Recipe API for fast, personalized recipe searches

Customer

We have developed a recipe API for the new Migusto website. It is now the central point of contact for all Migros recipes and is used by Migusto, Famigros and iMpuls.

The GraphQL interface delivers recipe data in real time (and at high speed). But not only that, thanks to an intelligent search, it is also the backbone of the Migusto search function.

Download iPhone AppDownload Android App

Main ingredients of the new API

  • Recipes with seasonal ingredients are weighted higher
  • Personal diets (vegan, gluten-free) are taken into account
  • Central recipe API for Migusto, Famigros and iMpuls
  • High performance and complex queries thanks to GraphQL

The relevant recipes on the plate immediately

The recipe API aggregates content from different sources and makes it available to platforms with different requirements. Recipes are regularly imported from the editorial system and the search index is optimized. The recipes are enriched live with community data (ratings, comments and questions) from Reactions: Tastes great, but even better with a little chili.

The interface can handle multiple clients. It delivers recipes for Migusto, Famigros and iMpuls. Thanks to GraphQL, a flexible query tailored to the client and the context is possible: only the required fields are queried, be it image and title or all steps and detailed nutritional information.

The right recipe straight away

The Migusto website relies on a lightning-fast search. The recipe API offers full-text search in a large number of recipes, combinable with filtering by categories such as "vegan" and "dessert". Even complex search queries deliver relevant results - thanks to Elasticsearch and Autocorrect, Bigram Matching, Stemming and synonym recognition with an average response time of just 40ms.

Looking for inspiration? The autocomplete function takes away the typing work and provides cooking ideas. Pizza... Margherita? Pizza bread? With prawns?

The right products in the shopping cart thanks to big data.

A recipe has ingredients, but products end up in the shopping cart. This is where the recipe API bridges the gap. It lists suitable products that can be added directly to the shopping list. If a preference is known, suitable products are suggested, such as organic milk. Hadoop is used to calculate personalized preferences from purchases at Migros. Thanks to the innovative use of big data, useful functions that make life easier are offered at every stage, from the recipe search and shopping to the finished menu.

Always the right recipes

The interface delivers recipes according to seasonal relevance. For example, if you search for a risotto in spring, a wild garlic risotto is suggested, whereas in fall a porcini mushroom risotto is weighted higher. Personal preferences such as a vegan or lactose-free diet are also included in the weighting, if known.

Ein Mann sitzt lächelnd auf dem Boden in einem Raum mit mehreren anderen Personen, die auf einem Sofa im Hintergrund sitzen und ebenfalls lachen.

Questions or interested? Thilo will be happy to help you.