Recommendations
GroupBy Recommendations platform provides enhancements to your customer journey by providing recommendations that are driven by your users’ actions.
For example, you can use this suite of product to:
- show trending products
- show popular search terms
- suggest alternative searches when no results were found
- order your filters and refinements by showcasing the more popular ones higher
All recommendations are driven by a beaconing code that sends your user’s actions to our engine. Once you add beaconing to your site, you are able to leverage the data from Recommendations to increase shopper conversion. Read more about Beacons here.
Results and Outputs
You will then get access to the following Recommendations data:
- Clickboost and Conversion-boost on your records
- Query recommendations (e.g. trending and popular searches)
- Navigation recommendations (e.g. more popular refinements)
- Product recommendations (e.g. bought-together products)
Sample Query Syntax
Here is a sample request to show how any given query to the Recommendations is built up.
This request is for a search recommendation. It is asking the service for:
- popular search queries
- that contained 'samsung’
- over the last week
Test the Recommendation V2 Endpoints and Recommendations V3 Endpoints
Windows and limits
Caching
As part of the Recommendations V2 engine, you have the ability to skip the caching layer.
The caching layer is on by default so that your production environment is protected.
If during development you want to skip the cache, there are limitations are in place on request with skip cache on to protect system performance.
Normal requests will work but rapid automated requests will not.
When caching is not skipped, these rate limits are not in place.
To skip the cache, add a Skip-caching header:
-H "Skip-Caching: true"
Windows
| Item | Limit or Window | Notes |
|---|---|---|
| Clickboost decay | 30 day window | Clickboost is a moving calculation based off the last 30 days of performance |
The windows are day, week, and month, representing a rolling 24 hour, 7 day and 30 day window respectively.