What does $lookup do, and why is it sometimes expensive?

Short answer

It is a left outer join: matching documents from another collection are attached as an array. It can be expensive because it runs per input document, the joined array counts toward the 16MB limit, and a missing index on the foreign field turns each lookup into a scan.

Why?

Non-matching parents still come through with an empty array. $match before $lookup shrinks the input set — usually the biggest win. A frequent $lookup is often a signal the two entities should have been embedded, or that the read should be two targeted queries in the app.

Example

db.users.aggregate([
  {
    $lookup: {
      from: "orders",
      localField: "_id",
      foreignField: "userId",
      as: "orders"
    }
  }
])

Interview tip

If asked about joins in MongoDB, mention that a hot join is often a modelling smell, not just a missing index.

Common mistake

Looking up orders for every user on a list page with no $match and no index on orders.userId.