Skip to main content
Out-of-stock products frustrate shoppers when they fill the first results, but removing them hurts too: users still search for them, and they come back in stock. The same applies to refurbished items, old article versions, or listings from sellers with poor ratings. Deboosting keeps these documents findable and moves them after the others. One rule with a deboost action covers this, and no condition is needed if you want it to apply to every search.

​
Set up from the Meilisearch Cloud dashboard

Open your project in the Meilisearch Cloud dashboard, select the Search rules tab, and click New rule. Give the rule an ID, such as out-of-stock-last, and a description. In the Actions block, click Add action, select Deboost documents, then click Next. Pick Use a filter, then fill in the dialog:
  • Index: the index holding the products, here products.
  • Filter: the condition selecting them, here stock = 0. The attribute you filter on must be in the filterableAttributes setting of the index, otherwise the action selects no document.
  • Strength: how far the selected products drop. Start with the medium value, then run a few searches of your own and adjust: the right strength depends on your data and on the way your users search.
Click Add action. Deboost documents dialog with the index, the strength, and the filter Leave the Conditions block with its default values, so the rule applies to every search, then click Create rule. Meilisearch indexes the rule, which can take a few seconds to a few minutes. Once it is live, out-of-stock products remain findable, but they lose ground to the products you can ship.

​
Set up on a self-hosted instance

Search rules must be turned on before you call the /dynamic-search-rules routes, otherwise Meilisearch returns an error. See Create your first rule.
  • weight multiplies the relevancy of the selected documents. Between 0 and 1, the documents rank lower. The closer to 0, the further down they go.
  • Deboosted documents stay in the results and in the facet counts. Only their position changes.
  • A rule without conditions applies to every search on the index.
  • The attributes used in filter must be in the filterableAttributes setting of the index, otherwise the action selects nothing.
Meilisearch answers with a summarized task containing a taskUid, because creating a rule is an asynchronous operation:
The rule applies to searches once this task reaches the succeeded status. To follow its progress, pass the taskUid to the task API. From then on, out-of-stock products lose ground to the ones you can ship, and customers still find them. They do not always fall to the bottom: a product that matches the query much better than the others can still come first. Check a few typical searches and lower the weight if you want a stronger effect.

​
Next steps

Hide documents

Remove documents from the results instead of lowering them

Boost documents

Rank a selection of documents higher

Specific behaviors

See how actions interact with filters and with each other

API reference

Endpoints, rule fields, and update behavior