The products chosen for promotion are frequently not the ones with the best margin or the highest sales. They are chosen for what accompanies them.

Transactions reveal associations

Every till receipt records a set of products bought together, and across millions of transactions the patterns of co-occurrence become measurable.

Some associations are obvious and follow from how the products are used. Others are less predictable and emerge only from the data.

What matters commercially is not that products appear together but whether promoting one reliably increases purchases of the other.

Frequency alone is misleading, because the most common pairings involve products that nearly everyone buys. The useful measure compares observed co-occurrence against what would be expected by chance.

Trip drivers justify their own losses

Certain products determine which shop a customer visits, and promoting them brings in a whole basket rather than a single sale.

Against that basket, a loss on the promoted item can be comfortably justified, which is why these lines carry the deepest reductions.

Identifying them requires looking at what customers came for rather than what they spent most on, and the two lists differ substantially.

Association is not the same as causation

Two products bought together may simply reflect the same type of shopper rather than any relationship between them.

Promoting one in that case brings forward purchases of both without generating anything additional, and the analysis records a lift that is not real.

Distinguishing the two requires testing, which is why promotional selection increasingly involves controlled trials rather than analysis of historic baskets alone.

Category management sets the boundaries

Promotions are planned at category level so that discounted items do not simply take sales from their neighbours on the same shelf.

Within a category, the aim is to select items whose promotion expands total category sales rather than redistributing them.

Products that substitute closely for each other are therefore rarely promoted together, since the combined effect would be smaller than either alone.

Placement completes the intention

Once items are selected, their physical position determines whether the associated purchases actually follow, and a promotion isolated from its complements underperforms.

This is why promoted lines appear near related products rather than only in dedicated promotional areas, and why some are duplicated in both locations.

The promotion and the layout are consequently planned together, since the discount alone accounts for only part of the intended effect.