Overall steps for the API integration:
- generate feed (or use platform integrations that already exist)
- create Findify account
- choose which capabilities you want to use: Search, Autocomplete, Collections, Recommendations
- integrate Analytics
-- all events from Feedback API
- validate with Findify team that we get all events properly
- go live
Analytics are an integral part of Findify that allow our ML and Personalization algorithms to operate. It is paramount that you integrate all the analytics events properly.
Some of the events rely on the request-id property that you will receive after performing a request to our backend. This helps us bind the response that we have sent with the actions that user has performed (e.g. clicked on a product in the search results).
Some of the events play a lesser role in the personalization (e.g. click-redirect event that happens when you click on a search suggestion that leads to a redirect), but it's still good to have them as in the future we might rely on such events for new features.
After setting up a page redirect, you can observe redirect property present in both autocomplete and search responses. You can utilize this data to redirect the user to a certain page.
After setting up a banner for search or collection, the API response will contain banner property with the details of a banner that can help you render it on the page.
Dynamic Recommendation Filters
Dynamic recommendation filters (happens when you choose ' Currently viewed' option in the recommendation filter setup) relies on the values sent from the frontend. If you want to utilize this ability, you will need to send the filter values in the recommendation request.
We strongly advice against caching responses from Findify as it will nullify the effect of ML and personalization.
We have an existing SDK written for JS (Github: Findify JS SDK) environments and several helper libraries that can help you build a JS integration and some React-specific libraries, if you are working in React environment.
Updated about a month ago