Personalisation

What personalisation means

Personalisation adapts content or offers to individual visitors. The term is broad and ranges from simple rules to calculated recommendations. Three stages can be distinguished, and they differ strongly in effort and effect.

The first stage is grouping: all visitors in a group see the same thing. A group can consist of the country of origin, an advertising channel or the customer history. The second stage is adaptation to the individual based on past behaviour, such as a list of recently viewed items. The third stage is calculated recommendations, where a model derives from the behaviour of many customers what might fit this one.

When the effort pays off

• Broad assortment: the larger the choice, the harder it is for the customer to find what fits. Personalisation works most strongly here.
• Returning visitors: anyone with data across several sessions can adapt better than on a first visit.
• A clear occasion: a recommendation for the matching refill carries; a general home page adaptation rarely does.
• Enough traffic: calculated recommendations need enough orders for patterns to become visible.

Which data is needed

The basis is behaviour: viewed items, searches made, goods bought, abandoned baskets and the origin of the visit. Master data such as size, location or customer group comes on top. Almost all of it is personal data, which makes personalisation a data protection question and not only a technical one.

Two points need particular attention. Anyone tracking behaviour across sessions needs a basis for it and usually consent. And anyone mixing customer data with visitor data has to be able to explain which purpose justifies which processing.

Why too much adaptation harms

The first reason is the feeling of being watched. When a visitor sees only that one category for weeks after a single search, the shop feels narrow and makes them suspicious. The second reason is reinforcement: recommendation systems that only confirm what was already bought prevent the discovery of new items and lower basket value over time.

The third reason is visible mistakes. A recommendation in the wrong place, such as accessories for a product the customer returned, looks inattentive. The fourth reason is the missing explanation: anyone who cannot say why something is being shown treats adaptation as manipulation rather than service.

In practice one rule follows: use personalisation where it obviously helps and leave the rest visibly unadapted. A shop offering a clear standard view plus precise recommendations works better than one adapting everything.

Frequent questions

• Do I need AI for it? No. Grouping and rules bring the first benefit.
• How do I measure success? By basket value and the repeat rate of the group involved, not by clicks on the recommendation.
• What if I have too few customers? Start with simple rules and expand later.