Accumulateurs $first, $last et $top/$bottom

Sometimes aggregation needs to select a single representative document from each group rather than computing a numeric aggregate. MongoDB provides several accumulators for this: $first, $last, and the newer $top and $bottom accumulators. Each has different semantics for determining which document is selected, and choosing the right one affects both correctness and performance.

$first returns the value from the first document encountered in the group. Since MongoDB does not guarantee document processing order within a $group unless you sort first, $first is meaningful only when you place a $sort stage immediately before the $group. This two-stage pattern—sort then group—is the standard way to reliably pick the earliest or highest-priority item per group.

$last returns the value from the last document encountered in the group. Again, this is meaningful only when combined with a preceding $sort. Sort ascending and use $last to get the most recent value per group. Sort descending and use $first to achieve the same result. The most common use case is finding each user's or customer's most recent activity.

The pattern of $sort before $group has an important performance implication: it prevents MongoDB from pushing the $group in front of the $sort during pipeline optimization. If the collection is large and you have an index on the sort field, MongoDB can use it for the sort phase but still must sort all documents before grouping. For very large collections, this can be expensive. The $top and $bottom accumulators were introduced partly to address this limitation.

Introduced in MongoDB 5.2, $top lets you pick the top document within each group according to a sort specification—without a separate $sort stage. You specify a sortBy expression and an output expression directly inside the accumulator. MongoDB evaluates the sort per group, which can be more efficient than sorting the entire collection first.

$bottom is the complement to $top—it returns the document that would appear last after sorting by the given key within each group. Use $bottom with sortBy: { timestamp: 1 } to get the earliest event per group, or use $top with the opposite sort direction—both approaches are equivalent. Choose whichever expresses your intent more clearly.

MongoDB also provides $topN and $bottomN accumulators that return an array of the top or bottom N documents per group according to a sort key. This is useful for leaderboard-style queries where you want the top 3 scores per level or the 5 most recent orders per customer, all computed in a single $group stage.

Use $first/$last when you are already sorting the pipeline for other reasons and want a simple, familiar API. Use $top/$bottom when you only need the selection behavior and want to avoid the cost of a full collection sort before $group. In practice, $top/$bottom with appropriate indexes can be significantly faster for large collections because MongoDB may be able to optimize the per-group selection more efficiently.

A classic real-world use case is maintaining a price history collection and querying the latest effective price for each product. Using $top with sort by date descending makes this query concise and efficient. The result is one document per product containing its current price without needing a separate 'current price' collection.

You can freely combine $first, $last, $top, $bottom, $sum, $avg, and $push in a single $group stage. Each accumulator operates independently on the documents in the group. This lets you compute summary statistics alongside picking representative documents in one efficient aggregation stage, avoiding multiple passes over the data.

$first and $last have been available since early MongoDB versions. The $top and $bottom accumulators were introduced in MongoDB 5.2, and $topN/$bottomN also became available around that time. If you are using an older MongoDB version (pre-5.2), fall back to the $sort then $first/$last pattern. Check your Atlas cluster version or self-hosted version before using newer operators.

Test your understanding of MongoDB & NoSQL Databases concepts from this lesson.

In this lesson you learned: $first/$last pick values from the first/last document in a group (requires prior $sort), $top/$bottom (MongoDB 5.2+) pick by sort key within the group without a separate $sort stage, and $topN/$bottomN return arrays of the top or bottom N items per group. Next up we explore window functions with $setWindowFields for running totals and moving averages.