means uncertain.

Safeguards - Customer-defined price floors and ceilings the recommended price is clamped to.

Stock awareness - An optional mode that caps the forecast at available inventory and reports the projected sell-out date. Off by default.

Seasonality multiplier - A per-date adjustment that lifts or lowers expected demand to reflect known seasonal patterns over the forecast window.

Forecast window - The date range the simulation covers. Defaults to the next three weeks; fully adjustable.