Sales history
Every price change you ever made was a natural experiment
When the price moved, what happened to volume? Two products, the same 20% price hike, and very different answers. Both already sitting in the store's order history.
Bike water bottle
↑ 20% price
→
↓ 30% volume
Price over time
Units sold per month
Price-sensitive. Volume fell further than the price rose. Sell Smart Price Optimization now recommends coming back down.
Bike helmet
↑ 20% price
→
↓ 10% volume
Price over time
Units sold per month
Barely sensitive. Volume held up, so the extra margin stuck — and there is likely more room above.
Overview
From sales data to the optimal price
We estimate the demand curve from past sales. From demand, we derive revenue and margin, then read off the price that maximizes your chosen goal.
1INPUTS
Your sales & price history, plus product description
2PRICING ENGINE
Combines your sales & price history and market context to estimate how demand responds to price. Similar products influence each other's estimates.
3DEMAND CURVE
Predicted demand at every price point, for all products
4REVENUE CURVE
Expected revenue at every price point, derived from the demand curve — its peak is the revenue-maximizing price
5MARGIN CURVE
Expected margin at every price point, derived from revenue and unit cost — its peak is the profit-maximizing price
PRICE HISTORY
Historical list prices over time
SALES HISTORY
Units sold & quantities
PRODUCT DESCRIPTION
Title, attributes & category
→
→
→
PRICING ENGINE
Combines price & sales history and market context to model demand
→
DEMAND CURVE
→
REVENUE CURVE
→
MARGIN CURVE
Cold start
New & slow-moving products
For a new product, we can't determine its own demand curve due to lack of past sales. But we can approximate it based on the sensitivity of similar products.
•We find similar products based on description, products in the same price range, and blend their estimates if they have enough data.
•This isn't only a new-product problem: many catalogs have little price variation, so many products lean on borrowed estimates until their own history grows.
Pooling demand curves from nearest neighbors
→
Similar products' curves
Pooled estimate
A weighted blend
Similarity scores become the weights. The less is known about the product, the more it leans on the group.
Similar products
55%
Price peers
30%
Own sales
15%
As real sales arrive, the green slice grows and the borrowed slices shrink.
Sell Smart Price Optimization · apps.shopify.com/smart-price-optimization
© 2026 Sell Smart