China’s clothes are getting dearer while its prices overall barely move. Apparel rose 1.5% in April, ahead of a headline rate of 1.2% that falling food holds down. Europe runs the other way, and the gap leaves one class of pricing software aimed at opposite goals: it defends margin in Beijing and clears glut in Berlin.

The strength is narrow, which is the point. Food fell 1.6% in the same month while non-food prices rose 1.8%, so clothing sits among the few categories still able to hold a number. A retailer there has room to price up, and the model is told to take it. The model’s task in Beijing is restraint.

Europe’s numbers invert the arithmetic. Euro-area inflation reached 3.0% in April (flash estimate), carried by services and energy, while the industrial-goods basket that holds clothing rose only 0.8%. The continent’s inflation is a services story, and clothing is the most cuttable line in the goods column. Demand for a coat is elastic in a way rent and electricity are not. That soft 0.8% is what keeps the headline looking respectable.

Germany shows the regime up close. Prices there rose 2.1% in January while clothing fell 0.1% on the year and fell 5.5% from December as the winter sales ran. The index climbed while the wardrobe was marked down inside it. To the shopper that was a sale; to the engine, a target met.

The tool answering both calls is the same tool. Markdown engines, among them 7Learnings, compute the shallowest discount that will clear a set quantity by a set date. In the console it is one slider: hold the price, or move the stock. Give such a model pricing power and it guards the margin; give it a glut and it sets the pace of the clearance. The intelligence is the same; only the objective moves.

The algorithm cannot tell whether it is defending a price or surrendering one; it knows only the figure it was set to reach.

A label selling in both markets runs one dashboard with two settings. The Shanghai store is told to protect the ticket; the Berlin store, to move the rack before the weather turns. Nothing in the software prefers one to the other. It optimises whatever it is pointed at, and head office points it differently by postcode. Let the stock cycle turn, and head office turns the same dials the other way.

The two jobs are not mirror images. Defending a price means refusing the discount a rival’s move invites, holding the ticket as prices fall around it. Clearing a glut means discounting early enough to beat the season and shallowly enough to keep something back. One setting tells the model to sit still, the other to move first.

The financial desks treat these prints as demand. China’s consumer inflation touched a three-year high in February, read as appetite returning, and Europe’s flat clothing as inflation finally cooling. That is not a story about two consumers — it is a story about two instructions. The shopper still believes the price is a fact about the world. It is a fact about a setting.