Personalization and Recommendations

SuggestAPI has two complementary intelligence layers in the backend:

They solve different problems.

What personalization does

Personalization changes search results for a specific user based on what SuggestAPI knows about that user.

In practice, that means you can:

store or update a user profile
record behavior such as clicks, views, and other useful signals
pass a user_id into search requests
let SuggestAPI boost results that better match that user’s preferences or past behavior

This is most useful when the same query should lead to different results for different users.

Examples:

a returning buyer sees brands or categories they engage with most often
a B2B user sees products closer to their industry, application, or spec preferences
a user who repeatedly clicks replacement parts sees those items rank higher for short queries

If you pass user_id to /v1/autocomplete, the service can fold profile-based scores into ranking when plan and schema settings allow it.

Recommendations return related items based on catalog structure, entity relationships, and behavioral patterns.

This is more about “what else should I show?” than “how should I rank this exact query for this exact user?”

In practice, recommendations can help you power:

This is useful on product pages, cart flows, upsell modules, and discovery surfaces outside the main search box.

Use personalization when you want the same search query to rank differently for different users.

Use recommendations when you want to return adjacent or related items even without a search query.

Many teams use both:

Start simple: