By integrating a grocery store into Whisk, the grocery store will be available within Whisk’s platform. This means the store can choose whether the store appears:
- Within every consenting partner on Whisk’s platform (this includes recipe publishers, Samsung Smart Appliances, users of Whisk’s APIs, and Whisk’s consumer apps). There are over 4M pages receiving 400M monthly impressions where your store may appear.
- Exclusively for one or more partners within Whisk’s network
- Within Whisk's consumer apps
Upon integration, a grocer becomes available to users based on these rules:
- Across all Whisk platforms/sites (option A) or on only specific sites/partners (option B)
- Users in the relevant locations (geo-location)
Technical integration
Every grocer integration is unique because your technology stack is unique. The following guidance is intended to be general in nature. Consultation with Whisk is necessary to develop a more specific plan for integration, including costs.
There are two parts to the initial integration work:
- Product mapping:
- Whisk needs access to your product catalog, for initial product mapping and for scheduling daily updates (stock, pricing, etc.).
- Products are mapped to Whisk’s internal ontology which allows Whisk to display the correct store products relevant to the shopping list items.
- Sending baskets:
- Whisk needs a solution for customers to send products to their ecommerce basket for your store.
- There are a couple of options and authentication can happen as part of the send to basket flow or on your site/app
Example user flow:
Integration process
Every grocer integration is custom as each grocer has a unique tech stack and product catalog makeup and structure. For this reason, consultation is required to assess and validate technical capabilities on the grocer side and align on the scope of the integration before cost and timing proposals can be put together.
A consultation with a grocer looks something like this:
General Integration Options
There are two components to the integration: 1. Product Matching and 2. Send to Basket
Product Mapping
Whisk uses two methods to map store products to ingredients.
- Using Whisks’ Food Genome (recommended)
Whisk syncs a store’s inventory into the Whisk platform. This can be achieved through an API or by uploading flat files with all product inventory. Inventory at the retailer is typically synced every 24 hours. Whisk then uses the following to make the links between products and shopping list items:
- Natural Language Processing: Whisk matches all products to its proprietary ontology, the Food Genome, automatically using Deep-learning based Natural Language Processing
- Product mapping: Most shopping list line-items have multiple possible products from retailers, so Whisk selects defaults based on algorithms which select “best matches”. These are based on the following criteria to select the most appropriate product:
- Item required (what’s closest)
- Quantity required
- Attributes (e.g. organic, fresh, frozen)
- Price
- Personal preferences
- Other attributes
- Using a store’s own matching (alternative)
-
If a store has its own product matching capability, this can be used instead. This typically allows Whisk to pass in a simplified ingredient (e.g. “<5>, <Carrot>” instead of “5 carrots cut into small pieces and washed”). It may also allow Whisk to authenticate a user in order to use their store loyalty scheme to find personalised matches based on a user’s in-store purchases.
Send to Basket
Whisk then needs a method through which to send lists of items to a users’ basket at the retailer. This can be achieved using:
- Authentication during Whisk send to basket flow:
- The grocer can develop/provide an OAuth/SSO APIs (best-practice, recommended)
- Authentication on grocer site
- Whisk needs an endpoint from the grocer to send products to a basket -- in the Whisk flow, the user is sent to the grocer site to login (if necessary) and view their items in their basket
Store Product Data
For details on the store product data points that are required and recommended, please see this article.