Text vs Vector Search in MongoDB Atlas

In this post, we’re gonna run a side-by-side search spin-off in MongoDB Atlas:

• Atlas Search (keyword / full-text) using the text operator (scored by BM25 by default).
• Atlas Vector Search (semantic / meaning-based) using the $vectorSearch aggregation stage (nearest-neighbor search over embeddings).

The goal isn’t to “replace” text search. It’s to experience, hands-on, why vector search is different.

• Keyword search answers: “Which documents contain these words?” (BM25-style ranking).
• Vector search answers: “Which documents are most similar in meaning?” (closest vectors).

To keep this truly “Hello World” and fast to reproduce, we’ll use Atlas’s sample dataset (sample_mflix) which includes an embedded_movies collection designed for vector search experiments.

Prerequisites

Accounts + cloud setup

• A MongoDB Atlas account and a project.
• An Atlas cluster you can connect to (Free Tier / Flex is fine for this tutorial).

Local tools

• Git (to clone the repo / commit your changes)
• mongosh (MongoDB Shell) or MongoDB Compass

Atlas data + indexes we’ll create

• Load Atlas Sample Dataset → sample_mflix database, including embedded_movies
• We will create:
  • A Search (text) index for keyword search using the text operator (BM25 scoring by default).
  • A Vector Search index so $vectorSearch can run.

Optional (nice to haves)

• Node.js (if you want a tiny “demo app” wrapper around the queries later).
• Atlas CLI (not required for this, but handy once you automate)