User-tailored results

Search Personalization

Tailor every search to every user. Surface what is most relevant to each person based on their preferences and behavior.

Trusted by leading companies

Same search, different results

Three users search for "laptop". Each sees results ranked by their unique preferences, no extra configuration needed.

Budget-conscious parent
Affordable, durable, good for kids, long battery life, value for money
""
1.Chromebook 14" Education

$249 · 12hr battery, spill-resistant

Budget + durable
2.Lenovo IdeaPad 3

$349 · reliable, great reviews

Value pick
3.Acer Aspire Go 15

$299 · lightweight, all-day battery

Family friendly

Personalized for every user

Same search, different results, based on who's searching. Experimental feature requiring Cohere API.

Natural language preferences

"Budget-conscious, values durability over brand names". Plain English, not config.

Semantic re-ranking

Results re-ranked by semantic match to preferences. Real-time, no reindexing.

Privacy-first

Only preference string sent. User data stays in your app. No PII stored.

Build from behavior

Construct preferences from purchases, views, explicit selections. Your logic.

Works with filters and search

Mix personalization with filters, facets, and keyword search. Results match what users want and who they are.

Real-time adaptation

Change preferences → results change immediately. No waiting.

Powered by leading reranking providers

Personalization uses re-ranking models to match user preferences to results semantically.

5 models from 3 providers

Cohere

Cohere

1 model

Rerank 3.5
Jina

Jina AI

1 model

jina-reranker-v2
Mixedbread AI

Mixedbread AI

3 models

mxbai-rerank-large-v2
mxbai-rerank-base-v2
mxbai-rerank-large-v1
+

Custom

Any provider

Meilisearch is compatible with any model offering a REST API and tool calling capabilities.

Built for every use case

From e-commerce to enterprise, personalization transforms search across industries.

E-commerce

Personalized product rankings based on purchase history, browsing patterns, and stated preferences.

Frequently asked questions

Personalization works by sending a natural language preference string alongside your search query. Meilisearch uses embeddings to semantically match user preferences against search results, re-ranking them so the most relevant items for that specific user appear first. No model training or ML pipeline is required.

Ready to get started?

Run Search Personalization on Meilisearch Cloud, or self-host the open-source engine.