MongoDB Aggregation Pipelines – Step-by-Step Explanation

The MongoDB Aggregation Pipeline is a powerful framework for processing and transforming data within MongoDB. It allows you to perform complex data analysis, filtering, grouping, reshaping, and calculations — all in the database, without pulling raw data into your application.

Think of it as a conveyor belt where documents flow through a series of stages, and each stage transforms the data before passing it to the next.

Why Use Aggregation Pipelines?

Core Concept: The Pipeline

db.collection.aggregate([
  { $stage1 },
  { $stage2 },
  { $stage3 },
  ...
])

• Each { $stage } is a stage in the pipeline.
• Documents flow from left to right.
• Output of one stage becomes input to the next.
• Final result is returned as an array of documents.

Key Aggregation Stages

Here are the most commonly used stages:

Step-by-Step Examples

Let’s use a sample collection: orders

{
  _id: 1,
  customer: "Alice",
  items: ["laptop", "mouse"],
  total: 1200,
  status: "completed",
  country: "USA"
},
{
  _id: 2,
  customer: "Bob",
  items: ["phone"],
  total: 800,
  status: "pending",
  country: "Canada"
}

1. $match – Filter Documents

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

Only completed orders pass through.

2. $project – Reshape Output

{
  $project: {
    customer: 1,
    total: 1,
    itemCount: { $size: "$items" }
  }
}

Result:

{ customer: "Alice", total: 1200, itemCount: 2 }

3. $group – Group & Aggregate

Goal: Total sales per country

{
  $group: {
    _id: "$country",           // Group by country
    totalSales: { $sum: "$total" },
    orderCount: { $sum: 1 },
    avgOrder: { $avg: "$total" }
  }
}

Result:

{ _id: "USA", totalSales: 1200, orderCount: 1, avgOrder: 1200 }
{ _id: "Canada", totalSales: 800, orderCount: 1, avgOrder: 800 }

Accumulators in $group:
- $sum, $avg, $min, $max, $push, $addToSet, $first, $last

4. $sort – Order Results

{ $sort: { totalSales: -1 } }

Sort by total sales descending.

5. $limit – Top N Results

{ $limit: 3 }

Only top 3 countries.

6. $unwind – Expand Arrays

{ $unwind: "$items" }

Turns:

{ items: ["laptop", "mouse"] }

Into two documents:

{ items: "laptop" }, { items: "mouse" }

Useful for analyzing individual array elements.

7. $lookup – Join Collections

Assume another collection: customers

{
  $lookup: {
    from: "customers",
    localField: "customer",
    foreignField: "name",
    as: "customerInfo"
  }
}

Adds customer details (e.g., email, phone) from another collection.

8. $addFields – Compute New Fields

{
  $addFields: {
    tax: { $multiply: ["$total", 0.08] },
    totalWithTax: { $add: ["$total", { $multiply: ["$total", 0.08] }] }
  }
}

Full Example: Top Customers by Spending

db.orders.aggregate([
  { $match: { status: "completed" } },

  { $group: {
      _id: "$customer",
      totalSpent: { $sum: "$total" },
      orders: { $sum: 1 }
  }},

  { $sort: { totalSpent: -1 } },

  { $limit: 5 },

  { $project: {
      customer: "$_id",
      totalSpent: 1,
      orders: 1,
      _id: 0
  }}
])

Output:

{ customer: "Alice", totalSpent: 1200, orders: 1 }

Advanced: Conditional Logic

Use $cond, $switch, $ifNull

{
  $addFields: {
    statusLabel: {
      $switch: {
        branches: [
          { case: { $eq: ["$status", "completed"] }, then: "Done" },
          { case: { $eq: ["$status", "pending"] }, then: "Waiting" }
        ],
        default: "Unknown"
      }
    }
  }
}

Performance Tips

1. Put $match early – filter as soon as possible.
2. Use indexes on fields in $match and $sort.
3. Project only needed fields early to reduce memory.
4. Avoid large $unwind on big arrays.
5. Use allowDiskUse: true for large datasets:

db.collection.aggregate(pipeline, { allowDiskUse: true })

Tools to Visualize Pipelines

• MongoDB Compass – Drag-and-drop pipeline builder
• MongoDB Atlas – Visual pipeline editor
• mongosh – Test in shell

Summary: Pipeline Flow

[Raw Docs]
     ↓
[$match] → filter
     ↓
[$project/$addFields] → reshape
     ↓
[$unwind] → expand arrays
     ↓
[$group] → aggregate
     ↓
[$sort] → order
     ↓
[$limit] → paginate
     ↓
[Result]

Official Docs & Resources

• MongoDB Aggregation Pipeline
• Aggregation Pipeline Stages
• Try it live: MongoDB Playground

Practice Tip: Start with small datasets. Build pipelines step by step in mongosh, adding one stage at a time and checking output with .pretty().

Let me know if you want a real-world example (e.g., e-commerce dashboard, log analysis, user analytics)!