Auchan: a large store should understand a small shopping mission

The format is concrete: Auchan publishes information for visitors to its Angers hypermarket. I use that setting for an independent navigation concept. This is not a report of technology installed by Auchan, nor an attempt to assign the company problems I have not measured.

A hypermarket can accommodate almost everything except the spare time of someone who came for one item. Variety is an advantage for the retailer. For the shopper, it can become geography they unexpectedly need to study. I am interested in AI that translates a large assortment into a small, understandable mission rather than another catalogue.

Search for a task, not just a product name

A shopper might need drawing supplies for a child. There is no single catalogue item for that request. An assistant could turn that task into a few categories, asking about age, budget or items already owned where relevant. Keep the exchange brief. Nobody needs an educational committee meeting before buying pencils and a sketchbook.

The list still needs to become a route using the verified map of that particular store. Temporary displays and accessible passages must be included. If the map is stale, offer a help point instead of invented precision. Updating locations needs to become part of merchandising work; otherwise the clever navigator continues inhabiting last year's floor plan.

At this point the shopper needs a choice between quick finding and inspiration. The shopper chooses whether to explore additional options. Offering something interesting is reasonable; secretly extending the route to encourage extra purchases is a different proposition. Trust in directions can disappear considerably faster than the next version of an app arrives.

Begin with several missions

Several ordinary missions provide a useful trial: school supplies, cleaning or a simple family dinner. Staff and testers should walk the proposed routes, including with a pushchair, before comparing them with ordinary search. Unavailable items must appear in the results. Hiding them behind an average journey time would miss the purpose of the exercise.

The three measures would be completed shopping lists, search time and requests for staff assistance after following the route. Basket value can be observed separately, but should not determine the arrows. Otherwise the navigator learns to be helpful to the organisation paying for it while becoming less helpful to the person trusting it.

The selected store and a voluntarily entered list should be sufficient. Family shopping histories, facial recognition and continuous location tracking are unnecessary. My standard is straightforward: customers know where to go, find what they need and do not feel recruited into an experiment. Sometimes a large shop benefits from behaving like a familiar small one that gives clear directions.