MongoDB $rand Operator

What Is $rand?

$rand is a MongoDB aggregation expression that returns a random floating-point number in the half-open range [0, 1)—at least 0, and always less than 1. Each document gets its own value.

$rand = random float in [0, 1)

{ $rand: {} }

no arguments — always empty {}
per document → independent value
not stable across runs

~20% sample:
  $lt: [ { $rand: {} }, 0.2 ]

exact count → use $sample stage instead

How to Use $rand in MongoDB

Call { $rand: {} }, then store it, scale it, or filter with it.

1. Choose an expression stage ($addFields or $project).
2. Write { $rand: {} } (empty object only).
3. Optionally scale with $multiply / $floor for ranges.
4. For percentage sampling, compare with $lt inside $match + $expr.

Pseudocode

for each document:
    r = random_float()   // 0 ≤ r < 1

    // store
    emit { ..., score: r }

    // or keep ~20%
    if r < 0.2:
        keep document
    else:
        skip

Cheat Sheet

Keep these facts handy while adding randomness to pipelines.

Live Preview

Generate a fresh [0, 1) value (like $rand), and optionally scale it to a dice roll 1–6.

$rand → (click Run for a new value)

range: 0 ≤ r < 1

Worked Walkthrough — Random Ticket Score

Three support tickets sit in a queue. We give each a random score in [0, 1), then keep only tickets with score < 0.2 (about 20%).

Documents:
  A, B, C

Each gets:
  score = $rand → e.g. 0.41, 0.07, 0.88
                 (your numbers will differ)

Keep if score < 0.2:
  A 0.41 → skip
  B 0.07 → keep
  C 0.88 → skip

About 1 in 5 docs kept on average.

MongoDB Examples

We use three tickets. First insert them, then attach a random score, then keep roughly 20% with $match. Run these in mongosh. Your random numbers will differ each run.

db.tickets.insertMany([
  { _id: 1, topic: "Login" },
  { _id: 2, topic: "Billing" },
  { _id: 3, topic: "Shipping" }
]);

db.tickets.find()

[
  { _id: 1, topic: "Login" },
  { _id: 2, topic: "Billing" },
  { _id: 3, topic: "Shipping" }
]

Login     →  $rand → 0.??
Billing   →  $rand → 0.??
Shipping  →  $rand → 0.??

each draw independent, in [0, 1)

db.tickets.aggregate([
  {
    $addFields: {
      score: { $rand: {} }
    }
  }
]);

[
  { _id: 1, topic: "Login", score: 0.412837291 },
  { _id: 2, topic: "Billing", score: 0.073019482 },
  { _id: 3, topic: "Shipping", score: 0.881204553 }
]

db.tickets.aggregate([
  {
    $match: {
      $expr: {
        $lt: [ { $rand: {} }, 0.2 ]
      }
    }
  }
]);

[
  { _id: 2, topic: "Billing" }
]

// Sometimes 0 docs, sometimes 2–3.
// Average keep rate ≈ 20%.

Performance Notes

Drawing a float is cheap, but filtering a huge collection still reads many docs. Prefer $sample when you only need a fixed-size random subset.

Key Takeaways