Aggregation Pipeline — Introduction

find() retrieves documents. The aggregation pipeline transforms them — filtering, grouping, reshaping, and computing new values, all in one operation made of chained stages, each one feeding its output into the next.

A left-to-right flow diagram showing three connected stage boxes labeled $match, $group, and $project, with documents flowing through arrows between them, narrowing at each stage

The Pipeline Shape

Call aggregate() with an array of stage objects — each object's key is the stage operator, starting with $:

db.orders.aggregate([
  { $match: { status: "completed" } },
  { $group: { _id: "$customerId", total: { $sum: "$amount" } } },
  { $project: { customerId: "$_id", total: 1, _id: 0 } }
])

$match — Filter Documents

$match works like a find() query, and is usually the first stage — filtering early means every following stage processes fewer documents:

{ $match: { status: "completed" } }

$group — Combine Documents

$group collapses many documents into one per distinct value of _id, computing a value for each group with an accumulator like $sum, $avg, $max, or $count:

{ $group: { _id: "$customerId", total: { $sum: "$amount" } } }

$project — Reshape the Output

$project works like a projection in find(), but can also compute new fields, not just include or exclude existing ones:

{ $project: { customerId: "$_id", total: 1, _id: 0 } }

These three stages go a long way

These three stages alone — $match, $group, $project — already cover a large share of real reporting queries: "total spend per customer," "orders per day," "average rating per product."