Relevancy Tuning

Ranking Adjustments

Ranking adjustments are a powerful way to fine-tune search results using your business data and a variety of different boosts options. These boosts provide the building blocks for promotions, merchandising and personalization.

Feature score weight

The feature score weight determines the importance of your ranking adjustments in relation to the overall relevance score. The bigger the feature score weight, the less important the index score becomes, potentially negatively affecting the textual relevance.

If set to 0.2, the total score value of all feature boosts makes up 20% of the relevance score.

Tip: To ensure search results remain relevant, we recommend to start with a weight of 0.2 and experiment to find the perfect ratio.

Filter boost

Filter boosts are a great way to promote and personalize results. Combined with conditions, they provide a powerful tool to adjust the ranking of your search results to optimize your business metrics.

A filter boost is defined by two fields, a filter expression and a boost score. Results that match the filter expression will be boosted according to the given score.


Assuming you want to run a promotion on Nike products in your online store. The following Filter boost will boost Nike products by a score of 0.8.

brand = "nike"0.8

But filter boosts aren't limited to pre-defined values like "nike" in the example above. The filter expression can reference dynamic parameters passed into the search.

To personalize results, you can pass information about the user, such as brand affinity, into the search query and improve the search results based on that data.

    "variables": {
        "q": "headphones",
        "brandAffinity": "Adidas"

In the above query example, we are passing a variable called "brandAffinity" to the search query. The variable can be referenced in the filter boost like so:

brand = brandAffinity0.7

Range boost

Boost results that fall within a specific numeric range. Boosting is applied linearly between the start and the end point. Use range boosts to boost highly rated products, popular products or inventory that needs to be sold.


Assume that you have a rating field that ranges between 0-5 and the values can be decimal values (not just whole numbers). You want to boost values between 2-4 in a linear manner, where 2 receives the smallest boost and 4 (and above) the maximum boost. Ratings between 0 and 2 will receive no boost.

The configuration for this would look as follows, where the score boost of 0.4 is the maximum boost applied to values of 4 and above.