Filters enable your customers to drill-down and refine the product recall set based on certain attributes. Relevant filters make product discovery easy and intuitive.
In some instances, you may come across irrelevant filters and filter values being displayed in the search results. Filters and filter values may become noisy when your product recall is noisy. The product recall can be noisy if you have noisy fields marked searchable. Irrelevant filters may confuse and frustrate users, leading to a poor search experience.
Precision mode available for search can sometimes remove relevant (or related) products that customers might want in the recall set.
Precision mode for filters and filter values feature targets and removes filter noise in search results. This mode helps clean up the irrelevant filters coming from the products buried in the search results.
For instance, your customer searches for the query “red throw pillows”.
Without Filter Precision applied, the product results show the following facets on page 1:
Relevant Filters: Color, Pillow Cases, Sheet Sets, Outdoor Pillows
Irrelevant Filters: Vase, Throws, Candles
When you apply Filter Precision mode, these irrelevant filters are dropped automatically. This allows your customer to narrow down the search results using the most relevant attributes/categories.
An important point to note is that Filter Precision mode doesn’t change the product recall set. It just eliminates the filter noise.
Less manual work, more scalability: Filter Precision reduces the need to create facet rules as it cleans up irrelevant facets automatically. This helps improve your site’s search experience at scale.
Improved relevance that can lead to more revenue: With a contextual search experience, customers can easily find the products they're looking for. This leads to faster conversions and more revenue for your business.
Better customer experience: Relevant facets make product discovery effortless for your customers. With precise filters, customers can easily refine the results based on meaningful attributes. This results in improved product findability and faster conversions.
When you enable Filter Precision mode, it displays the most relevant facets and facet values that match your customer’s search query.
The relevant facets and facet values are determined based on the products in the dominant categories of that query. Dominant categories are determined based on the top 50 products in the recall set. We can customize this limit for you, but the default value is 50.
For example, for the query "dress", the top 50 products might belong to "evening dress", "maxi dress", "cocktail dress" categories. The 78th product might be a pair of dress shoes that belong to the "men shoes", and "women shoes" categories.
In this case, "men shoes" and "women shoes" will not be a part of the dominant category set (since products from these categories don't appear in the top 50); "evening dress", "maxi dress", "cocktail dress" will be in our dominant categories list. Hence, the filters and filter values will be generated based on the products in the recall set which do exist in these dominant categories. The API response will not return any filters or filter values related to "dress shoes".
This feature is available to all customers using Bloomreach Search for their English language domains.
Live Customers: Please get in touch with the Bloomreach Support team if you’d like to enable this feature.
Integrating Customers: Please reach out to your Technical Services representative who is helping with your integration.
1) Can this feature be enabled on staging?
Yes, this feature can be enabled on Staging. Please reach out to your Technical Services representative who is helping with your integration.
2) Can this feature be AB tested?
Please get in touch with our Business Services team if you’d like to AB test this feature.
3) Does this feature work on the Recommendations API?
No, this feature is supported only on Search API.
4) Can this feature be applied to both product and SKU level filters?
5) Is the Filter Precision applicable to all facet types - text, numeric, single-valued, and multi-valued fields?
Updated 26 days ago