{"id":17801,"date":"2026-01-08T12:56:16","date_gmt":"2026-01-08T20:56:16","guid":{"rendered":"https:\/\/www.couchbase.com\/blog\/?p=17801"},"modified":"2026-03-31T16:25:34","modified_gmt":"2026-03-31T23:25:34","slug":"filtered-ann-search-with-composite-vector-indexes","status":"publish","type":"post","link":"https:\/\/www.couchbase.com\/blog\/filtered-ann-search-with-composite-vector-indexes\/","title":{"rendered":"Filtered ANN Search With Composite Vector Indexes (Part 1)"},"content":{"rendered":"

This post kicks off a multi-part series on composite vector indexing in Couchbase. We will start by building intuition, then progressively dive into internals, execution optimizations, and performance.<\/span><\/p>\n

The series will cover:<\/p>\n

    \n
  1. Why composite vector indexes matter, including concepts, terminology, and developer motivation. A Smart Grocery Recommendation System will be used as a running example.<\/span><\/li>\n
  2. How composite vector indexes are implemented inside the Couchbase Indexing Service.<\/span><\/a><\/li>\n
  3. How ORDER BY pushdown works for composite vector queries.<\/span><\/a><\/li>\n
  4. Real-world performance behavior and benchmarking results.<\/span><\/a><\/li>\n<\/ol>\n

    Smart Grocery Recommendation System With Filtered ANN<\/span><\/h2>\n

    Imagine you’re building a grocery-recommendation app. <\/span><\/p>\n

    A user opens it on a Sunday morning and types:<\/span><\/p>\n

    \u201cI love dark chocolate spread, but I\u2019m trying to cut sugar and add more protein. What else should I buy?\u201d<\/span><\/p>\n

    At this moment, your system needs to understand the user\u2019s intent, compare food items semantically, and apply strict nutritional filters.<\/span><\/p>\n

    This is exactly where Filtered Approximate Nearest Neighbor (Filtered ANN) comes in:<\/span><\/p>\n